1 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4 any RTX present on the RHS of a SET.
5 * compare-elim.c (try_eliminate_compare): Restore comment.
7 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10 may be also called on tcc_type nodes.
11 * langhooks.c (lhd_tree_size): Likewise.
13 2017-10-17 David Malcolm <dmalcolm@redhat.com>
15 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
16 format_warning_at_substring.
17 (maybe_warn): Convert source_range * param to a location_t. Pass
18 UNKNOWN_LOCATION rather than NULL to fmtwarn.
19 (format_directive): Remove code to extract source_ranges and
20 source_range * in favor of just a location_t.
21 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
23 * substring-locations.c (format_warning_va): Convert
24 source_range * param to a location_t.
25 (format_warning_at_substring): Likewise.
26 * substring-locations.h (format_warning_va): Likewise.
27 (format_warning_at_substring): Likewise.
29 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
31 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
33 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
34 and scatter/gather ops.
36 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
37 vec_gather_load and vec_scatter_store.
38 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
39 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
40 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
41 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
42 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
43 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
45 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
47 * reg-stack.c (compare_for_stack_reg): Add bool argument.
48 Detect FTST instruction and handle its register pops. Only pop
49 second operand if can_pop_second_op is true.
50 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
51 set can_pop_second_op to false in the compare_for_stack_reg call.
53 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
54 output_fp_compare for stack register operands.
55 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
56 instructions here. Do not emit stack register pops here. Assert
57 that FCOMPP pops next to top stack register. Rewrite function.
59 2017-10-17 Nathan Sidwell <nathan@acm.org>
62 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
63 use DECL_ASSEMBLER_NAME_RAW.
66 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
69 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
70 Wilco Dijkstra <wilco.dijkstra@arm.com>
72 * builtins.c (expand_builtin_update_setjmp_buf): Add a
73 converstion to Pmode from the buf_addr.
75 2017-10-17 Richard Biener <rguenther@suse.de>
77 * graphite-dependences.c (scop_get_reads_and_writes): Change
78 output parameters to references.
80 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
82 PR 71026/tree-optimization
83 * fold-const.c (distribute_real_division): Removed.
84 (fold_binary_loc): Remove calls to distribute_real_divison.
86 2017-10-17 Richard Biener <rguenther@suse.de>
88 * graphite-scop-detection.c
89 (scop_detection::stmt_has_simple_data_refs_p): Always use
90 the full nest as region.
91 (try_generate_gimple_bb): Likewise.
92 * sese.c (scalar_evolution_in_region): Simplify now that
93 SCEV can handle instantiation in regions.
94 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
95 in the non-loop part of a function if requested.
97 2017-10-17 Richard Biener <rguenther@suse.de>
99 PR tree-optimization/82563
100 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
102 (graphite_regenerate_ast_isl): Call it.
103 * graphite-scop-detection.c (build_scops): Remove entry edge split.
105 2017-10-17 Jakub Jelinek <jakub@redhat.com>
107 PR tree-optimization/82549
108 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
109 Formatting fixes. Instead of calling make_bit_field_ref with negative
112 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
114 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
115 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
116 _mm_maskz_reduce_ss): New.
117 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
118 __builtin_ia32_reducess_mask): Ditto..
119 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
120 * config/i386/sse.md (reduces<mode>): Renamed to ...
121 (reduces<mode><mask_scalar_name>): ... this.
122 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
124 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
125 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
127 2017-10-16 David Malcolm <dmalcolm@redhat.com>
129 * Makefile.in (OBJS): Add unique-ptr-tests.o.
130 * selftest-run-tests.c (selftest::run_tests): Call
131 selftest::unique_ptr_tests_cc_tests.
132 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
133 * unique-ptr-tests.cc: New file.
135 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
138 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
140 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
141 (make_hard_regno_born, make_hard_regno_dead): Update
142 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
144 2017-10-16 Jeff Law <law@redhat.com>
146 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
148 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
150 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
152 2017-10-16 Olivier Hainque <hainque@adacore.com>
154 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
155 with_cpu if we were configured for an e500v2 target cpu name.
157 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
159 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
160 * doc/invoke.texi: Document +nodsp as a valid extension for
163 2017-10-16 Martin Liska <mliska@suse.cz>
165 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
166 (test_set_range): Likewise.
167 (test_range_functions): Rename to ...
168 (test_bit_in_range): ... this.
169 (sbitmap_c_tests): Add new test.
171 2017-10-16 Tamar Christina <tamar.christina@arm.com>
173 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
174 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
175 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
177 2017-10-16 Tamar Christina <tamar.christina@arm.com>
179 * config/aarch64/aarch64-builtins.c
180 (aarch64_types_quadopu_lane_qualifiers): New.
181 (TYPES_QUADOPU_LANE): New.
182 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
183 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
184 (aarch64_<sur>dot_laneq<vsi2qi>): New.
185 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
186 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
187 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
188 (Vdottype, DOTPROD): New.
189 (sur): Add SDOT and UDOT.
191 2017-10-16 Tamar Christina <tamar.christina@arm.com>
193 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
194 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
195 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
196 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
197 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
198 (cortex-a75.cortex-a55): Likewise.
199 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
201 2017-10-16 Tamar Christina <tamar.christina@arm.com>
203 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
204 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
205 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
206 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
207 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
208 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
209 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
210 * config/arm/types.md (neon_dot, neon_dot_q): New.
211 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
213 2017-10-16 Tamar Christina <tamar.christina@arm.com>
215 * config/arm/arm.h (TARGET_DOTPROD): New.
216 * config/arm/arm.c (arm_arch_dotprod): New.
217 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
218 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
219 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
220 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
221 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
222 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
223 * doc/invoke.texi (armv8.2-a): Document dotprod
225 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
227 * i386.c (ix86_vec_cost): New function.
228 (ix86_rtx_costs): Handle vector operations better.
229 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
230 * x86-tune-costs.h: Add new costs to all tables.
232 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
234 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
236 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
237 divsd, sqrtss and sqrtsd
238 * x86-tune-costs.h: Add new entries to all costs.
239 (znver1_cost): Fix to match real instruction latencies.
241 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
242 Michael Collison <michael.collison@arm.com>
244 * compare-elim.c: Include emit-rtl.h.
245 (can_merge_compare_into_arith): New function.
246 (try_validate_parallel): Likewise.
247 (try_merge_compare): Likewise.
248 (try_eliminate_compare): Call the above when no previous clobber
250 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
253 2017-10-14 Jakub Jelinek <jakub@redhat.com>
257 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
258 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
262 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
263 to be any operand_equal_p operands. For & (B - 1) require
264 B to be power of 2. Recognize
265 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
267 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
270 * optabs.c (expand_memory_blockage): Fix call of
271 targetm.have_memory_blockage.
273 2017-10-14 Jakub Jelinek <jakub@redhat.com>
276 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
277 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
278 objects to extra_objs instead of overwriting it.
280 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
282 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
283 Use any_fp_register_operand as operand[3] predicate. Simplify
284 equality test for operands[2] and operands[4] memory location.
285 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
286 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
287 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
288 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
289 any_fp_register_operand as operand[1] predicate. Simplify
290 equality test for operands[0] and operands[3] memory location.
291 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
292 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
293 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
295 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
297 * target-insns.def: Add memory_blockage.
298 * optabs.c (expand_memory_blockage): New function.
299 (expand_asm_memory_barrier): Rename ...
300 (expand_asm_memory_blockage): ... to this.
301 (expand_mem_thread_fence): Call expand_memory_blockage
302 instead of expand_asm_memory_barrier.
303 (expand_mem_singnal_fence): Ditto.
304 (expand_atomic_load): Ditto.
305 (expand_atomic_store): Ditto.
306 * doc/md.texi (Standard Pattern Names For Generation):
307 Document memory_blockage instruction pattern.
309 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
311 * config/rl78/rl78.c (rl78_emit_libcall): New function.
312 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
313 * config/rl78/rl78.md: New define_expand "adddi3".
315 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
317 * cfghooks.c (verify_flow_info): Disable check that all probabilities
320 2017-10-13 Jeff Law <law@redhat.com>
322 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
324 2017-10-13 Jakub Jelinek <jakub@redhat.com>
327 * internal-fn.c (expand_mul_overflow): If both operands have
328 the same highpart of -1 or 0 and the topmost bit of lowpart
329 is different, overflow is if res <= 0 rather than res < 0.
331 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
333 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
334 TARGET_P9_VECTOR code for unaligned_load case.
336 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
338 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
340 2017-10-13 Nathan Sidwell <nathan@acm.org>
342 * tree-core.h (tree_contains_struct): Make bool.
343 * tree.c (tree_contains_struct): Likewise.
344 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
345 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
346 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
347 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
348 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
350 2017-10-13 Richard Biener <rguenther@suse.de>
352 * graphite-isl-ast-to-gimple.c
353 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
354 parameters and dominance check.
355 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
356 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
357 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
358 Do not update SSA form here or do intermediate IL verification.
359 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
360 (graphite_initialize): Remove check on the number of loops in
361 the function and inline into graphite_transform_loops.
362 (graphite_finalize): Inline into graphite_transform_loops.
363 (graphite_transform_loops): Perform SSA update and IL verification
365 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
367 2017-10-13 Richard Biener <rguenther@suse.de>
369 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
370 graphite_expression_type_precision): Avoid global constructor
372 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
373 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
374 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
375 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
376 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
377 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
379 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
382 * config/i386/i386.h (ix86_red_zone_size): New.
383 * config/i386/i386.md (push peephole2s): Replace
384 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
386 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
387 Alan Hayward <alan.hayward@arm.com>
388 David Sherwood <david.sherwood@arm.com>
390 * combine.c (can_change_dest_mode): Reject changes in
391 REGMODE_NATURAL_SIZE.
393 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
394 Alan Hayward <alan.hayward@arm.com>
395 David Sherwood <david.sherwood@arm.com>
397 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
398 (expand_debug_source_expr): Likewise.
399 * combine.c (combine_simplify_rtx): Likewise.
400 * cse.c (fold_rtx): Likewise.
401 * fwprop.c (canonicalize_address): Likewise.
402 * targhooks.c (default_shift_truncation_mask): Likewise.
404 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
405 Alan Hayward <alan.hayward@arm.com>
406 David Sherwood <david.sherwood@arm.com>
408 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
409 (widened_mode): Likewise.
410 (expand_unop): Likewise.
411 * ree.c (transform_ifelse): Likewise.
412 (merge_def_and_ext): Likewise.
413 (combine_reaching_defs): Likewise.
414 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
416 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
417 Alan Hayward <alan.hayward@arm.com>
418 David Sherwood <david.sherwood@arm.com>
420 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
421 * combine.c (gen_lowpart_for_combine): Likewise.
422 * dwarf2out.c (rtl_for_decl_location): Likewise.
423 * final.c (alter_subreg): Likewise.
424 * rtlhooks.c (gen_lowpart_general): Likewise.
425 (gen_lowpart_if_possible): Likewise.
427 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
428 Alan Hayward <alan.hayward@arm.com>
429 David Sherwood <david.sherwood@arm.com>
431 * calls.c (expand_call): Use subreg_lowpart_offset.
432 * cse.c (cse_insn): Likewise.
433 * regcprop.c (copy_value): Likewise.
434 (copyprop_hardreg_forward_1): Likewise.
436 2017-10-13 Jakub Jelinek <jakub@redhat.com>
439 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
440 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
441 =Q constraints to +Q and into insn condition add check
442 that operands[0] and operands[1] are equal.
443 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
444 =Q constraints to +Q and into insn condition add check
445 that operands[0] is equal to either operands[1] or operands[2].
448 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
449 instead of handling MINUS_EXPR twice (once for each argument),
450 canonicalize operand order and handle just once, use rtype where
451 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
454 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
455 any values of __C while still being pattern recognizable as a simple
458 2017-10-13 Richard Biener <rguenther@suse.de>
460 PR tree-optimization/82451
462 2017-10-02 Richard Biener <rguenther@suse.de>
464 PR tree-optimization/82355
465 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
466 a mapping for the enclosing loop but avoid generating one for
468 (copy_bb_and_scalar_dependences): Remove premature codegen
469 error on PHIs in blocks duplicated into multiple places.
470 * graphite-scop-detection.c
471 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
472 in the region use it as loop and nest to analyze the DR in.
473 (try_generate_gimple_bb): Likewise.
474 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
475 (add_loop_constraints): For blocks in a loop not in the region
476 create a dimension with a single iteration.
477 * sese.h (gbb_loop_at_index): Remove assert.
479 * cfgloop.c (loop_preheader_edge): For the loop tree root
480 return the single successor of the entry block.
481 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
482 Reset the SCEV hashtable and niters.
483 * graphite-scop-detection.c
484 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
485 assert that we only have POLYNOMIAL_CHREC that vary in loops
486 contained in the region.
487 (scop_detection::graphite_can_represent_expr): Adjust.
488 (scop_detection::stmt_has_simple_data_refs_p): For loops
489 not in the region set loop to NULL. The nest is now the
490 entry edge to the region.
491 (try_generate_gimple_bb): Likewise.
492 * sese.c (scalar_evolution_in_region): Adjust for
493 instantiate_scev change.
494 * tree-data-ref.h (graphite_find_data_references_in_stmt):
495 Make nest parameter the edge into the region.
496 (create_data_ref): Likewise.
497 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
498 entry edge into a region and adjust instantiate_scev calls.
499 (create_data_ref): Likewise.
500 (graphite_find_data_references_in_stmt): Likewise.
501 (find_data_references_in_stmt): Pass the loop preheader edge
502 from the nest argument.
503 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
504 parameter the edge into the region.
505 (instantiate_parameters): Use the loop preheader edge as entry.
506 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
508 (get_instantiated_value_entry): Make instantiate_below parameter
509 the edge into the region.
510 (instantiate_scev_name): Likewise. Adjust dominance checks,
511 when we cannot use loop-based instantiation instantiate by
512 walking use-def chains.
513 (instantiate_scev_poly): Adjust.
514 (instantiate_scev_binary): Likewise.
515 (instantiate_scev_convert): Likewise.
516 (instantiate_scev_not): Likewise.
517 (instantiate_array_ref): Remove.
518 (instantiate_scev_3): Likewise.
519 (instantiate_scev_2): Likewise.
520 (instantiate_scev_1): Likewise.
521 (instantiate_scev_r): Do not blindly handle N-operand trees.
522 Do not instantiate array-refs. Handle all constants and invariants.
523 (instantiate_scev): Make instantiate_below parameter
524 the edge into the region.
525 (resolve_mixers): Use the loop preheader edge for the region
526 parameter to instantiate_scev_r.
527 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
529 2017-10-13 Richard Biener <rguenther@suse.de>
531 PR tree-optimization/82525
532 * graphite-isl-ast-to-gimple.c
533 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
535 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
536 Fail code generation when we cannot represent the isl integer.
537 (binary_op_to_tree): Elide modulo operations that are no-ops
538 in the type we code generate. Remove now superfluous code
541 2017-10-13 Richard Biener <rguenther@suse.de>
543 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
544 (scop_detection::harmful_loop_in_region): Remove premature
546 (scop_detection::graphite_can_represent_scev): We can handle
547 pointer IVs just fine.
549 2017-10-13 Alan Modra <amodra@gmail.com>
551 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
552 "Clobbers and Scratch Registers". Add paragraph on
553 alternative to clobbers for scratch registers and OpenBLAS
556 2017-10-13 Alan Modra <amodra@gmail.com>
558 * doc/extend.texi (Clobbers): Correct vax example. Delete old
559 example of a memory input for a string of known length. Move
560 commentary out of table. Add a number of new examples
561 covering array memory inputs.
563 2017-10-12 Martin Liska <mliska@suse.cz>
565 PR tree-optimization/82493
566 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
567 (test_range_functions): New function.
568 (sbitmap_c_tests): Likewise.
569 * selftest-run-tests.c (selftest::run_tests): Run new tests.
570 * selftest.h (sbitmap_c_tests): New function.
572 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
574 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
576 * config/rs6000/amo.h: Fix spacing issue.
578 2017-10-12 Jakub Jelinek <jakub@redhat.com>
581 * config/i386/i386.md (*ashl<mode>3_mask_1,
582 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
583 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
586 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
588 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
589 (profile_probability): Set max_probability
590 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
592 * profile-count.c (profile_probability::differs_from_p): Do not
593 rely on max_probaiblity == 10000
595 2017-10-12 Jeff Law <law@redhat.com>
597 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
600 2017-10-12 Martin Sebor <msebor@redhat.com>
604 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
605 (handle_alias_pairs): Call it.
606 * common.opt (-Wattribute-alias): New option.
607 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
608 * doc/invoke.texi (-Wattribute-alias): Document.
610 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
613 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
615 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
617 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
618 (make_hard_regno_born, make_hard_regno_dead): Update
619 bb_killed_pseudos and bb_gen_pseudos.
621 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
623 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
625 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
627 * config/alpha/alpha.c (alpha_split_conditional_move):
628 Use std::swap instead of manually swapping.
629 (alpha_stdarg_optimize_hook): Ditto.
630 (alpha_canonicalize_comparison): Ditto.
632 2017-10-12 Bin Cheng <bin.cheng@arm.com>
634 * tree-loop-distribution.c (struct builtin_info): New struct.
635 (struct partition): Refactor fields into struct builtin_info.
636 (partition_free): Free struct builtin_info.
637 (build_size_arg_loc, build_addr_arg_loc): Delete.
638 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
639 information from struct builtin_info.
640 (find_single_drs): New function refactored from classify_partition.
641 Also moved builtin validity checks to this function.
642 (compute_access_range, alloc_builtin): New functions.
643 (classify_builtin_st, classify_builtin_ldst): New functions.
644 (classify_partition): Refactor code into functions find_single_drs,
645 classify_builtin_st and classify_builtin_ldst.
646 (distribute_loop): Don't do runtime alias check when distributing
648 (find_seed_stmts_for_distribution): New function.
649 (pass_loop_distribution::execute): Refactor code finding seed
650 stmts into above function. Support distribution for the innermost
651 two-level loop nest. Adjust dump information.
653 2017-10-12 Bin Cheng <bin.cheng@arm.com>
655 * tree-loop-distribution.c: Adjust the general comment.
656 (NUM_PARTITION_THRESHOLD): New macro.
657 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
658 (classify_partition): Skip builtin pattern of loop nest's inner loop.
659 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
660 in call to build_partition_graph.
661 (finalize_partitions): New parameter. Make loop distribution more
662 conservative by fusing more partitions.
663 (distribute_loop): Don't do runtime alias check in case of loop nest
665 (find_seed_stmts_for_distribution): New function.
666 (prepare_perfect_loop_nest): New function.
667 (pass_loop_distribution::execute): Refactor code finding seed stmts
668 and loop nest into above functions. Support loop nest distribution.
669 Adjust dump information accordingly.
671 2017-10-12 Bin Cheng <bin.cheng@arm.com>
673 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
674 and set PTYPE_SEQUENTIAL for merged partition.
676 2017-10-12 Richard Biener <rguenther@suse.de>
678 PR tree-optimization/69728
680 2017-09-19 Richard Biener <rguenther@suse.de>
682 PR tree-optimization/69728
683 * graphite-sese-to-poly.c (schedule_error): New global.
684 (add_loop_schedule): Handle empty domain by failing the
686 (build_original_schedule): Handle schedule_error.
688 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
689 domain by returning an unchanged schedule.
691 2017-10-12 Jakub Jelinek <jakub@redhat.com>
693 * genrecog.c (validate_pattern): For VEC_SELECT verify that
694 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
696 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
698 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
701 2017-10-12 Jakub Jelinek <jakub@redhat.com>
704 * expr.c (store_field): Don't optimize away bitsize == 0 store
705 from CALL_EXPR with addressable return type.
707 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
709 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
710 * config/rs6000/rs6000.md (sel): Delete mode attribute.
711 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
712 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
713 TARGET_ISEL instead of TARGET_ISEL<sel>.
715 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
717 * config/rs6000/rs6000.c
718 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
720 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
722 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
724 (reg_or_cint_operand): Fix comment.
725 (reg_or_zero_operand): New predicate.
726 * config/rs6000/rs6000-protos.h (output_isel): Delete.
727 * config/rs6000/rs6000.c (output_isel): Delete.
728 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
729 instead of reg_or_cint_operand. Output instruction directly (not via
731 (isel_unsigned_<mode>): Ditto.
732 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
733 gpc_reg_operand. Add an instruction alternative for this. Output
734 instruction directly.
735 (*isel_reversed_unsigned_<mode>): Ditto.
737 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
739 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
740 (TARGET_CANONICALIZE_COMPARISON): Define.
742 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
745 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
746 Check whether the dest is REG before adding REG_EQUIV note.
748 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
751 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
753 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
754 (make_hard_regno_born, make_hard_regno_dead): Update
755 bb_killed_pseudos and bb_gen_pseudos.
757 2017-10-11 Nathan Sidwell <nathan@acm.org>
759 * incpath.h (enum incpath_kind): Name enum, prefix values.
760 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
761 * incpath.c (heads, tails): Use INC_MAX.
762 (add_env_var_paths, add_standard_paths): Use incpath_kind.
763 (merge_include_chains, split_quote_chain,
764 register_include_chains): Update incpath_kind names.
765 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
766 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
768 (add_framework_path, darwin_register_objc_includes): Likewise.
769 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
771 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
773 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
774 Do not use float_operator operator predicate.
775 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
776 * config/i386/predicates.md (float_operator): Remove predicate.
778 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
780 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
781 (*jccxf_i387): Ditto.
782 (*jcc<mode>_i387): Ditto.
783 (*jccu<mode>_i387): Ditto.
784 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
785 (*jcc_*_i387 splitters): Remove.
786 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
787 * config/i386/i386.c (ix86_split_fp_branch): Remove.
788 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
791 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
793 * profile-count.h (slow_safe_scale_64bit): New function.
794 (safe_scale_64bit): New inline.
795 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
796 * profile-count.c: Include wide-int.h
797 (slow_safe_scale_64bit): New.
799 2017-10-11 Nathan Sidwell <nathan@acm.org>
801 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
802 HAS_DECL_ASSEMBLER_NAME_P.
803 * gimple-expr.c (gimple_decl_printable_name: Check
804 HAS_DECL_ASSEMBLER_NAME_P too.
805 * ipa-utils.h (type_in_anonymous_namespace_p): Check
806 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
807 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
808 * passes.c (rest_of_decl_compilation): Check
809 HAS_DECL_ASSEMBLER_NAME_P too.
810 * recog.c (verify_changes): Likewise.
811 * tree-pretty-print.c (dump_decl_name): Likewise.
812 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
814 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
815 (SET_DECL_ASSEMBLER_NAME): Use it.
816 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
817 (COPY_DECL_ASSEMBLER_NAME): Likewise.
818 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
820 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
822 * config.gcc (i386, x86_64): Add extra objects.
823 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
824 (ix86_min_insn_size): Declare.
825 (ix86_issue_rate): Declare.
826 (ix86_adjust_cost): Declare.
827 (ia32_multipass_dfa_lookahead): Declare.
828 (ix86_macro_fusion_p): Declare.
829 (ix86_macro_fusion_pair_p): Declare.
830 (ix86_bd_has_dispatch): Declare.
831 (ix86_bd_do_dispatch): Declare.
832 (ix86_core2i7_init_hooks): Declare.
833 (ix86_atom_sched_reorder): Declare.
834 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
835 (COSTS_N_BYTES): Move to x86-tune-costs.h.
836 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
837 (rip_relative_addr_p): Rename to ...
838 (ix86_rip_relative_addr_p): ... this one; export.
839 (memory_address_length): Update.
840 (ix86_issue_rate): Move to x86-tune-sched.c.
841 (ix86_flags_dependent): Move to x86-tune-sched.c.
842 (ix86_agi_dependent): Move to x86-tune-sched.c.
843 (exact_dependency_1): Move to x86-tune-sched.c.
844 (exact_store_load_dependency): Move to x86-tune-sched.c.
845 (ix86_adjust_cost): Move to x86-tune-sched.c.
846 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
847 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
848 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
849 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
850 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
851 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
852 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
853 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
854 (min_insn_size): Rename to ...
855 (ix86_min_insn_size): ... this one; export.
856 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
857 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
858 (core2i7_first_cycle_multipass_backtrack): Move to
859 x86-tune-sched-core.c.
860 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
861 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
862 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
863 (ix86_avoid_jump_mispredicts): Update.
864 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
865 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
866 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
867 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
868 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
869 (MAX_INSN): Move to ix86-tune-sched-bd.c.
870 (MAX_IMM): Move to ix86-tune-sched-bd.c.
871 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
872 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
873 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
874 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
875 (MAX_STORE): Move to ix86-tune-sched-bd.c.
876 (BIG): Move to ix86-tune-sched-bd.c.
877 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
878 (enum insn_path): Move to ix86-tune-sched-bd.c.
879 (get_mem_group): Move to ix86-tune-sched-bd.c.
880 (is_cmp): Move to ix86-tune-sched-bd.c.
881 (dispatch_violation): Move to ix86-tune-sched-bd.c.
882 (is_branch): Move to ix86-tune-sched-bd.c.
883 (is_prefetch): Move to ix86-tune-sched-bd.c.
884 (init_window): Move to ix86-tune-sched-bd.c.
885 (allocate_window): Move to ix86-tune-sched-bd.c.
886 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
887 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
888 (process_end_window): Move to ix86-tune-sched-bd.c.
889 (allocate_next_window): Move to ix86-tune-sched-bd.c.
890 (find_constant): Move to ix86-tune-sched-bd.c.
891 (get_num_immediates): Move to ix86-tune-sched-bd.c.
892 (has_immediate): Move to ix86-tune-sched-bd.c.
893 (get_insn_path): Move to ix86-tune-sched-bd.c.
894 (get_insn_group): Move to ix86-tune-sched-bd.c.
895 (count_num_restricted): Move to ix86-tune-sched-bd.c.
896 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
897 (add_insn_window): Move to ix86-tune-sched-bd.c.
898 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
899 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
900 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
901 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
902 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
903 (do_dispatch): Move to ix86-tune-sched-bd.c.
904 (has_dispatch): Move to ix86-tune-sched-bd.c.
905 * config/i386/t-i386: Add new object files.
906 * config/i386/x86-tune-costs.h: New file.
907 * config/i386/x86-tune-sched-atom.c: New file.
908 * config/i386/x86-tune-sched-bd.c: New file.
909 * config/i386/x86-tune-sched-core.c: New file.
910 * config/i386/x86-tune-sched.c: New file.
912 2017-10-11 Liu Hao <lh_mouse@126.com>
914 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
915 the generic version below instead.
916 (should_colorize): Recognize Windows consoles as terminals
918 * pretty-print.c [__MINGW32__] (write_all): New function.
919 [__MINGW32__] (find_esc_head): Likewise.
920 [__MINGW32__] (find_esc_terminator): Likewise.
921 [__MINGW32__] (eat_esc_sequence): Likewise.
922 [__MINGW32__] (mingw_ansi_fputs): New function that handles
924 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
927 2017-10-11 Richard Biener <rguenther@suse.de>
929 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
930 Properly call analyze_scalar_evolution with the loop of the stmt.
932 2017-10-11 Richard Biener <rguenther@suse.de>
934 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
935 * tree-core.h (tree_base): Add chrec_var union member.
936 * tree.h (CHREC_VAR): Remove.
937 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
938 * tree-chrec.h (build_polynomial_chrec): Adjust.
939 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
940 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
942 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
944 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
945 * match.pd: ... here.
946 ((T) X == (T) Y): Relax condition.
948 2017-10-11 Bin Cheng <bin.cheng@arm.com>
950 PR tree-optimization/82472
951 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
953 (break_alias_scc_partitions): Update postorder number.
955 2017-10-11 Martin Liska <mliska@suse.cz>
958 * opts.c (parse_no_sanitize_attribute): Do not use error_value
960 * opts.h (parse_no_sanitize_attribute): Remove last argument.
962 2017-10-11 Martin Liska <mliska@suse.cz>
964 * print-rtl.c (print_insn): Move declaration of idbuf
965 to same scope as name.
967 2017-10-11 Martin Liska <mliska@suse.cz>
972 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
975 2017-10-11 Martin Liska <mliska@suse.cz>
978 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
981 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
983 * config.gcc (default_gnu_indirect_function): Default to yes for
984 arm*-*-linux* with glibc.
986 2017-10-11 Richard Biener <rguenther@suse.de>
988 * tree-scalar-evolution.c (get_scalar_evolution): Handle
989 default-defs and types we do not want to analyze.
990 (interpret_loop_phi): Replace unreachable code with an assert.
991 (compute_scalar_evolution_in_loop): Remove and inline ...
992 (analyze_scalar_evolution_1): ... here, replacing condition with
993 what makes the intent clearer. Remove handling of cases
994 get_scalar_evolution now handles.
996 2017-10-10 Jim Wilson <wilson@tuliptree.org>
998 PR rtl-optimization/81434
999 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
1000 comment for main loop. In sched_group_found if, also add checks for
1001 pass and min_cost_group.
1003 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
1005 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
1006 (rs6000_insn_cost): New function.
1007 * config/rs6000/rs6000.md (cost): New attribute.
1009 2017-10-10 Jakub Jelinek <jakub@redhat.com>
1010 H.J. Lu <hongjiu.lu@intel.com>
1014 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
1015 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
1016 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
1017 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
1018 to OPTION_MASK_ISA_AVX512VL - builtins that have both
1019 OPTION_MASK_ISA_MMX and some other bit set require both
1020 mmx and the ISAs without the mmx bit.
1021 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
1022 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
1023 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
1024 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
1025 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
1026 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
1027 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
1028 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
1029 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
1030 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
1031 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
1032 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
1033 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
1034 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
1035 Add OPTION_MASK_ISA_MMX.
1037 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
1039 * config.gcc (armv7*-*-freebsd*): New target.
1040 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
1042 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
1044 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
1045 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
1048 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
1050 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
1051 a template parameter.
1052 (WIDE_INT_REF_FOR): Update accordingly.
1053 * tree.h (wi::int_traits <const_tree>): Delete.
1054 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
1055 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
1056 (wi::tree_to_wide_ref): New typedef.
1057 (wi::to_wide): New function.
1058 * calls.c (get_size_range): Use wi::to_wide when operating on
1060 * cgraph.c (cgraph_node::create_thunk): Likewise.
1061 * config/i386/i386.c (ix86_data_alignment): Likewise.
1062 (ix86_local_alignment): Likewise.
1063 * dbxout.c (stabstr_O): Likewise.
1064 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
1065 * expr.c (const_vector_from_tree): Likewise.
1066 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
1067 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
1068 (fold_negate_expr_1, int_const_binop_1, const_binop)
1069 (fold_convert_const_int_from_real, optimize_bit_field_compare)
1070 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
1071 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
1072 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
1073 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
1074 (fold_not_const, round_up_loc): Likewise.
1075 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
1076 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
1077 (alloca_call_type): Likewise.
1078 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
1079 * godump.c (go_output_typedef): Likewise.
1080 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
1081 * internal-fn.c (get_min_precision): Likewise.
1082 * ipa-cp.c (ipcp_store_vr_results): Likewise.
1083 * ipa-polymorphic-call.c
1084 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1085 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
1086 (ipa_modify_call_arguments): Likewise.
1087 * match.pd: Likewise.
1088 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
1089 * print-tree.c (print_node_brief, print_node): Likewise.
1090 * stmt.c (expand_case): Likewise.
1091 * stor-layout.c (layout_type): Likewise.
1092 * tree-affine.c (tree_to_aff_combination): Likewise.
1093 * tree-cfg.c (group_case_labels_stmt): Likewise.
1094 * tree-data-ref.c (dr_analyze_indices): Likewise.
1095 (prune_runtime_alias_test_list): Likewise.
1096 * tree-dump.c (dequeue_and_dump): Likewise.
1097 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
1098 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
1099 * tree-pretty-print.c (dump_generic_node): Likewise.
1100 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
1101 (simple_iv_with_niters): Likewise.
1102 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1103 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
1104 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
1105 * tree-ssa-loop-niter.c (split_to_var_and_offset)
1106 (refine_value_range_using_guard, number_of_iterations_ne_max)
1107 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
1108 (get_cst_init_from_scev, record_nonwrapping_iv)
1109 (scev_var_range_cant_overflow): Likewise.
1110 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
1111 * tree-ssa-pre.c (compute_avail): Likewise.
1112 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1113 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
1114 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
1115 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
1116 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
1117 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
1118 (dump_case_nodes, try_switch_expansion): Likewise.
1119 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
1120 (vect_do_peeling): Likewise.
1121 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
1122 * tree-vect-stmts.c (vectorizable_load): Likewise.
1123 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
1124 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
1125 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
1126 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
1127 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
1128 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
1129 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
1130 (evrp_dom_walker::before_dom_children): Likewise.
1131 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
1132 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
1133 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
1134 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
1135 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
1136 (get_range_pos_neg): Likewise.
1137 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
1138 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
1139 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
1140 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
1141 * config/avr/avr.c (avr_fold_builtin): Likewise.
1142 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
1143 * config/msp430/msp430.c (msp430_attr): Likewise.
1144 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
1145 * config/powerpcspe/powerpcspe-c.c
1146 (altivec_resolve_overloaded_builtin): Likewise.
1147 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
1148 (rs6000_expand_ternop_builtin): Likewise.
1149 * config/rs6000/rs6000-c.c
1150 (altivec_resolve_overloaded_builtin): Likewise.
1151 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
1152 (rs6000_expand_ternop_builtin): Likewise.
1153 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
1155 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1157 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
1158 when copying loop nest with only one inner loop.
1160 2017-10-10 Richard Biener <rguenther@suse.de>
1162 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
1163 blocks if SCEV is active.
1164 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
1166 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
1167 (scev_initialize): Assert we are not yet initialized.
1169 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1171 * tree-loop-distribution.c (generate_loops_for_partition): Remove
1172 inner loop's exit stmt by making it always exit the loop, otherwise
1173 we would generate an infinite empty loop.
1175 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1177 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
1178 renaming variables in new preheader if it's deleted.
1180 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1182 * tree-loop-distribution.c (struct partition): Remove unused field
1183 loops of the structure.
1184 (partition_alloc, partition_free): Ditto.
1185 (build_rdg_partition_for_vertex): Ditto.
1187 2017-10-09 Jeff Law <law@redhat.com>
1189 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
1190 return type to match prototype and documentation.
1192 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1194 * config/rs6000/rs6000.c (processor_costs): Move to ...
1195 * config/rs6000/rs6000.h: ... here.
1196 (rs6000_cost): Declare.
1198 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
1200 * except.c (setjmp_fn): New global variable.
1201 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
1202 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
1203 if DONT_USE_BUILTIN_SETJMP is defined.
1205 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1207 * target.def (insn_cost): New hook.
1208 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
1209 * doc/tm.texi: Regenerate.
1210 * rtlanal.c (insn_cost): Use the new hook.
1212 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1214 * combine.c (combine_validate_cost): Compute the new insn_cost,
1215 not just pattern_cost.
1216 (try_combine): Adjust comment.
1218 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1220 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
1222 * combine.c (uid_insn_cost): Adjust comment.
1223 (combine_validate_cost): Adjust comment. Use pattern_cost instead
1225 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
1226 * dse.c (find_shift_sequence): Ditto.
1227 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
1228 (bb_valid_for_noce_process_p): Use pattern_cost.
1229 * rtl.h (insn_rtx_cost): Delete.
1230 (pattern_cost): New prototype.
1231 (insn_cost): New prototype.
1232 * rtlanal.c (insn_rtx_cost): Rename to...
1233 (pattern_cost): ... this.
1236 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
1238 * config/i386/i386.md (*jcc_2): Remove insn pattern.
1239 (*jcc<mode>_0_r_i387): Ditto.
1240 (*jccxf_r_i387): Ditto.
1241 (*jcc<mode>_r_i387): Ditto.
1242 (*jccu<mode>_r_i387): Ditto.
1243 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
1244 (*jcc): Rename from *jcc_1.
1246 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1248 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
1249 deferred rescans after the lvx/stvx recombination pre-pass.
1251 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1253 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
1254 memory operation instruction support.
1255 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
1256 (rs6000-ibm-aix[789]*): Likewise.
1257 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
1258 Document new functions.
1260 2017-10-09 Richard Biener <rguenther@suse.de>
1262 PR tree-optimization/82397
1263 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
1264 equality only for semantically equal trees.
1266 2017-10-09 Richard Biener <rguenther@suse.de>
1268 PR tree-optimization/82449
1269 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
1270 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
1271 allow constant addresses.
1272 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
1275 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1277 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
1280 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1283 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
1286 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1289 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
1291 2017-10-09 Jakub Jelinek <jakub@redhat.com>
1294 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
1295 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
1297 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
1299 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
1300 (WI_BINARY_PREDICATE_RESULT): Likewise.
1301 (wi::binary_traits::operator_result): New type.
1302 (wi::binary_traits::predicate_result): Likewise.
1303 (generic_wide_int::operator~, unary generic_wide_int::operator-)
1304 (generic_wide_int::operator==, generic_wide_int::operator!=)
1305 (generic_wide_int::operator&, generic_wide_int::and_not)
1306 (generic_wide_int::operator|, generic_wide_int::or_not)
1307 (generic_wide_int::operator^, generic_wide_int::operator+
1308 (binary generic_wide_int::operator-, generic_wide_int::operator*):
1310 (operator~, unary operator-, operator==, operator!=, operator&)
1311 (operator|, operator^, operator+, binary operator-, operator*): New
1313 * expr.c (get_inner_reference): Use wi::bit_and_not.
1314 * fold-const.c (fold_binary_loc): Likewise.
1315 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
1316 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
1317 (bit_value_binop): Likewise.
1318 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
1319 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
1320 (extract_range_from_binary_expr_1): Likewise.
1321 (masked_increment): Likewise.
1322 (simplify_bit_ops_using_ranges): Likewise.
1324 2017-10-09 Martin Jambor <mjambor@suse.cz>
1327 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
1328 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
1329 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
1331 (hsa_fixup_mov_insn_type): New function.
1332 (hsa_op_with_type::get_in_type): Use it.
1333 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
1334 immediates in an assert.
1335 (hsa_op_with_type::extend_int_to_32bit): New method.
1336 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
1337 types. Convert to dest type if necessary.
1338 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
1339 (reg_for_gimple_ssa): Pass false as min32int to
1340 hsa_type_for_scalar_tree_type.
1341 (gen_hsa_addr): Fixup type when creating addresable temporary.
1342 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
1343 (gen_hsa_unary_operation): Extend operands and convert to dest type if
1344 necessary. Call hsa_fixup_mov_insn_type.
1345 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
1346 extend operands and convert to dest type if necessary.
1347 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
1348 to dest type if necessary.
1349 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
1350 if conversion nt necessary and size matches.
1351 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
1352 to dest type if necessary.
1353 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
1354 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
1356 (gen_hsa_clrsb): Likewise.
1357 (gen_hsa_ffs): Likewise.
1358 (gen_hsa_divmod): Extend operands and convert to dest type if
1360 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
1362 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1364 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
1365 Remove empty default arguments. Use a brace block as output
1367 (conditional return): Ditto.
1369 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
1370 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
1372 (group_ending_nop): Ditto.
1373 (doloop_end): Ditto.
1374 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
1375 (splitters for those): Ditto.
1377 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1379 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
1380 a conditional jump (and the compare for it) so that pc_rtx is the
1382 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
1383 for the deleted and renamed ctr<mode>_internal[234] patterns.
1384 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
1385 Delete second conditional return pattern.
1386 (ctr<mode>_internal2): Delete this second bdnz pattern.
1387 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
1388 (ctr<mode>_internal4): Delete this second bdz pattern.
1390 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1392 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
1393 (always_initialized_rtx_for_ssa_name_p): New predicate.
1394 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
1395 (finish_out_of_ssa): Free new field of SA.
1396 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
1397 * tree-ssa-coalesce.c: Include tree-ssa.h.
1398 (get_parm_default_def_partitions): Remove extern keyword.
1399 (get_undefined_value_partitions): New function.
1400 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
1401 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
1403 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
1405 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1407 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
1409 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1411 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
1412 for targets that preffer 128bit.
1414 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1416 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
1418 2017-10-08 Olivier Hainque <hainque@adacore.com>
1420 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
1421 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
1423 (thumb_set_return_address): Likewise.
1425 2017-10-08 Olivier Hainque <hainque@adacore.com>
1427 * common/config/arm/arm-common.c (arm_except_unwind_info):
1428 Handle DWARF2_UNWIND_INFO.
1430 2017-10-07 Michael Collison <michael.collison@arm.com>
1432 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
1435 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
1437 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
1438 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
1439 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
1440 defined, force the creation of a new block for a dispatch label.
1442 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
1444 * invoke.texi (Wsuggest-attribute=cold): Document.
1445 * common.opt (Wsuggest-attribute=cold): New
1446 * ipa-pure-const.c (warn_function_cold): New function.
1447 * predict.c (compute_function_frequency): Use it.
1448 * predict.h (warn_function_cold): Declare.
1450 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
1452 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
1455 2017-10-06 Martin Liska <mliska@suse.cz>
1457 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
1458 keyword for member functions.
1459 (struct sanopt_tree_couple): New struct.
1460 (struct sanopt_tree_couple_hash): New function.
1461 (struct sanopt_ctx): Add new hash_map.
1462 (has_dominating_ubsan_ptr_check): New function.
1463 (record_ubsan_ptr_check_stmt): Likewise.
1464 (maybe_optimize_ubsan_ptr_ifn): Likewise.
1465 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
1466 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
1468 2017-10-06 Sudakshina Das <sudi.das@arm.com>
1471 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
1472 aarch64_simd_valid_immediate on CONST_VECTORs.
1473 (aarch64_reg_or_bic_imm): Likewise.
1475 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
1477 PR rtl-optimization/82396
1478 * haifa-sched.c (ready_sort_real): Disable qsort checking.
1480 2017-10-06 Sebastian Pop <sebpop@gmail.com>
1482 * graphite-dependences.c (scop_get_reads): Move code to...
1483 (scop_get_must_writes): Move code to...
1484 (scop_get_may_writes): Move code to...
1485 (scop_get_reads_and_writes): ... here.
1486 (scop_get_dependences): Call scop_get_reads_and_writes.
1488 2017-10-06 Jakub Jelinek <jakub@redhat.com>
1490 PR tree-optimization/82434
1491 * fold-const.h (can_native_encode_type_p,
1492 can_native_encode_string_p): Remove.
1493 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
1494 don't encode anything, just return what would be otherwise returned.
1495 (native_encode_fixed, native_encode_complex, native_encode_vector):
1497 (native_encode_string): Likewise. Inline by hand
1498 can_native_encode_string_p.
1499 (can_native_encode_type_p): Remove.
1500 (can_native_encode_string_p): Remove.
1501 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
1502 STRING_CSTs using can_native_encode_string_p, test all
1503 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
1504 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
1505 argument from native_encode_expr.
1506 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
1507 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
1510 2017-10-06 Richard Biener <rguenther@suse.de>
1512 PR tree-optimization/82397
1513 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
1514 operand_equal_p but rely on data_ref_compare_tree for detecting
1516 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
1517 to match up with dr_group_sort_cmp.
1519 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1522 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
1524 * config/s390/s390-builtin-types.def: Regenerate.
1526 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1529 * config/s390/s390-builtin-types.def: Regenerate.
1530 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
1531 Change flag from B_VXE to B_VX.
1532 (s390_vec_min_dbl): Remove B_VXE flag.
1534 2017-10-06 Richard Biener <rguenther@suse.de>
1536 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
1537 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
1538 translate_isl_ast_to_gimple::is_valid_rename,
1539 translate_isl_ast_to_gimple::get_rename,
1540 translate_isl_ast_to_gimple::get_def_bb_for_const,
1541 translate_isl_ast_to_gimple::get_new_name,
1542 translate_isl_ast_to_gimple::collect_all_ssa_names,
1543 translate_isl_ast_to_gimple::copy_loop_phi_args,
1544 translate_isl_ast_to_gimple::collect_all_ssa_names,
1545 translate_isl_ast_to_gimple::copy_loop_phi_args,
1546 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
1547 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
1548 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
1549 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
1550 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
1551 translate_isl_ast_to_gimple::copy_cond_phi_args,
1552 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
1553 translate_isl_ast_to_gimple::edge_for_new_close_phis,
1554 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
1555 translate_isl_ast_to_gimple::rename_uses,
1556 translate_isl_ast_to_gimple::rename_all_uses): Remove.
1557 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
1558 (set_rename_for_each_def): Likewise.
1559 (graphite_copy_stmts_from_block): Handle debug stmt resetting
1560 here. Handle rewriting SCEV analyzable uses here.
1561 (copy_bb_and_scalar_dependences): Generate code for PHI
1563 (graphite_regenerate_ast_isl): Adjust.
1564 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
1565 (add_write, add_read): New functions.
1566 (build_cross_bb_scalars_def): Use it and simplify.
1567 (build_cross_bb_scalars_use): Likewise.
1568 (graphite_find_cross_bb_scalar_vars): Inline into...
1569 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
1570 simulating out-of-SSA. Compute liveout and add dependencies.
1571 (build_scops): Force an empty entry block.
1572 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
1574 (sese_build_liveouts): Declare.
1575 (sese_trivially_empty_bb_p): Likewise.
1576 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
1577 compute liveout and debug_liveout.
1578 (sese_bad_liveouts_use): Remove.
1579 (sese_reset_debug_liveouts_bb): Likewise.
1580 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
1581 (sese_build_liveouts): Build liveout and debug_liveout and store
1583 (new_sese_info): Adjust.
1584 (free_sese_info): Likewise.
1585 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
1586 do not build liveout here.
1587 (move_sese_in_condition): Adjust region entry.
1588 (scev_analyzable_p): Match up with chrec_apply requirements.
1589 (sese_trivially_empty_bb_p): New.
1590 * tree-into-ssa.c (get_reaching_def): Properly support generating
1591 default-defs for incremental rewrite of anonymous names.
1593 2017-10-06 Richard Biener <rguenther@suse.de>
1595 * graphite-sese-to-poly.c (extract_affine): For casts increasing
1596 precision do not perform modulo reduction.
1598 2017-10-06 Richard Biener <rguenther@suse.de>
1600 PR tree-optimization/82436
1601 * tree-vect-slp.c (vect_supported_load_permutation_p): More
1602 conservatively choose the vectorization factor when checking
1603 whether we can perform the required load permutation.
1604 (vect_transform_slp_perm_load): Assert when we may not fail.
1606 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1608 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
1609 message for incompatible -msdata=* and -mcall-* options.
1611 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1613 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
1614 rate for post-reload scheduling.
1616 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1618 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
1620 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1622 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
1623 to improve monte carlo in scimark.
1625 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1627 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
1628 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
1629 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
1630 pentium4_cost, nocona_cost): Set reassociation width to 1.
1631 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
1632 width to 2 for fp operations and 1 otherwise.
1633 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
1635 (atom_cost): Set reassociation width to 2.
1636 (slm_cost, generic_cost): Set fp reassociation width
1637 to 2 and 1 otherwise.
1638 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
1639 (core_cost): Set fp reassociation width to 4 and vector to 2.
1640 (ix86_reassociation_width): Rewrite using cost table; special case
1641 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
1642 and TARGET_AVX128_OPTIMAL.
1643 * config/i386/i386.h (processor_costs): Add
1644 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
1645 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
1646 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
1647 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
1648 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
1649 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
1651 2017-10-05 Nathan Sidwell <nathan@acm.org>
1653 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
1655 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1657 * config/arm/arm.c (arm_test_fpu_data): New.
1658 (arm_run_selftests): Call arm_test_fpu_data.
1660 2017-10-04 Nathan Sidwell <nathan@acm.org>
1662 * toplev.c (toplev::main): Remove excess parens on pretty_printer
1664 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
1666 2017-10-04 Sudakshina Das <sudi.das@arm.com>
1668 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
1669 check type for aarch64_simd_valid_immediate.
1670 (aarch64_output_simd_mov_immediate): Update prototype.
1671 (aarch64_simd_valid_immediate): Update prototype.
1672 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
1673 support for ORR-immediate.
1674 (and<mode>3): modified pattern to add support for BIC-immediate.
1675 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
1676 now checks for valid immediate for BIC and ORR based on new enum
1678 (aarch64_output_simd_mov_immediate): Function now used to output
1679 BIC/ORR imm as well based on new enum argument.
1680 * config/aarch64/constraints.md (Do): New vector immediate constraint.
1682 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
1683 (aarch64_reg_or_bic_imm): Likewise.
1685 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1687 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
1688 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
1690 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1694 PR rtl-optimization/82396
1695 * haifa-sched.c (autopref_multipass_init): Simplify
1697 (autopref_rank_data): Simplify sort order.
1698 * sched-int.h (autopref_multipass_data_): Remove
1699 multi_mem_insn_p, min_offset and max_offset.
1701 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1703 * doc/sourcebuild.texi: Document vect_peeling_profitable.
1705 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1707 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
1710 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1712 * doc/sourcebuild.texi: Document vect_long_mult.
1714 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
1716 PR tree-optimization/82413
1717 * fold-const.c (build_range_check): Use widest_int when comparing
1718 the maximum ETYPE value with HIGH.
1720 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1722 PR rtl-optimization/82396
1723 * haifa-sched.c (autopref_multipass_init): Simplify
1725 (autopref_rank_data): Simplify sort order.
1726 * sched-int.h (autopref_multipass_data_): Remove
1727 multi_mem_insn_p, min_offset and max_offset.
1729 2017-10-04 Jakub Jelinek <jakub@redhat.com>
1731 PR tree-optimization/82381
1732 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
1733 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
1736 PR tree-optimization/82374
1737 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
1738 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
1739 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
1740 current_function_decl to the new decl.
1742 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
1744 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
1745 helper macro for IEEE float128 hardware built-in functions.
1746 (SQRTF128_ODD): Add built-in functions with the round-to-odd
1748 (TRUNCF128_ODD): Likewise.
1749 (ADDF128_ODD): Likewise.
1750 (SUBF128_ODD): Likewise.
1751 (MULF128_ODD): Likewise.
1752 (DIVF128_ODD): Likewise.
1753 (FMAF128_ODD): Likewise.
1754 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
1755 UNSPEC_TRUNC_ROUND_TO_ODD.
1756 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
1757 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
1758 floating point round to odd instructions.
1759 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
1760 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
1761 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
1762 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
1763 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
1764 (trunc<mode>sf2_hw): Change the truncate with round to odd
1765 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
1766 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
1767 to odd hardware instructions.
1768 (sub<mode>3_odd): Likewise.
1769 (mul<mode>3_odd): Likewise.
1770 (div<mode>3_odd): Likewise.
1771 (sqrt<mode>2_odd): Likewise.
1772 (fma<mode>4_odd): Likewise.
1773 (fms<mode>4_odd): Likewise.
1774 (nfma<mode>4_odd): Likewise.
1775 (nfms<mode>4_odd): Likewise.
1776 (trunc<mode>df2_odd): Change the truncate with round to odd
1777 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
1779 * doc/extend.texi (PowerPC built-in functions): Update documentation
1780 for existing IEEE float128-bit built-in functions. Add built-in
1781 functions that generate the IEEE 128-bit floating point round to
1784 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
1786 PR rtl-optimization/77729
1787 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
1788 to (X&(C1&~C2))|C2 transformations.
1790 2017-10-03 Martin Jambor <mjambor@suse.cz>
1792 PR tree-optimization/82363
1793 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
1794 mismatch, mark lacc written regardless of racc.
1796 2017-10-03 Jakub Jelinek <jakub@redhat.com>
1798 PR tree-optimization/82381
1799 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
1800 stmt_to_insert nor wheather SSA_NAMEs are default defs.
1801 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
1802 fallthrough into reassoc_stmt_dominates_stmt_p.
1805 * combine.c (combine_instructions): Don't combine in unreachable
1808 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1811 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
1812 function to not use the have_cpu variable. Do not set cpu_index,
1813 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
1815 (rs6000_valid_attribute_p): Remove duplicate initializations of
1816 old_optimize and func_optimize.
1817 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
1818 (rs6000_activate_target_options): Make global.
1819 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
1822 2017-10-02 Jakub Jelinek <jakub@redhat.com>
1824 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
1825 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
1826 Set *poffset to 0 and *psize and *pmax_size to -1 if
1827 *poffset + *psize overflows in HOST_WIDE_INT.
1829 PR tree-optimization/82387
1830 PR tree-optimization/82388
1831 PR tree-optimization/82389
1832 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
1833 instead of live_bytes non-NULL.
1835 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
1838 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
1841 2017-10-02 Richard Biener <rguenther@suse.de>
1843 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
1844 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
1845 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
1847 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1849 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
1850 requested precision matches the type's.
1851 * calls.c (alloc_max_size): Calculate the new candidate size as
1852 a widest_int and use wi::to_widest when comparing it with the
1853 current candidate size.
1854 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
1855 zero rather than integer_zero_node.
1856 * match.pd: Check for a no-op conversion before using wi::add
1857 rather than after. Use tree_to_uhwi when summing small shift
1858 counts into an unsigned int.
1860 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1861 Alan Hayward <alan.hayward@arm.com>
1862 David Sherwood <david.sherwood@arm.com>
1865 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
1866 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
1867 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
1868 POINTER_AND_FP_REGS.
1870 2017-10-02 Richard Biener <rguenther@suse.de>
1872 PR tree-optimization/82355
1873 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1874 a mapping for the enclosing loop but avoid generating one for
1876 (copy_bb_and_scalar_dependences): Remove premature codegen
1877 error on PHIs in blocks duplicated into multiple places.
1878 * graphite-scop-detection.c
1879 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1880 in the region use it as loop and nest to analyze the DR in.
1881 (try_generate_gimple_bb): Likewise.
1882 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1883 (add_loop_constraints): For blocks in a loop not in the region
1884 create a dimension with a single iteration.
1885 * sese.h (gbb_loop_at_index): Remove assert.
1887 2017-10-01 Kevin Buettner <kevinb@redhat.com>
1889 * omp-expand.c (adjust_context_scope): New function.
1890 (expand_parallel_call): Call adjust_context_scope.
1892 2017-10-01 Jeff Law <law@redhat.com>
1894 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
1895 dom_opt_dom_walker class with direct access to private members.
1896 Add comments. Call test_for_singularity.
1897 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
1898 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
1899 m_dummy_cond anymore.
1900 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
1902 (pass_dominator:execute): Build the dummy_cond here and pass it
1903 to the dom_opt_dom_walker ctor.
1904 (test_for_singularity): New function.
1906 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
1907 Maya Rashish <coypu@sdf.org>
1909 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
1910 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
1911 (alpha*-*-netbsd*) Use nbsd_tm_file.
1912 (arm*-*-netbsdelf*) Likewise.
1913 (i[34567]86-*-netbsdelf*) Likewise.
1914 (x86_64-*-netbsd*) Likewise.
1915 (mips*-*-netbsd*) Likewise.
1916 (powerpc-*-netbsd*) Likewise.
1917 (sh*-*-netbsd*) Likewise.
1918 (sparc-*-netbsdelf*) Likewise.
1919 (sparc64-*-netbsd*) Likewise.
1920 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
1922 (vax-*-netbsdelf*) Likewise.
1923 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
1924 (UINT_FAST8_TYPE) Likewise.
1925 (INT_FAST16_TYPE) Check CHAR_FAST16.
1926 (UINT_FAST16_TYPE) Likewise.
1928 2017-09-30 Jakub Jelinek <jakub@redhat.com>
1931 * config/i386/i386.md
1932 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
1933 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
1934 *divmodsi4_zext_2): New define_insn_and_split.
1935 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
1936 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
1937 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
1938 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
1939 New define_insn_and_split.
1940 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
1941 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
1942 operands[1] having DImode when mode is SImode.
1944 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
1945 always SImode for DIV and MOD in REG_EQUAL notes.
1947 2017-09-29 Yury Gribov <tetra2005@gmail.com>
1950 * match.pd: Fix handling of NaNs in pattern.
1952 2017-09-29 Jeff Law <law@redhat.com>
1954 * sbitmap.c (bitmap_bit_in_range_p): New function.
1955 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
1956 * tree-ssa-dse.c (live_bytes_read): New function.
1957 (dse_classify_store): Ignore reads of dead bytes.
1959 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
1960 typos and whitespace errors.
1961 * config/i386/predicates.md (address_no_seg_operand): Likewise.
1962 * config/s390/s390.c (s390_emit_prologue): Likewise.
1964 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1967 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
1968 with a symbol for LRA.
1970 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1972 PR rtl-optimization/82338
1973 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
1975 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
1977 * genmodes.c (calc_wider_mode): Suppress qsort macro.
1978 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
1979 (qsort_chk): Declare.
1980 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
1981 (qsort_chk): New function.
1983 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1985 PR tree-optimization/82337
1986 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
1987 phi definition if the PHI result appears in an abnormal PHI.
1988 (find_basis_for_base_expr): Don't record a basis if the LHS of the
1989 basis appears in an abnormal PHI.
1991 2017-09-29 Richard Biener <rguenther@suse.de>
1993 * graphite-isl-ast-to-gimple.c
1994 (translate_isl_ast_to_gimple::set_codegen_error): New function.
1995 (binary_op_to_tree): Use it.
1996 (get_rename_from_scev): Likewise.
1997 (copy_loop_phi_nodes): Likewise.
1998 (copy_bb_and_scalar_dependences): Likewise.
1999 (translate_pending_phi_nodes): Likewise.
2001 2017-09-29 Jakub Jelinek <jakub@redhat.com>
2004 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
2005 for movabsq $(i32 << shift), r64.
2007 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
2009 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
2010 index when encoding %esp as %rsp to avoid 0x67 prefix.
2012 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
2014 * config/i386/i386.md (*movsf_internal, *movdf_internal):
2015 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
2017 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
2019 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
2020 Extensions with more than 16 double VFP registers.
2021 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
2022 to_clear_mask and all code related to it. Replace the remaining
2023 entry by a sbitmap and adapt code accordingly.
2025 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
2027 * brig-builtins.def: Change pure attributes to const.
2029 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2031 * config.gcc (default_gnu_indirect_function): Default to yes for
2032 sparc*-*-linux* with glibc.
2034 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2036 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
2037 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
2038 when creating .init_array and .fini_array sections with priority
2041 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
2044 * config/aarch64/aarch64.c
2045 (aarch64_builtin_support_vector_misalignment): Always return false
2046 when misalignment is unknown.
2048 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
2050 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
2051 this function to return false if the definition used by the swap
2052 instruction is artificial, or if the memory address from which the
2053 constant value is loaded is not represented by a base address held
2054 in a register or if the base address register is a frame or stack
2055 pointer. Additionally, return false if the base address of the
2056 loaded constant is a SYMBOL_REF but is not considered to be a
2058 (replace_swapped_load_constant): New function.
2059 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
2060 loaded constant vector with a load of a swapped constant vector.
2062 2017-09-27 Carl Love <cel@us.ibm.com>
2064 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
2065 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
2066 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
2069 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
2071 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
2072 first, always call autopref_rank_data otherwise.
2074 2017-09-27 Richard Biener <rguenther@suse.de>
2076 * graphite-scop-detection.c (find_scop_parameters): Move
2077 loop bound handling ...
2078 (gather_bbs::before_dom_children): ... here, avoiding the need
2079 to build scop_info->loop_nest.
2080 (record_loop_in_sese): Remove.
2081 * sese.h (sese_info_t::loop_nest): Remove.
2082 * sese.c (new_sese_info): Do not allocate loop_nest.
2083 (free_sese_info): Do not free loop_nest.
2085 2017-09-27 Jakub Jelinek <jakub@redhat.com>
2088 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
2089 lhs from calls if the lhs has addressable type.
2091 2017-09-27 Richard Biener <rguenther@suse.de>
2093 * graphite.h (scop::max_alias_set): New member.
2094 * graphite-scop-detection.c: Remove references to non-existing
2095 --param in comments.
2096 (build_alias_sets): Record the maximum alias set used for drs.
2097 (build_scops): Support zero as unlimited for
2098 --param graphite-max-arrays-per-scop.
2099 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
2101 (build_poly_sr_1): ... here. Compute alias set based on the
2102 maximum alias set used for drs rather than
2103 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
2105 2017-09-27 Richard Biener <rguenther@suse.de>
2107 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
2108 --param loop-block-tile-size=0 to disable tiling.
2110 2017-09-27 Richard Biener <rguenther@suse.de>
2112 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
2113 (graphite-max-nb-scop-params): Document special value zero.
2114 * domwalk.h (dom_walker::STOP): New symbolical constant.
2115 (dom_walker::dom_walker): Add optional parameter for bb to
2117 (dom_walker::~dom_walker): Declare.
2118 (dom_walker::before_dom_children): Document STOP return value.
2119 (dom_walker::m_user_bb_to_rpo): New member.
2120 (dom_walker::m_bb_to_rpo): Likewise.
2121 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
2122 mapping here if not provided by the user.
2123 (dom_walker::~dom_walker): Free bb to RPO mapping if not
2124 provided by the user.
2125 (dom_walker::STOP): Define.
2126 (dom_walker::walk): Do not compute bb to RPO mapping here.
2127 Support STOP return value from before_dom_children to stop
2129 * graphite-optimize-isl.c (optimize_isl): If the schedule
2130 is the same still generate code if -fgraphite-identity
2131 or -floop-parallelize-all are given.
2132 * graphite-scop-detection.c: Include cfganal.h.
2133 (gather_bbs::gather_bbs): Get and pass through bb to RPO
2135 (gather_bbs::before_dom_children): Return STOP for BBs
2137 (build_scops): Compute bb to RPO mapping and pass it to
2138 the domwalk. Treat --param graphite-max-nb-scop-params=0
2139 as not limiting the number of params.
2140 * graphite.c (graphite_initialize): Remove limit on the
2141 number of basic-blocks in a function.
2142 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
2143 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
2144 default value of 10.
2146 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2148 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
2149 Adjust code to eliminate needing to do the shift right 32-bits
2150 operation after XSCVDPSPN.
2152 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2154 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
2155 ((X / Y) != 0 -> X >= Y): Likewise.
2157 2017-09-26 Carl Love <cel@us.ibm.com>
2159 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
2160 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
2161 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
2162 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
2163 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
2164 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
2165 definitions and overloading.
2166 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
2167 statement for P9V_BUILTIN_XST_LEN_R.
2168 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
2169 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
2170 define_expand and define_insn for the instructions and builtins.
2171 * doc/extend.texi: Update the built-in documentation file for the new
2173 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
2174 define_insn for the instructions
2176 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
2179 * gcc/config/netbsd-protos.h: New file.
2180 * gcc/config/netbsd.c: New file.
2181 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
2182 * gcc/config/t-netbsd: New file.
2183 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
2184 (tmake_file) Add t-netbsd.
2185 (extra_objs) Add netbsd.o.
2187 2017-09-26 Janus Weil <janus@gcc.gnu.org>
2191 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
2193 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2195 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
2196 sign extension from a vector register to a GPR by doing a 32-bit
2197 direct move and then an EXTSW.
2198 (extendsi<mode>2 splitter): Likewise.
2199 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
2200 right or vector extract after doing XSCVDPSPN. Use
2201 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
2203 (movdi_from_sf_zero_ext): Likewise.
2204 (reload_gpr_from_vsxsf): Likewise.
2205 (p8_mfvsrd_4_disf): Delete, no longer used.
2206 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
2207 then needing to move the SFmode to a GPR to use the XSCVDPSP
2208 instruction instead of FRSP and XSCVDPSPN.
2209 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
2210 it is adjacent to the other XSCVSPDP insns.
2211 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
2212 SFmode to be in traditional Altivec registers.
2213 (vsx_xscvdpspn): Eliminate useless alternative constraint.
2214 (vsx_xscvspdpn): Likewise.
2215 (vsx_xscvspdpn_scalar): Likewise.
2217 2017-09-26 Martin Jambor <mjambor@suse.cz>
2219 * tree-sra.c (compare_access_positions): Put integral types first,
2220 stabilize sorting of integral types, remove conditions putting
2221 non-full-precision integers last.
2222 (sort_and_splice_var_accesses): Disable scalarization if a
2223 non-integert would be represented by a non-full-precision integer.
2225 2017-09-26 Joseph Myers <joseph@codesourcery.com>
2227 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
2228 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
2229 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
2230 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
2231 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
2232 conditionals inside the function instead of around it. Call
2233 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
2234 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
2236 2017-09-26 Richard Biener <rguenther@suse.de>
2238 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
2240 (scop_detection::build_scop_breadth): ... this. Removed.
2241 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
2242 (scop_detection::harmful_stmt_in_bb): Likewise.
2243 (scop_detection::graphite_can_represent_stmt): Likewise.
2244 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
2245 (scop_detection::can_represent_loop): Remove recursion, fold in ...
2246 (scop_detection::can_represent_loop_1): ... this. Removed.
2247 (scop_detection::harmful_loop_in_region): Simplify after inlining
2248 the above and remove more quadraticness.
2249 (build_scops): Adjust.
2250 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
2253 2017-09-26 Jakub Jelinek <jakub@redhat.com>
2256 * config/i386/i386.c (ix86_print_operand_address_as): Only test
2257 REGNO (base) == SP_REG if base is a REG.
2260 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
2261 if it is different SSA_NAME.
2262 (optimize_range_tests_cmp_bitwise): New function.
2263 (optimize_range_tests): Call it.
2265 2017-09-26 Richard Biener <rguenther@suse.de>
2267 PR tree-optimization/82321
2268 * graphite.c (canonicalize_loop_closed_ssa): Properly check
2269 for the def being inside the loop.
2271 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2273 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
2275 * config/s390/s390-builtins.def: Fix constraint on op4.
2277 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2279 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
2280 independent expanders.
2281 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
2282 ("vec_ordered", "vec_unordered"): New expanders.
2284 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2286 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
2289 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2291 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
2292 vec_unpacks_lo_v16qi.
2293 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
2295 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2297 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
2298 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
2299 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
2301 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2303 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
2305 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
2307 ("*vec_slb<mode>"): New insn pattern.
2308 ("vec_shr_<mode>"): New expander.
2309 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
2310 and force the shift count operand to V16QImode.
2311 ("vec_srb<mode>"): Set shift count mode to V16QI.
2313 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2315 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
2316 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
2317 ("vec_widen_smult_hi_<mode>"): New expander definitions.
2319 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
2322 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
2324 2017-09-26 Richard Biener <rguenther@suse.de>
2326 PR tree-optimization/82320
2327 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
2330 2017-09-25 Jeff Law <law@redhat.com>
2332 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
2333 prototype for new argument.
2334 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
2335 mostly extracted from rs6000_emit_allocate_stack.
2336 (rs6000_emit_probe_stack_range_stack_clash): New function.
2337 (rs6000_emit_allocate_stack): Call
2338 rs6000_emit_probe_stack_range_stack_clash as needed.
2339 (rs6000_emit_probe_stack_range): Add additional argument
2340 to call to gen_probe_stack_range{si,di}.
2341 (output_probe_stack_range): New.
2342 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
2343 (output_probe_stack_range_stack_clash): New.
2344 (rs6000_emit_prologue): Emit notes into dump file as requested.
2345 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
2346 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
2347 Add additional operand and pass it to output_probe_stack_range.
2349 2017-09-25 Bin Cheng <bin.cheng@arm.com>
2351 PR tree-optimization/82163
2352 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
2353 (checking_verify_loop_closed_ssa): New parameter.
2354 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
2355 (check_loop_closed_ssa_stmt): Delete.
2356 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
2357 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
2358 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
2361 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
2363 * brig-builtins.def: Treat HSAIL barrier builtins as
2364 setjmp/longjump style functions.
2366 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2368 * target.def (constant_alignment): New hook.
2369 * defaults.h (CONSTANT_ALIGNMENT): Delete.
2370 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
2371 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
2372 * doc/tm.texi: Regenerate.
2373 * targhooks.h (default_constant_alignment): Declare.
2374 (constant_alignment_word_strings): Likewise.
2375 * targhooks.c (default_constant_alignment): New function.
2376 (constant_alignment_word_strings): Likewise.
2377 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
2378 instead of CONSTANT_ALIGNMENT.
2379 * varasm.c (align_variable, get_variable_align, build_constant_desc)
2380 (force_const_mem): Likewise.
2381 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
2382 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
2383 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
2384 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2385 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
2387 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
2388 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2389 constant_alignment_word_strings.
2390 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
2391 (CONSTANT_ALIGNMENT): Likewise.
2392 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2393 (arm_constant_alignment): New function.
2394 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
2395 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2396 constant_alignment_word_strings.
2397 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
2398 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2399 constant_alignment_word_strings.
2400 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
2401 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2402 (cris_constant_alignment): New function.
2403 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
2404 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2405 (epiphany_constant_alignment): New function.
2406 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
2407 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2408 constant_alignment_word_strings.
2409 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
2410 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2411 constant_alignment_word_strings.
2412 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
2413 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2414 constant_alignment_word_strings.
2415 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
2416 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
2417 * config/i386/i386.c (ix86_constant_alignment): Make static.
2418 Use the same interface as the target hook.
2419 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2420 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
2421 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2422 constant_alignment_word_strings.
2423 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
2424 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
2425 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2426 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
2427 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2428 constant_alignment_word_strings.
2429 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
2430 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2431 constant_alignment_word_strings.
2432 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
2433 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2434 constant_alignment_word_strings.
2435 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
2436 * config/microblaze/microblaze.c (microblaze_constant_alignment):
2438 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2439 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
2440 * config/mips/mips.c (mips_constant_alignment): New function.
2441 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2442 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
2443 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
2444 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2445 (mmix_constant_alignment): Make static. Use the same interface
2447 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
2448 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2449 constant_alignment_word_strings.
2450 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
2451 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2452 constant_alignment_word_strings.
2453 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
2454 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2455 constant_alignment_word_strings.
2456 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
2457 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2458 (rs6000_constant_alignment): New function.
2459 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
2460 * config/riscv/riscv.c (riscv_constant_alignment): New function.
2461 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2462 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
2463 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2464 (rs6000_constant_alignment): New function.
2465 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
2466 * config/s390/s390.c (s390_constant_alignment): New function.
2467 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2468 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
2469 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2470 constant_alignment_word_strings.
2471 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
2472 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2473 (sparc_constant_alignment): New function.
2474 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
2475 * config/spu/spu.c (spu_constant_alignment): New function.
2476 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2477 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
2478 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2479 constant_alignment_word_strings.
2480 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
2481 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2482 constant_alignment_word_strings.
2483 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
2484 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2485 constant_alignment_word_strings.
2486 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
2487 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2488 (visium_constant_alignment): New function.
2489 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
2490 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2491 (xtensa_constant_alignment): New function.
2492 * system.h (CONSTANT_ALIGNMENT): Poison.
2494 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
2496 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2497 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
2498 (rs6000_builtin_valid_without_lhs): New helper function.
2499 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2500 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
2502 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2504 * target.h (vec_perm_indices): Use unsigned short rather than
2506 (auto_vec_perm_indices): Likewise.
2507 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
2508 Use unsigned int rather than unsigned char.
2509 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
2511 2017-09-25 Richard Biener <rguenther@suse.de>
2513 * cfgloop.h (sort_sibling_loops): Declare.
2514 * cfgloop.c (sort_sibling_loops_cmp): New helper.
2515 (sort_sibling_loops): New function sorting the sibling loop list
2517 * graphite.c (graphite_transform_loops): Sort sibling loops.
2519 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
2521 * target.def (vec_perm_const_ok): Change sel parameter to
2523 * optabs-query.c (can_vec_perm_p): Update accordingly.
2524 * doc/tm.texi: Regenerate.
2525 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
2526 auto_vec_perm_indices and remove separate nelt field.
2527 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
2528 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
2529 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
2530 (aarch64_expand_vec_perm_const): Update accordingly.
2531 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
2532 to vec_perm_indices.
2533 * config/arm/arm.c (expand_vec_perm_d): Change perm to
2534 auto_vec_perm_indices and remove separate nelt field.
2535 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
2536 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
2537 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
2539 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
2540 to vec_perm_indices.
2541 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
2542 sel to vec_perm_indices.
2543 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
2544 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
2545 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
2547 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
2550 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
2553 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
2554 on the FUNCTION_DECL function context if it has a DIE that is a
2557 2017-09-25 Richard Biener <rguenther@suse.de>
2559 PR tree-optimization/82285
2560 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
2563 2017-09-25 Tom de Vries <tom@codesourcery.com>
2567 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
2570 2017-09-25 Richard Biener <rguenther@suse.de>
2572 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
2573 ISL errors other than isl_error_quota happen. Dump if the
2574 schedule is the same.
2575 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
2576 errors instead of aborting inside ISL.
2578 2017-09-25 Iain Sandoe <iain@codesourcery.com>
2581 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
2582 of libgcc_eh for m64.
2583 * config/i386/darwin64.h: Likewise.
2585 2017-09-25 Richard Biener <rguenther@suse.de>
2588 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
2589 attribute for incomplete types nor twice for complete ones.
2591 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
2594 * config/i386/i386.c (ix86_print_operand_address_as): Encode
2595 %esp as %rsp to avoid 0x67 prefix if there is no index or base
2598 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
2601 * config/i386/i386.opt (mprefer-avx256): Use
2602 ix86_target_flags variable.
2603 * config/i386/i386.c (ix86_target_string): Move
2604 -mprefer-avx256 to flag2_opts.
2606 2017-09-22 Jakub Jelinek <jakub@redhat.com>
2609 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
2610 and x != -1 | y != -1 into (x & y) != -1.
2612 2017-09-22 Steve Ellcey <sellcey@cavium.com>
2614 * config.gcc: Add new case statement to set
2615 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
2616 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
2617 s390x-*-linux* case statements. Added aarch64 to the list of
2618 supported architectures.
2620 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2622 PR tree-optimization/82289
2623 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
2624 STMT_VINFO_RELEVANT_P.
2626 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2627 Alan Hayward <alan.hayward@arm.com>
2628 David Sherwood <david.sherwood@arm.com>
2630 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
2631 for VR_RANGE only; don't allow VR_ANTI_RANGE.
2632 (extract_range_from_binary_expr_1): Don't call
2633 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
2635 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2636 Alan Hayward <alan.hayward@arm.com>
2637 David Sherwood <david.sherwood@arm.com>
2639 * target.def (preferred_vector_alignment): New hook.
2640 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
2642 * doc/tm.texi: Regenerate.
2643 * targhooks.h (default_preferred_vector_alignment): Declare.
2644 * targhooks.c (default_preferred_vector_alignment): New function.
2645 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
2647 (DR_TARGET_ALIGNMENT): New macro.
2648 (aligned_access_p): Update commentary.
2649 (vect_known_alignment_in_bytes): New function.
2650 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
2652 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
2653 Calculate the misalignment based on the target alignment rather than
2655 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
2656 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
2657 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
2658 the target alignment, rather than masking the element misalignment
2659 with the number of elements in a vector. Also use the target
2660 alignment when calculating the maximum number of peels.
2661 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
2662 instead of TYPE_ALIGN_UNIT.
2663 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
2664 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
2665 (vect_create_addr_base_for_vector_ref): Update call accordingly.
2666 (vect_create_data_ref_ptr): Likewise.
2667 (vect_setup_realignment): Realign by ANDing with
2668 -DR_TARGET_MISALIGNMENT.
2669 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
2670 the number of peels based on DR_TARGET_ALIGNMENT.
2671 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
2672 with the guaranteed alignment boundary when deciding whether
2674 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
2675 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
2676 (ensure_base_align): Remove stmt_info parameter. Get the
2677 target base alignment from DR_TARGET_ALIGNMENT.
2678 (vectorizable_store): Update call accordingly. Interpret
2679 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
2681 (vectorizable_load): Likewise.
2683 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2684 Alan Hayward <alan.hayward@arm.com>
2685 David Sherwood <david.sherwood@arm.com>
2687 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
2688 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
2689 (vect_enhance_data_refs_alignment): Likewise.
2691 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2693 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2694 error. Only quit immediately if parsing is complete.
2695 (BEGIN): Initialize fatal_err and parse_done.
2696 (begin fpu, end fpu): Check number of arguments.
2697 (begin arch, end arch): Likewise.
2698 (begin cpu, end cpu): Likewise.
2699 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2700 (optalias): Likewise.
2702 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2704 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2705 * config/arm/arm-isa.h: Delete. Move definitions to ...
2706 * arm-cpus.in: ... here. Use new feature and fgroup values.
2707 * config/arm/arm.c (arm_option_override): Use lower case for feature
2709 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2710 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2711 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2713 (print_isa_bits_for): New function.
2714 (gen_isa): New function.
2715 (gen_comm_data): Use print_isa_bits_for.
2716 (define feature): New keyword.
2717 (define fgroup): New keyword.
2718 * config/arm/t-arm (TM_H): Remove.
2719 (GTM_H): Add arm-isa.h.
2720 (arm-isa.h): Add rule to generate file.
2721 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2722 case for feature bit names.
2724 2017-09-22 Richard Biener <rguenther@suse.de>
2726 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
2728 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
2729 print of no dependency loops ...
2730 * graphite.c (graphite_transform_loops): ... here.
2731 (canonicalize_loop_closed_ssa_form): Work from inner to outer
2733 (same_close_phi_node, remove_duplicate_close_phi,
2734 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
2735 (canonicalize_loop_closed_ssa): ... here and simplify.
2736 * graphite-optimize-isl.c: Include tree-vectorizer.h.
2737 (optimize_isl): Use dump_printf_loc to tell when we stopped
2738 optimizing because of an ISL timeout.
2740 2017-09-22 Richard Biener <rguenther@suse.de>
2742 PR tree-optimization/82291
2743 * tree-if-conv.c (predicate_mem_writes): Make sure to
2744 remove writes in blocks predicated with false.
2746 2017-09-22 Richard Biener <rguenther@suse.de>
2748 * sese.c: Include cfganal.h.
2749 (if_region_set_false_region): Remove.
2750 (create_if_region_on_edge): Likewise.
2751 (move_sese_in_condition): Re-implement without destroying
2754 2017-09-22 Richard Biener <rguenther@suse.de>
2756 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
2757 Verify both BBs contain loop PHI nodes before dispatching to
2759 (graphite_regenerate_ast_isl): Do not recompute dominators,
2760 do not verify three times. Restructure for clarity.
2761 * graphite-scop-detection.c (same_close_phi_node,
2762 remove_duplicate_close_phi, make_close_phi_nodes_unique,
2763 defined_in_loop_p, canonicalize_loop_closed_ssa,
2764 canonicalize_loop_closed_ssa_form): Simplify, remove excess
2765 checking and SSA rewrite, move to ...
2766 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
2767 (graphite_initialize): Do not pass in ctx, do not reset the
2768 SCEV cache, compute only dominators.
2769 (graphite_transform_loops): Allocate ISL ctx after
2770 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
2771 Maintain post-dominators only around build_scops.
2772 * sese.c (if_region_set_false_region): Make static. Free
2773 and recompute dominators.
2774 (move_sese_in_condition): Assert we don't get called with
2775 post-dominators computed.
2776 * sese.h (if_region_set_false_region): Remove.
2778 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
2780 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
2781 mode attribute for TARGET_AVX512VL.
2783 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
2785 * config/i386/i386.opt (mprefer-avx256): New option.
2786 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
2788 (ix86_preferred_simd_mode): Return 256-bit AVX modes
2789 for TARGET_PREFER_AVX256.
2790 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
2792 2017-09-21 Jeff Law <law@redhat.com>
2794 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
2795 Fix dump output if the only stack space is for pushed registers.
2797 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2799 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
2802 2017-09-21 Martin Sebor <msebor@redhat.com>
2805 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
2806 code (in C++) or code that triggers warnings.
2808 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
2810 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
2812 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
2814 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
2815 * sched-rgn.c: Ditto.
2816 * sel-sched-ir.c: Ditto.
2818 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
2820 * toplev.h (set_random_seed): Adjust return type.
2821 * toplev.c (init_local_tick): Move eager initialization of random_seed
2822 to get_random_seed. Adjust comment.
2823 (init_random_seed): Inline to get_random_seed, delete.
2824 (get_random_seed): Initialize random_seed lazily.
2825 (set_random_seed): Do not return previous value.
2826 (print_switch_value): Do not call get_random_seed.
2828 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
2830 * cgraph.c (delete_function_version): New, broken out from...
2831 (cgraph_node::delete_function_version): ...here. Rename to
2832 cgraph_node::delete_function_version_by_decl. Update all uses.
2833 (cgraph_node::remove): Call delete_function_version.
2835 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2838 * tree-inline.c (expand_call_inline): Emit clobber stmts for
2839 VAR_DECLs to which addressable non-volatile parameters are mapped
2840 and for id->retvar after the return value assignment. Clear
2841 id->retval and id->retbnd after inlining.
2843 2017-09-21 Richard Biener <rguenther@suse.de>
2845 PR tree-optimization/82276
2846 PR tree-optimization/82244
2847 * tree-vrp.c (build_assert_expr_for): Set
2848 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
2850 (remove_range_assertions): Revert earlier change.
2852 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
2855 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
2857 2017-09-21 Richard Biener <rguenther@suse.de>
2859 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2860 Restore valid IL after code generation errors.
2861 * graphite.c (graphite_transform_loops): Diagnose code
2862 generation issues as MSG_MISSED_OPTIMIZATION and continue
2863 with processing SCOPs.
2865 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2866 Alan Hayward <alan.hayward@arm.com>
2867 David Sherwood <david.sherwood@arm.com>
2869 * calls.c (compute_argument_addresses): Use simplify_gen_binary
2870 rather than choosing between plus_constant and gen_rtx_<CODE>.
2871 * expr.c (emit_push_insn): Likewise.
2872 (expand_expr_real_2): Likewise.
2874 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2875 Alan Hayward <alan.hayward@arm.com>
2876 David Sherwood <david.sherwood@arm.com>
2878 * loop-unroll.c (split_iv): Call copy_rtx on the step.
2880 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2881 Alan Hayward <alan.hayward@arm.com>
2882 David Sherwood <david.sherwood@arm.com>
2884 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
2885 calling tree_to_uhwi.
2887 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2888 Alan Hayward <alan.hayward@arm.com>
2889 David Sherwood <david.sherwood@arm.com>
2891 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
2892 INTEGER_CST rather than a negative test for ADDR_EXPR.
2894 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2895 Alan Hayward <alan.hayward@arm.com>
2896 David Sherwood <david.sherwood@arm.com>
2898 * tree-vrp.c (extract_range_from_binary_expr_1): Check
2899 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
2901 2017-09-21 Richard Biener <rguenther@suse.de>
2903 PR tree-optimization/71351
2904 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
2905 graphite_create_new_loop_guard): Remove, fold remaining parts
2907 (translate_isl_ast_node_for): ... here and simplify.
2909 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2912 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
2913 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
2914 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
2915 alternative always use QI mode, for -Os imov (=R,R) alternative
2916 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
2919 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2920 Jeff Law <law@redhat.com>
2922 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
2923 (allocate_stack_space): New function, partially extracted from
2925 (s390_emit_prologue): Track offset to most recent stack probe.
2926 Code to allocate space moved into allocate_stack_space.
2927 Dump actions when no stack is allocated.
2928 (s390_prologue_plus_offset): New function.
2929 (s390_emit_stack_probe): Likewise.
2931 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
2933 * common.opt (Wa, Wl, Wp, g, gz=): Add
2935 (gno-column-info): Remove.
2936 (gcolumn-info): Drop RejectNegative.
2938 (gno-record-gcc-switches): Remove.
2939 (grecord-gcc-switches): Drop RejectNegative.
2940 (gno-split-dwarf): Remove.
2941 (gsplit-dwarf): Drop RejectNegative.
2942 (gno-strict-dwarf): Remove.
2943 (gstrict-dwarf): Drop RejectNegative.
2944 * config/darwin.opt (gfull, gused): Add RejectNegative.
2945 * dwarf2out.c (gen_producer_string): Drop
2946 gno-record-gcc-switches handler.
2947 * optc-gen.awk: Add g to prefixes with negative forms.
2948 * opts-common.c (remapping_prefix_p): New.
2949 (find_opt): Check it.
2950 (generate_canonical_option): Test g prefix.
2951 (option_map): Add -gno- mapping.
2952 (add_misspelling_candidates): Check remapping_prefix_p.
2954 2017-09-20 Jeff Law <law@redhat.com>
2956 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
2957 thinko in stack clash protection support.
2959 * explow.c (compute_stack_clash_protection_loop_data): Use
2960 CONST_INT_P instead of explicit test. Verify object is a
2961 CONST_INT_P before looking at INTVAL.
2962 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
2963 instead of explicit test.
2965 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2968 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
2969 address instead of to r1 and r11.
2971 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
2973 * config.gcc: Support "knm".
2974 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
2975 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2977 * config/i386/i386.c (m_KNM): Define.
2978 (processor_target_table): Add "knm".
2980 (ix86_option_override_internal): Add "knm".
2981 (ix86_issue_rate): Add PROCESSOR_KNM.
2982 (ix86_adjust_cost): Ditto.
2983 (ia32_multipass_dfa_lookahead): Ditto.
2984 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
2985 (fold_builtin_cpu): Add M_INTEL_KNM.
2986 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
2987 (processor_type): Add PROCESSOR_KNM.
2988 * config/i386/x86-tune.def: Add m_KNM.
2989 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
2991 2017-09-20 Richard Biener <rguenther@suse.de>
2993 PR tree-optimization/80213
2994 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
2995 are allowed in empty BBs as well.
2996 (canonicalize_loop_closed_ssa): Also look for other complex
2998 (scop_detection::get_sese): Include the loop-closed PHI block
3000 (scop_detection::merge_sese): Remove code adding extra blocks.
3001 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
3002 (build_scops): Assert the final returned scop is invalid.
3004 2017-09-20 Richard Biener <rguenther@suse.de>
3006 PR tree-optimization/82264
3007 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
3009 (vn_phi_lookup): Likewise.
3010 (vn_phi_insert): Likewise.
3012 2017-09-20 Jakub Jelinek <jakub@redhat.com>
3014 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
3015 that fits into uhwi or shwi, add DW_AT_const_value regardless
3016 of early_dwarf without going through RTL, using add_AT_unsigned
3019 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
3020 (DEBUG_LTO_ABBREV_SECTION): Likewise.
3021 (DEBUG_LTO_MACINFO_SECTION): Likewise.
3022 (DEBUG_MACRO_SECTION): Likewise.
3023 (DEBUG_LTO_MACRO_SECTION): Likewise.
3024 (DEBUG_STR_DWO_SECTION): Likewise.
3025 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
3026 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
3027 (DEBUG_LTO_DWO_LINE_SECTION): Define.
3028 (DEBUG_LTO_LINE_STR_SECTION): Define.
3029 (init_sections_and_labels): Initialize debug_line_str_section
3030 variable. Initialize debug_loc_section for -gdwarf-5 to
3031 DEBUG_LOCLISTS_SECTION. Formatting fixes.
3033 2017-09-20 Richard Biener <rguenther@suse.de>
3035 * graphite-sese-to-poly.c (extract_affine): Properly handle
3036 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
3038 2017-09-20 Richard Biener <rguenther@suse.de>
3040 PR tree-optimization/81373
3041 * graphite-scop-detection.c (build_cross_bb_scalars_def):
3042 Force SESE live-out defs to be handled even if they are
3045 2017-09-19 Jeff Law <law@redhat.com>
3047 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
3048 nothing for stack adjustments with REG_STACK_CHECK.
3049 * sched-deps.c (parse_add_or_inc): Reject insns with
3050 REG_STACK_CHECK from dependency breaking.
3051 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
3052 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
3053 * reg-notes.def (STACK_CHECK): New note.
3055 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
3056 (ix86_expand_prologue): Dump stack clash info as needed.
3057 Call ix86_adjust_stack_and_probe_stack_clash as needed.
3059 * function.c (dump_stack_clash_frame_info): New function.
3060 * function.h (dump_stack_clash_frame_info): Prototype.
3061 (enum stack_clash_probes): New enum.
3063 * config/alpha/alpha.c (alpha_expand_prologue): Also check
3064 flag_stack_clash_protection.
3065 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
3066 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
3067 (arm_frame_pointer_required): Likewise.
3068 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
3069 (ia64_expand_prologue): Likewise.
3070 * config/mips/mips.c (mips_expand_prologue): Likewise.
3071 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
3072 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3073 (sparc_flat_expand_prologue): Likewise.
3074 * config/spu/spu.c (spu_expand_prologue): Likewise.
3076 * explow.c: Include "params.h".
3077 (anti_adjust_stack_and_probe_stack_clash): New function.
3078 (get_stack_check_protect): Likewise.
3079 (compute_stack_clash_protection_loop_data): Likewise.
3080 (emit_stack_clash_protection_loop_start): Likewise.
3081 (emit_stack_clash_protection_loop_end): Likewise.
3082 (allocate_dynamic_stack_space): Use get_stack_check_protect.
3083 Use anti_adjust_stack_and_probe_stack_clash.
3084 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
3085 (emit_stack_clash_protection_loop_start): Likewise.
3086 (emit_stack_clash_protection_loop_end): Likewise.
3087 * rtl.h (get_stack_check_protect): Prototype.
3088 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
3089 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
3090 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
3092 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
3094 * doc/tm.texi: Rebuilt.
3095 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3096 get_stack_check_protect.
3097 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
3098 * config/arm/arm.c (arm_expand_prologue): Likewise.
3099 (arm_frame_pointer_required): Likewise.
3100 * config/i386/i386.c (ix86_expand_prologue): Likewise.
3101 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
3102 * config/mips/mips.c (mips_expand_prologue): Likewise.
3103 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
3104 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
3105 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3106 (sparc_flat_expand_prologue): Likewise.
3108 * common.opt (-fstack-clash-protection): New option.
3109 * flag-types.h (enum stack_check_type): Note difference between
3110 -fstack-check= and -fstack-clash-protection.
3111 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
3112 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
3113 * toplev.c (process_options): Issue warnings/errors for cases
3114 not handled with -fstack-clash-protection.
3115 * doc/invoke.texi (-fstack-clash-protection): Document new option.
3116 (-fstack-check): Note additional problem with -fstack-check=generic.
3117 Note that -fstack-check is primarily for Ada and refer users
3118 to -fstack-clash-protection for stack-clash-protection.
3119 Document new params for stack clash protection.
3121 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
3123 * config/i386/i386.c (ix86_split_long_move): Do not handle
3124 address used for LEA in a special way.
3126 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3128 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
3130 2017-09-19 Martin Sebor <msebor@redhat.com>
3133 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
3134 of incompatible types.
3136 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
3138 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3139 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
3140 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3141 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
3143 2017-09-19 Richard Biener <rguenther@suse.de>
3145 PR tree-optimization/82244
3146 * tree-vrp.c (remove_range_assertions): Do not propagate
3147 a constant to abnormals but replace the assert with a copy.
3149 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
3151 PR rtl-optimization/57878
3152 PR rtl-optimization/68988
3153 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
3154 avoidance test involving non_reload_pseudos. Move frequency test
3155 below the general fragmentation avoidance test.
3157 2017-09-19 Richard Biener <rguenther@suse.de>
3159 PR tree-optimization/69728
3160 * graphite-sese-to-poly.c (schedule_error): New global.
3161 (add_loop_schedule): Handle empty domain by failing the
3163 (build_original_schedule): Handle schedule_error.
3165 2017-09-19 Richard Biener <rguenther@suse.de>
3167 * graphite-scop-detection.c (scop_detection::can_represent_loop):
3168 Do not iterate to sibling loops but only to siblings of inner
3171 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
3174 * config/m68k/m68k.md (moveq feeding equality comparison): Check
3175 that the registers are different.
3177 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
3179 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
3180 to processor_model and "amdfam17h" to arch_names_table.
3181 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
3183 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3186 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
3188 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3189 Alan Hayward <alan.hayward@arm.com>
3190 David Sherwood <david.sherwood@arm.com>
3192 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
3194 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
3196 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
3197 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
3198 vect_schedule_slp_instance.
3199 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
3200 Update call to vect_slp_analyze_node_operations. Simplify return
3202 (vect_slp_analyze_bb_1): Update call accordingly.
3203 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
3204 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
3205 (vect_schedule_slp): Update call accordingly.
3207 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3208 Alan Hayward <alan.hayward@arm.com>
3209 David Sherwood <david.sherwood@arm.com>
3211 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
3212 with types that aren't in fact scalar.
3214 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3216 * tree-vect-slp.c (vect_record_max_nunits): New function,
3218 (vect_build_slp_tree_1): ...here.
3219 (vect_build_slp_tree_2): Call it for phis too.
3221 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3223 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
3224 to vect_get_vec_def_for_operand when getting the mask operand.
3226 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3227 Alan Hayward <alan.hayward@arm.com>
3228 David Sherwood <david.sherwood@arm.com>
3230 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
3233 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3234 Alan Hayward <alan.hayward@arm.com>
3235 David Sherwood <david.sherwood@arm.com>
3237 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
3238 calculation for vector booleans.
3240 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3241 Alan Hayward <alan.hayward@arm.com>
3242 David Sherwood <david.sherwood@arm.com>
3244 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
3246 (vect_transform_stmt): ...here.
3247 (vect_analyze_stmt): Use it instead of calling
3248 vectorizable_live_operation directly.
3250 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
3252 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
3253 non-SIMT targets in acc vector loops.
3255 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
3257 * configure.ac: Add arc and check if assembler supports gdwarf2.
3258 * configure: Regenerate.
3260 2017-09-18 Richard Biener <rguenther@suse.de>
3262 PR tree-optimization/82220
3263 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
3264 epilogue niters from the min_profitable_iters compute.
3266 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3269 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
3270 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
3271 (ix86_init_pic_reg): Revert 2017-09-01 changes.
3273 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
3276 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
3277 switching to a new text section.
3279 2017-09-18 Richard Biener <rguenther@suse.de>
3281 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
3283 (build_alias_set): Reject aliases with no access function.
3285 2017-09-18 Richard Biener <rguenther@suse.de>
3287 PR tree-optimization/79622
3288 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
3290 (build_cross_bb_scalars_use): Likewise.
3292 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
3294 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
3296 2017-09-18 Alan Modra <amodra@gmail.com>
3299 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
3300 stack_pointer_rtx for count 0. Update comments. Break up
3301 large rtl expression.
3303 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
3305 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
3306 Increase to 20 bytes.
3307 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
3308 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
3309 or avx version of the stub.
3311 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
3314 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
3315 compute the minimum stack alignment. Also update preferred stack
3316 boundary for leaf functions.
3318 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
3320 PR tree-optimization/82228
3321 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
3324 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3326 * common/config/nds32/nds32-common.c
3327 (nds32_option_optimization_table): Refine formatting.
3328 (nds32_option_optimization_table): Use -fsched-pressure and
3329 -fomit-frame-pointer for specific optimization level.
3331 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3333 * config/nds32/nds32.c: Refine formatting and comments.
3334 * config/nds32/nds32.h: Likewise.
3335 * config/nds32/nds32.md: Likewise.
3336 * config/nds32/nds32-cost.c: Likewise.
3337 * config/nds32/nds32-isr.c: Likewise.
3338 * config/nds32/nds32-md-auxiliary.c: Likewise.
3339 * config/nds32/nds32-multiple.md: Likewise.
3340 * config/nds32/nds32-predicates.c: Likewise.
3342 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
3343 Jakub Jelinek <jakub@redhat.com>
3345 Add support for -std=c++2a.
3346 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
3348 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
3350 2017-09-15 Steve Ellcey <sellcey@cavium.com>
3353 * doc/extend.texi (Common Function Attributes): Add
3354 references to ARM, AArch64, and S/390 specific attributes.
3355 (Function Specific Option Pragmas): Add AArch64 and S/390
3356 to list of back ends that support the target pragma.
3358 2017-09-15 Nathan Sidwell <nathan@acm.org>
3360 * doc/standards.texi: Fix C++17 description. Update URLs for
3363 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
3365 * common.opt (Wcast-align=strict): New warning option.
3366 * doc/invoke.texi: Document -Wcast-align=strict.
3368 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
3370 * cgraph.h (cgraph_thunk_info): Add comments.
3371 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
3372 assert for VIRTUAL_* arguments stricter.
3374 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
3376 PR tree-optimization/71026
3377 * match.pd: Move RDIV patterns from fold-const.c
3378 * fold-const.c (distribute_real_division): Removed.
3379 (fold_binary_loc): Remove calls to distribute_real_divison.
3381 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3383 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
3384 c++1z and gnu++1z as deprecated. Change other references to
3385 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
3386 Change -Wc++1z-compat to -Wc++17-compat.
3387 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
3388 * dwarf2out.c (highest_c_language): Handle C++17.
3389 (gen_compile_unit_die): Likewise.
3391 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3393 PR rtl-optimization/82192
3394 * combine.c (make_extraction): Don't look through non-paradoxical
3395 SUBREGs or TRUNCATE if pos + len is or might be bigger than
3398 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3399 Alan Hayward <alan.hayward@arm.com>
3400 David Sherwood <david.sherwood@arm.com>
3402 * target.def (function_arg_offset): New hook.
3403 * targhooks.h (default_function_arg_offset): Declare.
3404 * targhooks.c (default_function_arg_offset): New function.
3405 * function.c (locate_and_pad_parm): Use
3406 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
3407 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
3408 (TARGET_FUNCTION_ARG_OFFSET): ...this.
3409 * doc/tm.texi: Regenerate.
3410 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
3411 * config/spu/spu.c (spu_function_arg_offset): New function.
3412 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
3413 * system.h (FUNCTION_ARG_OFFSET): Poison.
3415 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3416 Alan Hayard <alan.hayward@arm.com>
3417 David Sherwood <david.sherwood@arm.com>
3419 * target.def (truly_noop_truncation): New hook.
3420 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
3421 than TRULY_NOOP_TRUNCATION.
3422 * hooks.h (hook_bool_uint_uint_true): Declare.
3423 * hooks.c (hook_bool_uint_uint_true): New function.
3424 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
3425 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
3426 * doc/tm.texi: Regenerate.
3427 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
3428 rather than TRULY_NOOP_TRUNCATION in comments.
3429 (simplify_comparison): Likewise.
3430 (record_truncated_value): Likewise.
3431 * expmed.c (extract_bit_field_1): Likewise.
3432 (extract_split_bit_field): Likewise.
3433 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
3434 instead of TRULY_NOOP_TRUNCATION.
3435 * function.c (assign_parm_setup_block): Likewise.
3436 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
3437 * rtlhooks.c: Include target.h.
3438 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
3439 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
3440 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
3441 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
3442 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
3443 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
3444 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
3445 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
3446 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
3447 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
3448 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
3449 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
3450 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
3451 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
3452 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
3453 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
3454 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
3455 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
3456 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
3457 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
3458 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
3459 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
3460 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
3461 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
3462 * config/mips/mips.c (mips_truly_noop_truncation): New function.
3463 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3464 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3465 rather than TRULY_NOOP_TRUNCATION in comments.
3466 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
3467 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
3468 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
3469 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
3470 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
3471 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
3472 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
3473 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
3474 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
3475 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
3476 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
3477 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3478 rather than TRULY_NOOP_TRUNCATION in comments.
3479 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
3480 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
3481 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
3482 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
3483 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
3484 TRULY_NOOP_TRUNCATION condition.
3485 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
3486 (TRULY_NOOP_TRUNCATION): Delete.
3487 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
3488 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
3489 * config/spu/spu.c (spu_truly_noop_truncation): New function.
3490 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3491 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
3492 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
3493 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
3494 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3495 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3496 rather than TRULY_NOOP_TRUNCATION in comments.
3497 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
3498 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
3499 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
3500 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
3501 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
3502 * system.h (TRULY_NOOP_TRUNCATION): Poison.
3504 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
3507 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
3508 (*cmp_ior): Likewise.
3509 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
3510 (*ior_scc_scc_cmp): Likewise.
3511 (*and_scc_scc): Likewise.
3512 (*and_scc_scc_cmp): Likewise.
3514 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3515 Alan Hayard <alan.hayward@arm.com>
3516 David Sherwood <david.sherwood@arm.com>
3518 * target.def (can_change_mode_class): New hook.
3519 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
3520 (hard_regno_nregs): Likewise.
3521 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
3522 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
3523 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
3524 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
3525 (LOAD_EXTEND_OP): Update accordingly.
3526 * doc/tm.texi: Regenerate.
3527 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
3528 CANNOT_CHANGE_MODE_CLASS.
3529 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
3530 (REG_CAN_CHANGE_MODE_P): ...this new macro.
3531 * combine.c (simplify_set): Update accordingly.
3532 * emit-rtl.c (validate_subreg): Likewise.
3533 * recog.c (general_operand): Likewise.
3534 * regcprop.c (mode_change_ok): Likewise.
3535 * reload1.c (choose_reload_regs): Likewise.
3536 (inherit_piecemeal_p): Likewise.
3537 * rtlanal.c (simplify_subreg_regno): Likewise.
3538 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
3539 instead of CANNOT_CHANGE_MODE_CLASS.
3540 (reload_cse_simplify_operands): Likewise.
3541 * reload.c (push_reload): Use targetm.can_change_mode_class
3542 instead of CANNOT_CHANGE_MODE_CLASS.
3543 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
3544 REG_CANNOT_CHANGE_MODE_P.
3545 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3546 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
3547 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3548 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3549 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3550 (arm_can_change_mode_class): New function.
3551 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3552 than CANNOT_CHANGE_MODE_CLASS in comments.
3553 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3554 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
3555 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
3556 (ix86_can_change_mode_class): ...this new function, inverting the
3557 sense of the return value.
3558 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3559 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3560 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3561 (ia64_can_change_mode_class): New function.
3562 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3563 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
3564 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
3565 (m32c_can_change_mode_class): ...this new function, inverting the
3566 sense of the return value.
3567 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3568 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3569 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
3570 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
3571 (mips_can_change_mode_class): ...this new function, inverting the
3572 sense of the return value.
3573 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3574 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3575 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3576 (msp430_can_change_mode_class): New function.
3577 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3578 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
3579 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3580 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3581 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3582 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
3583 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3584 (pa_cannot_change_mode_class): Replace with...
3585 (pa_can_change_mode_class): ...this new function, inverting the
3586 sense of the return value.
3587 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3588 than CANNOT_CHANGE_MODE_CLASS in comments.
3589 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3590 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
3591 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3592 (pdp11_cannot_change_mode_class): Replace with...
3593 (pdp11_can_change_mode_class): ...this new function, inverting the
3594 sense of the return value.
3595 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3596 * config/powerpcspe/powerpcspe-protos.h
3597 (rs6000_cannot_change_mode_class_ptr): Delete.
3598 * config/powerpcspe/powerpcspe.c
3599 (rs6000_cannot_change_mode_class_ptr): Delete.
3600 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3601 (rs6000_option_override_internal): Assign to
3602 targetm.can_change_mode_class instead of
3603 rs6000_cannot_change_mode_class_ptr.
3604 (rs6000_cannot_change_mode_class): Replace with...
3605 (rs6000_can_change_mode_class): ...this new function, inverting the
3606 sense of the return value.
3607 (rs6000_debug_cannot_change_mode_class): Replace with...
3608 (rs6000_debug_can_change_mode_class): ...this new function.
3609 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3610 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
3611 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3612 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3613 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
3615 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
3616 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3617 (rs6000_option_override_internal): Assign to
3618 targetm.can_change_mode_class instead of
3619 rs6000_cannot_change_mode_class_ptr.
3620 (rs6000_cannot_change_mode_class): Replace with...
3621 (rs6000_can_change_mode_class): ...this new function, inverting the
3622 sense of the return value.
3623 (rs6000_debug_cannot_change_mode_class): Replace with...
3624 (rs6000_debug_can_change_mode_class): ...this new function.
3625 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3626 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
3627 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
3628 (s390_can_change_mode_class): ...this new function, inverting the
3629 sense of the return value.
3630 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3631 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3632 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
3633 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3634 (sh_cannot_change_mode_class): Replace with...
3635 (sh_can_change_mode_class): ...this new function, inverting the
3636 sense of the return value.
3637 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3638 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3639 (sparc_can_change_mode_class): New function.
3640 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3641 * config/spu/spu.c (spu_can_change_mode_class): New function.
3642 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3643 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3644 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3645 (visium_can_change_mode_class): New function.
3646 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
3648 2017-09-15 Richard Biener <rguenther@suse.de>
3650 PR tree-optimization/82217
3651 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
3652 but not undefined case.
3654 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3657 * postreload.c (reload_cse_simplify_operands): Skip
3658 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
3660 2017-09-15 Richard Biener <rguenther@suse.de>
3662 PR tree-optimization/68823
3663 * graphite-scop-detection.c (build_alias_set): If we have a
3664 possible dependence check whether we can handle them by just
3665 looking at the DRs DR_ACCESS_FNs.
3666 (build_scops): If build_alias_set fails, fail the SCOP.
3668 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3670 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
3671 to support float128 built-in functions that require the ISA 3.0
3673 (BU_FLOAT128_3_HW): Likewise.
3674 (SQRTF128): Add support for the IEEE 128-bit square root and fma
3676 (FMAF128): Likewise.
3678 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
3679 support for built-in functions that need the ISA 3.0 IEEE 128-bit
3680 floating point instructions.
3681 (rs6000_invalid_builtin): Likewise.
3682 (rs6000_builtin_mask_names): Likewise.
3683 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
3684 (RS6000_BTM_FLOAT128_HW): Likewise.
3685 (RS6000_BTM_COMMON): Likewise.
3686 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
3688 * doc/extend.texi (RS/6000 built-in functions): Document the
3689 IEEE 128-bit floating point square root and fused multiply-add
3692 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
3694 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
3695 reg (r2) isn't in the set of registers defined in the prologue.
3697 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3698 Alan Hayward <alan.hayward@arm.com>
3699 David Sherwood <david.sherwood@arm.com>
3701 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
3702 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
3703 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
3704 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
3705 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
3707 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
3708 max_vectorization_factor.
3709 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
3711 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3712 Alan Hayward <alan.hayward@arm.com>
3713 David Sherwood <david.sherwood@arm.com>
3715 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
3716 (vect_worthwhile_without_simd_p): Declare.
3717 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
3718 (vectorizable_reduction): Use it.
3719 * tree-vect-stmts.c (vectorizable_shift): Likewise.
3720 (vectorizable_operation): Likewise.
3722 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3723 Alan Hayward <alan.hayward@arm.com>
3724 David Sherwood <david.sherwood@arm.com>
3726 * tree-vectorizer.h (vect_get_num_copies): New function.
3727 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
3728 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3729 (vectorizable_induction): Likewise.
3730 (vectorizable_live_operation): Likewise.
3731 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3732 (vectorizable_bswap): Likewise.
3733 (vectorizable_call): Likewise.
3734 (vectorizable_conversion): Likewise.
3735 (vectorizable_assignment): Likewise.
3736 (vectorizable_shift): Likewise.
3737 (vectorizable_operation): Likewise.
3738 (vectorizable_store): Likewise.
3739 (vectorizable_load): Likewise.
3740 (vectorizable_condition): Likewise.
3741 (vectorizable_comparison): Likewise.
3742 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
3744 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3745 Alan Hayward <alan.hayward@arm.com>
3746 David Sherwood <david.sherwood@arm.com>
3748 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
3749 of vect_init_vector.
3751 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3752 Alan Hayward <alan.hayward@arm.com>
3753 David Sherwood <david.sherwood@arm.com>
3755 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
3756 an inline wrapper that provides a location.
3757 (gimple_build_vector): Likewise.
3758 * gimple-fold.c (gimple_build_vector_from_val): New function.
3759 (gimple_build_vector): Likewise.
3760 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
3761 functions to build the initial value. Always return a gimple value.
3762 (get_initial_defs_for_reduction): Likewise. Only compute
3764 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
3765 vect_init_vector on the results from get_initial_def(s)_for_reduction.
3766 (vectorizable_induction): Use gimple_build_vector rather than
3769 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3770 Alan Hayward <alan.hayward@arm.com>
3771 David Sherwood <david.sherwood@arm.com>
3773 * target.h (vec_perm_indices): New typedef.
3774 (auto_vec_perm_indices): Likewise.
3775 * optabs-query.h: Include target.h
3776 (can_vec_perm_p): Take a vec_perm_indices *.
3777 * optabs-query.c (can_vec_perm_p): Likewise.
3778 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
3779 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3780 * tree-vect-generic.c (lower_vec_perm): Likewise.
3781 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
3782 (vect_grouped_load_supported): Likewise.
3783 (vect_shift_permute_load_chain): Likewise.
3784 (vect_permute_store_chain): Use auto_vec_perm_indices.
3785 (vect_permute_load_chain): Likewise.
3786 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
3787 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
3788 Update uses of can_vec_perm_p.
3789 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
3790 mode with a number of elements. Take a vec_perm_indices *.
3791 (vect_create_epilog_for_reduction): Update accordingly.
3792 Use auto_vec_perm_indices.
3793 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
3794 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
3795 (vect_transform_slp_perm_load): Likewise.
3796 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
3797 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
3798 (vect_gen_perm_mask_checked): Likewise.
3799 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
3800 (vect_gen_perm_mask_checked): Likewise.
3801 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
3802 (vectorizable_store): Likewise.
3803 (vectorizable_load): Likewise.
3804 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
3805 (vectorizable_bswap): Likewise.
3807 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3808 Alan Hayward <alan.hayward@arm.com>
3809 David Sherwood <david.sherwood@arm.com>
3811 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
3812 * tree.c (build_vector): Likewise.
3813 (build_vector_from_ctor): Update accordingly.
3814 (build_vector_from_val): Likewise.
3815 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
3816 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3817 * tree-vect-generic.c (add_rshift): Likewise.
3818 (expand_vector_divmod): Likewise.
3819 (optimize_vector_constructor): Likewise.
3820 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3821 (vect_transform_slp_perm_load): Likewise.
3822 (vect_schedule_slp_instance): Likewise.
3823 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
3824 (vectorizable_call): Likewise.
3825 (vect_gen_perm_mask_any): Likewise. Add elements in order.
3826 * expmed.c (make_tree): Likewise.
3827 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
3828 a vector passed to build_vector.
3829 (fold_convert_const): Likewise.
3830 (exact_inverse): Likewise.
3831 (fold_ternary_loc): Likewise.
3832 (fold_relational_const): Likewise.
3833 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
3834 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
3835 (const_unop): Likewise. Store the reduction accumulator in a
3836 variable rather than an array.
3837 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
3838 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
3839 the new vector, rather than constructing it after the input arrays.
3840 (native_interpret_vector): Use auto_vec<tree> when building
3841 a vector passed to build_vector. Add elements in order.
3842 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
3843 auto_vec<tree> when building a vector passed to build_vector.
3844 (vect_create_epilog_for_reduction): Likewise.
3845 (vectorizable_induction): Likewise.
3846 (get_initial_def_for_reduction): Likewise. Fix indentation of
3848 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
3850 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
3851 passed to build_vector.
3853 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3854 Alan Hayward <alan.hayward@arm.com>
3855 David Sherwood <david.sherwood@arm.com>
3857 * tree-core.h (tree_base::u): Add an "nelts" field.
3858 (tree_vector): Use VECTOR_CST_NELTS as the length.
3859 * tree.c (tree_size): Likewise.
3860 (make_vector): Initialize VECTOR_CST_NELTS.
3861 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
3862 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
3863 TYPE_VECTOR_SUBPARTS.
3864 * expr.c (const_vector_mask_from_tree): Consistently use "units"
3865 as the number of units, setting it from VECTOR_CST_NELTS.
3866 (const_vector_from_tree): Likewise.
3867 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
3868 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
3869 (fold_negate_expr_1): Likewise.
3870 (fold_convert_const): Likewise.
3871 (const_binop): Likewise. Differentiate the number of output and
3873 (const_unop): Likewise.
3874 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
3875 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
3876 in cases that did the opposite.
3878 2017-09-14 Richard Biener <rguenther@suse.de>
3880 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
3883 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
3885 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
3887 2017-09-14 Jakub Jelinek <jakub@redhat.com>
3890 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
3891 if and where to split a bb, except for splitting before debug insn
3892 sequences followed by non-label real insn. Delete debug insns
3893 in between basic blocks.
3895 * combine.c (make_compound_operation_int): Formatting fixes.
3897 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
3898 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3899 * config/netbsd.h (LINK_EH_SPEC): Likewise.
3900 * config/sol2.h (LINK_EH_SPEC): Likewise.
3901 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3902 * config/s390/linux.h (LINK_SPEC): Likewise.
3903 * config/freebsd.h (LINK_EH_SPEC): Likewise.
3904 * config/openbsd.h (LINK_EH_SPEC): Likewise.
3905 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3906 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
3907 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
3908 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3909 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
3910 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
3912 2017-09-13 Jakub Jelinek <jakub@redhat.com>
3914 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
3916 (ENDFILE_LINUX_SPEC): Likewise.
3917 (LINK_EH_SPEC): Likewise.
3918 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
3919 (LINK_OS_LINUX_SPEC32): Likewise.
3920 (LINK_OS_LINUX_SPEC64): Likewise.
3921 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
3922 (LINK_OS_LINUX_SPEC): Likewise.
3924 2017-09-13 Martin Liska <mliska@suse.cz>
3927 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
3928 CASE_HIGH is NULL_TREE.
3930 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3931 Alan Hayward <alan.hayward@arm.com>
3932 David Sherwood <david.sherwood@arm.com>
3934 * target.def (secondary_memory_needed): New hook.
3935 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
3936 instead of SECONDARY_MEMORY_NEEDED.
3937 (secondary_memory_needed_mode): Likewise.
3938 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
3939 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
3940 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
3941 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
3942 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
3943 * doc/tm.texi: Regenerate.
3944 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
3945 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
3946 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3947 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
3948 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
3949 * config/i386/i386.c (inline_secondary_memory_needed): Put the
3950 mode argument first and change the reg_class arguments to reg_class_t.
3951 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
3952 Make static. Update the call to inline_secondary_memory_needed.
3953 (ix86_register_move_cost): Update the call to
3954 inline_secondary_memory_needed.
3955 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3956 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
3958 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
3959 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3961 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
3962 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
3963 * config/mips/mips.c (mips_secondary_memory_needed): Make static
3964 and match hook interface. Add comment from mips.h.
3965 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3966 * config/mmix/mmix.md (truncdfsf2): Refer to
3967 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3969 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
3970 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
3971 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3972 (pa_secondary_memory_needed): New function.
3973 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
3974 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
3975 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3976 (pdp11_secondary_memory_needed): Make static and match hook interface.
3977 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
3978 * config/powerpcspe/powerpcspe-protos.h
3979 (rs6000_secondary_memory_needed_ptr): Delete.
3980 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
3982 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3983 (rs6000_option_override_internal): Assign to
3984 targetm.secondary_memory_needed rather than
3985 rs6000_secondary_memory_needed_ptr.
3986 (rs6000_secondary_memory_needed): Match hook interface.
3987 (rs6000_debug_secondary_memory_needed): Likewise.
3988 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
3989 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
3990 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
3991 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3992 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
3993 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
3995 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
3996 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3997 (rs6000_option_override_internal): Assign to
3998 targetm.secondary_memory_needed rather than
3999 rs6000_secondary_memory_needed_ptr.
4000 (rs6000_secondary_memory_needed): Match hook interface.
4001 (rs6000_debug_secondary_memory_needed): Likewise.
4002 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
4003 * config/s390/s390.c (s390_secondary_memory_needed): New function.
4004 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4005 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
4006 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4007 (sparc_secondary_memory_needed): New function.
4008 * lra-constraints.c (check_and_process_move): Refer to
4009 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4011 (curr_insn_transform): Likewise.
4012 (process_alt_operands): Use targetm.secondary_memory_needed
4013 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4014 (check_secondary_memory_needed_p): Likewise.
4015 (choose_split_class): Likewise.
4016 * reload.c: Unconditionally include code that was previously
4017 conditional on SECONDARY_MEMORY_NEEDED.
4018 (push_secondary_reload): Use targetm.secondary_memory_needed
4019 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4020 (push_reload): Likewise.
4021 * reload1.c: Unconditionally include code that was previously
4022 conditional on SECONDARY_MEMORY_NEEDED.
4023 (choose_reload_regs): Use targetm.secondary_memory_needed
4024 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4025 (gen_reload): Likewise.
4026 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
4028 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4029 Alan Hayward <alan.hayward@arm.com>
4030 David Sherwood <david.sherwood@arm.com>
4032 * target.def (secondary_memory_needed_mode): New hook:
4033 * targhooks.c (default_secondary_memory_needed_mode): Declare.
4034 * targhooks.h (default_secondary_memory_needed_mode): New function.
4035 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
4036 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
4037 * doc/tm.texi: Regenerate.
4038 * lra-constraints.c (check_and_process_move): Use
4039 targetm.secondary_memory_needed_mode instead of
4040 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
4041 (curr_insn_transform): Likewise.
4042 * reload.c (get_secondary_mem): Likewise.
4043 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4044 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
4046 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4047 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4048 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
4049 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4050 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
4052 * config/powerpcspe/powerpcspe-protos.h
4053 (rs6000_secondary_memory_needed_mode): Delete.
4054 * config/powerpcspe/powerpcspe.c
4055 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4056 (rs6000_secondary_memory_needed_mode): Make static.
4057 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4058 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
4060 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4062 (rs6000_secondary_memory_needed_mode): Make static.
4063 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4064 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
4065 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4066 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4067 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4069 (sparc_secondary_memory_needed_mode): New function.
4070 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
4072 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
4074 * config/aarch64/constraints.md (Umq): New constraint.
4075 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
4077 (mov<mode>): Update condition.
4079 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4081 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
4082 when bitposition is the same.
4084 2017-09-13 Richard Biener <rguenther@suse.de>
4086 * dwarf2out.c (output_die_symbol): Remove.
4087 (output_die): Do not output a DIEs symbol.
4089 2017-09-13 Richard Biener <rguenther@suse.de>
4092 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
4093 default-def to avoid breaking iterator update with the weird
4094 interaction with cgraph_update_edges_for_call_stmt_node.
4096 2017-09-13 Richard Biener <rguenther@suse.de>
4098 * tree-cfg.c (verify_gimple_assign_binary): Add verification
4099 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
4100 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
4101 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
4103 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
4105 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
4106 Disable pc relative literal load irrespective of
4107 TARGET_FIX_ERR_A53_84341 for default.
4109 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
4111 * config/sparc/sparc.c (output_return): Output the source location of
4112 the insn in the delay slot, if any.
4113 (output_sibcall): Likewise.
4115 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
4118 * common.opt (-static-pie): New alias.
4119 (shared): Negate static-pie.
4120 (-no-pie): Update help text.
4122 (static-pie): New option.
4123 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
4124 -static-pie support.
4125 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
4126 (LINK_EH_SPEC): Likewise.
4127 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4128 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4129 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4130 * gcc.c (LINK_COMMAND_SPEC): Likewise.
4131 (init_gcc_specs): Likewise.
4132 (init_spec): Likewise.
4133 (display_help): Update help message for -pie.
4134 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
4137 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
4139 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
4140 (movdi_aarch64): Likewise.
4141 (movti_aarch64): Likewise.
4143 2017-09-12 Simon Wright <simon@pushface.org>
4146 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
4147 calculation of the minor version, always output as 0.
4149 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4152 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4153 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
4154 on it early, rather than manual conversion late. For
4155 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
4156 instead of performing manual conversion.
4158 2017-09-12 Carl Love <cel@us.ibm.com>
4160 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
4161 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
4163 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4164 VMULOSW): Add definitions.
4165 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4166 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4167 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4168 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4169 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4171 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4173 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
4175 (movti_aarch64): Likewise.
4176 (movdf_aarch64): Likewise.
4177 (movtf_aarch64): Likewise.
4178 (load_pairdi): Likewise.
4179 (store_pairdi): Likewise.
4180 (load_pairdf): Likewise.
4181 (store_pairdf): Likewise.
4182 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
4183 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
4184 (ldr_got_small_<mode>): Likewise.
4185 (ldr_got_small_28k_<mode>): Likewise.
4186 (ldr_got_tiny): Likewise.
4187 * config/aarch64/iterators.md (ldst_sz): New.
4188 (ldpstp_sz): Likewise.
4189 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
4191 (thunderx_load): Split load_8 to load_16.
4192 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
4194 (thunderx2t99_storepair_basic): Split store_8 to store_16.
4195 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
4196 (xgene1_store_pair): Split store_8 to store_16.
4197 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
4198 (falkor_st_0_st_sd): Split store_8 to store_16.
4200 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4202 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
4203 and store1/2/3/4 to store_4/8/12/16.
4204 * config/aarch64/aarch64.md: Update for rename.
4205 * config/arm/arm.md: Likewise.: Likewise.
4206 * config/arm/arm.c: Likewise.
4207 * config/arm/thumb1.md: Likewise.
4208 * config/arm/thumb2.md: Likewise.
4209 * config/arm/vfp.md: Likewise.
4210 * config/arm/arm-generic.md: Likewise.
4211 * config/arm/arm1020e.md: Likewise.
4212 * config/arm/arm1026ejs.md: Likewise.
4213 * config/arm/arm1136jfs.md: Likewise.
4214 * config/arm/arm926ejs.md: Likewise.
4215 * config/arm/cortex-a15.md: Likewise.
4216 * config/arm/cortex-a17.md: Likewise.
4217 * config/arm/cortex-a5.md: Likewise.
4218 * config/arm/cortex-a53.md: Likewise.
4219 * config/arm/cortex-a57.md: Likewise.
4220 * config/arm/cortex-a7.md: Likewise.
4221 * config/arm/cortex-a8.md: Likewise.
4222 * config/arm/cortex-a9.md: Likewise.
4223 * config/arm/cortex-m4.md: Likewise.
4224 * config/arm/cortex-m7.md: Likewise.
4225 * config/arm/cortex-r4.md: Likewise.
4226 * config/arm/exynos-m1.md: Likewise.
4227 * config/arm/fa526.md: Likewise.
4228 * config/arm/fa606te.md: Likewise.
4229 * config/arm/fa626te.md: Likewise.
4230 * config/arm/fa726te.md: Likewise.
4231 * config/arm/fmp626.md: Likewise.
4232 * config/arm/iwmmxt.md: Likewise.
4233 * config/arm/ldmstm.md: Likewise.
4234 * config/arm/marvell-pj4.md: Likewise.
4235 * config/arm/xgene1.md: Likewise.
4236 * config/aarch64/thunderx.md: Likewise.
4237 * config/aarch64/thunderx2t99.md: Likewise.
4238 * config/aarch64/falkor.md: Likewise.
4240 2017-09-12 Martin Liska <mliska@suse.cz>
4242 * attribs.c (private_lookup_attribute): New function.
4243 * attribs.h (private_lookup_attribute): Declared here.
4244 (lookup_attribute): Called from this place.
4246 2017-09-12 Richard Biener <rguenther@suse.de>
4248 PR tree-optimization/82157
4249 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
4250 stmts with side-effects.
4252 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4253 Alan Hayward <alan.hayward@arm.com>
4254 David Sherwood <david.sherwood@arm.com>
4256 * target.def (hard_regno_nregs): New hook.
4257 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
4258 * targhooks.h (default_hard_regno_nregs): Declare.
4259 * targhooks.c (default_hard_regno_nregs): New function.
4260 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
4261 (TARGET_HARD_REGNO_NREGS): ...this hook.
4262 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
4263 (CLASS_MAX_NREGS): Likewise.
4264 * doc/tm.texi: Regenerate.
4265 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
4266 instead of HARD_REGNO_NREGS.
4267 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
4268 HARD_REGNO_NREGS in the comment.
4269 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
4270 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
4271 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
4272 Return an unsigned int.
4273 (TARGET_HARD_REGNO_NREGS): Redefine.
4274 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
4275 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
4276 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4277 (arc_hard_regno_nregs): New function.
4278 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
4279 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
4280 (arm_hard_regno_nregs): New function.
4281 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
4282 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
4283 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
4284 (TARGET_HARD_REGNO_NREGS): Redefine.
4285 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
4286 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
4287 (HARD_REGNO_NREGS): Delete.
4288 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
4289 (cr16_hard_regno_nregs): New function.
4290 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
4291 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
4292 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
4293 (cris_hard_regno_nregs): New function.
4294 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
4295 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
4296 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4297 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
4298 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
4299 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
4300 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
4301 (frv_hard_regno_nregs): Make static. Take and return an
4303 (frv_class_max_nregs): Remove outdated copy of documentation.
4304 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
4305 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
4306 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
4307 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
4308 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
4309 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
4310 (TARGET_HARD_REGNO_NREGS): Redefine.
4311 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
4312 (CLASS_MAX_NREGS): Update comment.
4313 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
4314 (ia64_hard_regno_nregs): New function.
4315 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
4316 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
4317 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
4318 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
4319 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
4321 (m32c_hard_regno_nregs): Likewise. Make static.
4322 (TARGET_HARD_REGNO_NREGS): Redefine.
4323 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
4324 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
4325 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
4326 (m68k_hard_regno_nregs): New function.
4327 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
4328 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
4329 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
4330 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
4331 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
4332 Take and return an unsigned int.
4333 (TARGET_HARD_REGNO_NREGS): Redefine.
4334 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
4335 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4336 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
4337 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
4338 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
4339 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
4340 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
4341 (msp430_hard_regno_nregs): Make static. Take and return an
4343 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
4344 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
4345 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
4346 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
4347 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
4348 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
4349 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
4350 (TARGET_HARD_REGNO_NREGS): Redefine.
4351 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
4352 (PA_HARD_REGNO_NREGS): ...this.
4353 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
4354 (PA_HARD_REGNO_NREGS): ...this.
4355 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4356 (pa_hard_regno_nregs): New function.
4357 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
4358 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
4359 (pdp11_hard_regno_nregs): New function.
4360 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
4361 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
4362 (rs6000_hard_regno_nregs_hook): New function.
4363 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
4364 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4365 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
4366 Take and return an unsigned int. Move earlier in file.
4367 (TARGET_HARD_REGNO_NREGS): Redefine.
4368 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
4369 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
4370 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
4371 (rl78_hard_regno_nregs): Make static. Take and return an
4373 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
4374 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
4375 (rs6000_hard_regno_nregs_hook): New function.
4376 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
4377 * config/rx/rx.c (rx_hard_regno_nregs): New function.
4378 (TARGET_HARD_REGNO_NREGS): Redefine.
4379 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
4380 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
4381 instead of HARD_REGNO_NREGS.
4382 (s390_hard_regno_nregs): New function.
4383 (s390_hard_regno_mode_ok): Add comment from s390.h.
4384 (TARGET_HARD_REGNO_NREGS): Redefine.
4385 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
4386 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
4387 (sh_hard_regno_nregs): New function.
4388 (sh_pass_in_reg_p): Use it.
4389 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
4390 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4391 (sparc_hard_regno_nregs): New function.
4392 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
4393 * config/spu/spu.c (spu_hard_regno_nregs): New function.
4394 (spu_function_arg_advance): Use it, supplying a valid register number.
4395 (TARGET_HARD_REGNO_NREGS): Redefine.
4396 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
4397 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
4398 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
4399 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
4400 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
4401 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
4402 (CLASS_MAX_NREGS): Remove copy of old documentation.
4403 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
4404 (visium_hard_regno_nregs): New function.
4405 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
4406 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
4407 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4408 xtensa_hard_regno_nregs): New function.
4409 * system.h (HARD_REGNO_NREGS): Poison.
4411 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4413 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
4414 hard_regno_nregs instead of HARD_REGNO_NREGS.
4415 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
4416 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
4417 (c6x_expand_epilogue): Likewise.
4418 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
4419 (frv_read_iacc_argument): Likewise.
4420 * config/sh/sh.c: Include regs.h.
4421 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4422 (regs_used): Likewise.
4423 (output_stack_adjust): Likewise.
4424 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
4425 * expmed.c: Include regs.h.
4426 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4427 * ree.c: Include regs.h.
4428 (combine_reaching_defs): Use hard_regno_nregs instead of
4430 (add_removable_extension): Likewise.
4432 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4434 * regs.h (hard_regno_nregs): Turn into a function.
4435 (end_hard_regno): Update accordingly.
4436 * caller-save.c (setup_save_areas): Likewise.
4437 (save_call_clobbered_regs): Likewise.
4438 (replace_reg_with_saved_mem): Likewise.
4439 (insert_restore): Likewise.
4440 (insert_save): Likewise.
4441 * combine.c (can_change_dest_mode): Likewise.
4442 (move_deaths): Likewise.
4443 (distribute_notes): Likewise.
4444 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
4445 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
4446 (rs6000_split_multireg_move): Likewise.
4447 (rs6000_register_move_cost): Likewise.
4448 (rs6000_memory_move_cost): Likewise.
4449 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
4450 (rs6000_split_multireg_move): Likewise.
4451 (rs6000_register_move_cost): Likewise.
4452 (rs6000_memory_move_cost): Likewise.
4453 * cselib.c (cselib_reset_table): Likewise.
4454 (cselib_lookup_1): Likewise.
4455 * emit-rtl.c (set_mode_and_regno): Likewise.
4456 * function.c (aggregate_value_p): Likewise.
4457 * ira-color.c (setup_profitable_hard_regs): Likewise.
4458 (check_hard_reg_p): Likewise.
4459 (calculate_saved_nregs): Likewise.
4460 (assign_hard_reg): Likewise.
4461 (improve_allocation): Likewise.
4462 (calculate_spill_cost): Likewise.
4463 * ira-emit.c (modify_move_list): Likewise.
4464 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
4465 (ira_hard_reg_in_set_p): Likewise.
4466 * ira.c (setup_reg_mode_hard_regset): Likewise.
4467 (clarify_prohibited_class_mode_regs): Likewise.
4468 (check_allocation): Likewise.
4469 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4470 (lra_setup_reg_renumber): Likewise.
4471 (setup_try_hard_regno_pseudos): Likewise.
4472 (spill_for): Likewise.
4473 (assign_hard_regno): Likewise.
4474 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
4475 * lra-constraints.c (in_class_p): Likewise.
4476 (lra_constraint_offset): Likewise.
4477 (simplify_operand_subreg): Likewise.
4478 (lra_constraints): Likewise.
4479 (split_reg): Likewise.
4480 (split_if_necessary): Likewise.
4481 (invariant_p): Likewise.
4482 (inherit_in_ebb): Likewise.
4483 * lra-lives.c (process_bb_lives): Likewise.
4484 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
4485 (get_hard_regs): Likewise.
4486 (do_remat): Likewise.
4487 * lra-spills.c (assign_spill_hard_regs): Likewise.
4488 * mode-switching.c (create_pre_exit): Likewise.
4489 * postreload.c (reload_combine_recognize_pattern): Likewise.
4490 * recog.c (peep2_find_free_register): Likewise.
4491 * regcprop.c (kill_value_regno): Likewise.
4492 (set_value_regno): Likewise.
4493 (copy_value): Likewise.
4494 (maybe_mode_change): Likewise.
4495 (find_oldest_value_reg): Likewise.
4496 (copyprop_hardreg_forward_1): Likewise.
4497 * regrename.c (check_new_reg_p): Likewise.
4498 (regrename_do_replace): Likewise.
4499 * reload.c (push_reload): Likewise.
4500 (combine_reloads): Likewise.
4501 (find_dummy_reload): Likewise.
4502 (operands_match_p): Likewise.
4503 (find_reloads): Likewise.
4504 (find_equiv_reg): Likewise.
4505 (reload_adjust_reg_for_mode): Likewise.
4506 * reload1.c (count_pseudo): Likewise.
4507 (count_spilled_pseudo): Likewise.
4508 (find_reg): Likewise.
4509 (clear_reload_reg_in_use): Likewise.
4510 (free_for_value_p): Likewise.
4511 (allocate_reload_reg): Likewise.
4512 (choose_reload_regs): Likewise.
4513 (reload_adjust_reg_for_temp): Likewise.
4514 (emit_reload_insns): Likewise.
4515 (delete_output_reload): Likewise.
4516 * rtlanal.c (subreg_get_info): Likewise.
4517 * sched-deps.c (sched_analyze_reg): Likewise.
4518 * sel-sched.c (init_regs_for_mode): Likewise.
4519 (mark_unavailable_hard_regs): Likewise.
4520 (choose_best_reg_1): Likewise.
4521 (verify_target_availability): Likewise.
4522 * valtrack.c (dead_debug_insert_temp): Likewise.
4523 * var-tracking.c (track_loc_p): Likewise.
4524 (emit_note_insn_var_location): Likewise.
4525 * varasm.c (make_decl_rtl): Likewise.
4526 * reginfo.c (choose_hard_reg_mode): Likewise.
4527 (init_reg_modes_target): Refer directly to
4528 this_target_regs->x_hard_regno_nregs.
4530 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4532 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
4533 instead of hard_regno_nregs.
4535 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4537 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
4538 end_hard_regno instead of hard_regno_nregs.
4539 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
4540 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
4541 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
4542 * ira-color.c (improve_allocation): Likewise.
4543 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4544 * lra-lives.c (mark_regno_live): Likewise.
4545 (mark_regno_dead): Likewise.
4546 * lra-remat.c (operand_to_remat): Likewise.
4547 * lra.c (collect_non_operand_hard_regs): Likewise.
4548 * postreload.c (reload_combine_note_store): Likewise.
4549 (move2add_valid_value_p): Likewise.
4550 * reload.c (regno_clobbered_p): Likewise.
4552 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4554 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
4556 * config/v850/v850.c (v850_reorg): Likewise.
4557 * reload.c (refers_to_regno_for_reload_p): Likewise.
4558 (find_equiv_reg): Likewise.
4559 * reload1.c (reload_reg_reaches_end_p): Likewise.
4561 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4563 * caller-save.c (add_used_regs): Use REG_NREGS instead of
4565 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
4566 * config/arm/arm.c (output_move_neon): Likewise.
4567 (arm_attr_length_move_neon): Likewise.
4568 (neon_split_vcombine): Likewise.
4569 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
4570 (c6x_mark_reg_written): Likewise.
4571 (c6x_dwarf_register_span): Likewise.
4572 * config/i386/i386.c (ix86_save_reg): Likewise.
4573 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
4574 (rws_access_reg): Likewise.
4575 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
4576 * mode-switching.c (create_pre_exit): Likewise.
4577 * ree.c (combine_reaching_defs): Likewise.
4578 (add_removable_extension): Likewise.
4579 * regcprop.c (find_oldest_value_reg): Likewise.
4580 (copyprop_hardreg_forward_1): Likewise.
4581 * reload.c (reload_inner_reg_of_subreg): Likewise.
4582 (push_reload): Likewise.
4583 (combine_reloads): Likewise.
4584 (find_dummy_reload): Likewise.
4585 (reload_adjust_reg_for_mode): Likewise.
4586 * reload1.c (find_reload_regs): Likewise.
4587 (forget_old_reloads_1): Likewise.
4588 (reload_reg_free_for_value_p): Likewise.
4589 (reload_adjust_reg_for_temp): Likewise.
4590 (emit_reload_insns): Likewise.
4591 (delete_output_reload): Likewise.
4592 * sel-sched.c (choose_best_reg_1): Likewise.
4593 (choose_best_pseudo_reg): Likewise.
4595 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4596 Alan Hayward <alan.hayward@arm.com>
4597 David Sherwood <david.sherwood@arm.com>
4599 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
4600 * target.def (slow_unaligned_access): New hook.
4601 * targhooks.h (default_slow_unaligned_access): Declare.
4602 * targhooks.c (default_slow_unaligned_access): New function.
4603 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
4604 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
4605 * doc/tm.texi: Regenerate.
4606 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
4607 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
4608 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
4610 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
4611 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
4613 (rs6000_slow_unaligned_access): New function.
4614 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4615 (expand_block_compare): Likewise.
4616 (expand_strn_compare): Likewise.
4617 (rs6000_rtx_costs): Likewise.
4618 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
4619 (riscv_slow_unaligned_access): Likewise.
4620 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
4621 (riscv_slow_unaligned_access_p): ...this and make static.
4622 (riscv_option_override): Update accordingly.
4623 (riscv_slow_unaligned_access): New function.
4624 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4625 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
4626 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4627 (rs6000_slow_unaligned_access): New function.
4628 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4629 (rs6000_rtx_costs): Likewise.
4630 * config/rs6000/rs6000-string.c (expand_block_compare)
4631 (expand_strn_compare): Use targetm.slow_unaligned_access instead
4632 of SLOW_UNALIGNED_ACCESS.
4633 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
4634 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
4635 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
4636 of SLOW_UNALIGNED_ACCESS.
4637 * expmed.c (simple_mem_bitfield_p): Likewise.
4638 * expr.c (alignment_for_piecewise_move): Likewise.
4639 (emit_group_load_1): Likewise.
4640 (emit_group_store): Likewise.
4641 (copy_blkmode_from_reg): Likewise.
4642 (emit_push_insn): Likewise.
4643 (expand_assignment): Likewise.
4644 (store_field): Likewise.
4645 (expand_expr_real_1): Likewise.
4646 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
4647 * lra-constraints.c (simplify_operand_subreg): Likewise.
4648 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4649 * gimple-ssa-store-merging.c: Likewise in block comment at start
4651 * tree-ssa-strlen.c: Include target.h.
4652 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
4653 of SLOW_UNALIGNED_ACCESS.
4654 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
4656 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4658 PR rtl-optimization/82185
4659 * expmed.c (emit_store_flag_int): Only test tem if it has been
4662 2017-09-12 Richard Biener <rguenther@suse.de>
4665 * match.pd ((FTYPE) N CMP CST): Fix typo.
4667 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
4669 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
4671 (mips_near_type_p): Add 'short_call' attribute as a synonym
4673 * doc/extend.texi (short_call): Document new function attribute.
4675 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4678 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
4679 assertion check that in the condition.
4680 (get_atomic_generic_size): Likewise. Before testing if parameter
4681 has pointer type, if it has array type, call for C++
4682 default_conversion to perform array-to-pointer conversion.
4684 2017-09-12 Richard Biener <rguenther@suse.de>
4686 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
4687 for operations we cannot scalarize.
4689 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
4691 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
4692 vectors heap vectors. Clean up comments.
4693 Make visited_bbs a reference.
4694 (profitable_jump_thread_path): Make GC
4695 vectors heap vectors. Clean up comments.
4697 (convert_and_register_jump_thread_path): Make GC vectors heap
4699 (check_subpath_and_update_thread_path): Same. Clean up comments.
4700 Make visited_bbs a reference.
4701 (handle_phi): Abstract common code to to
4702 register_jump_thread_path_if_profitable.
4703 Rename VAR_BB to DEF_BB.
4705 Make GC vectors heap vectors.
4706 Make visited_bbs a reference.
4707 (handle_assignment): Same.
4708 (register_jump_thread_path_if_profitable): New.
4709 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
4711 Make GC vectors heap vectors. Clean up comments.
4712 Make visited_bbs a reference.
4713 (find_jump_threads_backwards): Make visited_bbs live in the stack.
4714 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
4717 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
4720 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
4721 words of E_DImode object are reachable by xtensa_uimm8x4 access.
4723 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
4725 Revert r251800 and r251799.
4727 2017-09-11 Martin Jambor <mjambor@suse.cz>
4730 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
4731 arguments in advance.
4732 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
4733 use it to find predecessor edges.
4734 (naive_outof_ssa): Collect vector of predecessors.
4736 2017-09-08 Jason Merrill <jason@redhat.com>
4738 PR c++/70029 - ICE with ref-qualifier and -flto
4739 * langhooks.h (struct lang_hooks_for_types): Add
4740 copy_lang_qualifiers.
4741 * attribs.c (build_type_attribute_qual_variant): Use it.
4742 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
4744 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
4745 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
4747 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
4750 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
4751 (*mulsi3_sp64): New instruction.
4752 (mulsi3): New expander.
4754 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
4756 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
4758 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4760 * sancov.c: Include memmodel.h.
4762 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
4765 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
4768 2017-09-07 Carl Love <cel@us.ibm.com>
4770 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
4771 the sldi instruction.
4773 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
4775 * sancov.c: Include tm_p.h.
4777 2017-09-07 Jakub Jelinek <jakub@redhat.com>
4780 * output.h (switch_to_other_text_partition): New declaration.
4781 * varasm.c (switch_to_other_text_partition): New function.
4782 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
4783 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
4784 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
4785 to the other text partition before emitting LCL label and switch back
4786 after emitting the word after it.
4788 2017-09-07 Richard Biener <rguenther@suse.de>
4790 * passes.def (pass_split_crit_edges): Remove instance before PRE.
4791 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
4792 critical edges here, after loop init.
4793 (pass_data_pre): Remove PROP_no_crit_edges flags.
4794 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
4795 for valueization of call args to avoid leaking VN_TOP.
4796 (visit_use): Assert we do not visit default defs.
4797 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
4798 Use error_mark_node to more easily detect leaking VN_TOP.
4799 All default-defs are varying, not VN_TOP. Mark them visited.
4800 (run_scc_vn): Make code match comment.
4802 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4804 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
4805 OPTION_MASK_FLOAT128_KEYWORD.
4806 (POWERPC_MASKS): Likewise.
4807 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
4808 support for the -mfloat128-type option, and make -mfloat128
4809 default on PowerPC Linux systems. Define or undefine
4810 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
4811 Define __float128 to be __ieee128 if IEEE 128-bit support is
4812 enabled, or undefine it.
4813 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
4814 Delete defining __FLOAT128_TYPE__.
4815 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
4816 -mfloat128-type option and make -mfloat128 default on PowerPC
4818 (TARGET_FLOAT128_TYPE): Likewise.
4819 (-mfloat128-type): Likewise.
4820 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4821 Delete the -mfloat128-type option and make -mfloat128 default on
4822 PowerPC Linux systems. Always use __ieee128 to be the keyword for
4823 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
4824 128-bit floating point is enabled. Change tests from using
4825 -mfloat128-type to -mfloat128.
4826 (rs6000_mangle_type): Use the correct mangling for the __float128
4827 type even if normal long double is restricted to 64-bits.
4828 (floatn_mode): Enable the _Float128 type by default on VSX Linux
4830 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
4831 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
4832 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
4834 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
4835 documentation for -mfloat128.
4837 2017-09-06 Olivier Hainque <hainque@adacore.com>
4839 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
4841 2017-09-06 Wish Wu <wishwu007@gmail.com>
4842 Jakub Jelinek <jakub@redhat.com>
4844 * asan.c (initialize_sanitizer_builtins): Add
4845 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
4846 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
4847 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
4848 BT_FN_VOID_UINT64_PTR variables.
4849 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
4850 (BT_FN_VOID_UINT16_UINT16): Likewise.
4851 (BT_FN_VOID_UINT32_UINT32): Likewise.
4852 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
4853 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
4854 (BT_FN_VOID_UINT64_PTR): Likewise.
4855 * common.opt (flag_sanitize_coverage): New variable.
4856 (fsanitize-coverage=trace-pc): Remove.
4857 (fsanitize-coverage=): Add.
4858 * flag-types.h (enum sanitize_coverage_code): New enum.
4859 * fold-const.c (fold_range_test): Disable non-short-circuit
4860 optimization if flag_sanitize_coverage.
4861 (fold_truth_andor): Likewise.
4862 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
4863 * opts.c (COVERAGE_SANITIZER_OPT): Define.
4864 (coverage_sanitizer_opts): New array.
4865 (get_closest_sanitizer_option): Add OPTS argument, handle also
4866 OPT_fsanitize_coverage_.
4867 (parse_sanitizer_options): Adjusted to also handle
4868 OPT_fsanitize_coverage_.
4869 (common_handle_option): Add OPT_fsanitize_coverage_.
4870 * sancov.c (instrument_comparison, instrument_switch): New function.
4871 (sancov_pass): Add trace-cmp support.
4872 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
4873 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
4874 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
4875 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
4876 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
4877 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
4878 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
4879 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
4880 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
4881 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
4883 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4885 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4886 error. Only quit immediately if parsing is complete.
4887 (BEGIN): Initialize fatal_err and parse_done.
4888 (begin fpu, end fpu): Check number of arguments.
4889 (begin arch, end arch): Likewise.
4890 (begin cpu, end cpu): Likewise.
4891 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4892 (optalias): Likewise.
4894 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4896 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4897 * config/arm/arm-isa.h: Delete. Move definitions to ...
4898 * arm-cpus.in: ... here. Use new feature and fgroup values.
4899 * config/arm/arm.c (arm_option_override): Use lower case for feature
4901 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4902 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4903 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4905 (print_isa_bits_for): New function.
4906 (gen_isa): New function.
4907 (gen_comm_data): Use print_isa_bits_for.
4908 (define feature): New keyword.
4909 (define fgroup): New keyword.
4910 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
4911 (arm-isa.h): Add rule to generate file.
4912 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4913 case for feature bit names.
4915 2017-09-06 Richard Biener <rguenther@suse.de>
4917 * tree-ssa-pre.c (NECESSARY): Remove.
4918 (create_expression_by_pieces): Do not touch pass-local flags.
4919 (insert_into_preds_of_block): Likewise.
4920 (do_pre_regular_insertion): Likewise.
4921 (eliminate_insert): Likewise.
4922 (eliminate_dom_walker::before_dom_children): Likewise.
4923 (fini_eliminate): Do not look at inserted_exprs.
4924 (mark_operand_necessary): Remove.
4925 (remove_dead_inserted_code): Replace with simple work-list
4926 algorithm based on inserted_exprs and SSA uses.
4927 (pass_pre::execute): Re-order fini_eliminate and
4928 remove_dead_inserted_code.
4930 2017-09-06 Olivier Hainque <hainque@adacore.com>
4932 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
4933 for VxWorks 7. Adjust surrounding comments.
4935 2017-09-06 Richard Biener <rguenther@suse.de>
4937 * gimple-ssa-strength-reduction.c
4938 (find_candidates_dom_walker::before_dom_children): Also allow
4941 2017-09-06 Richard Biener <rguenther@suse.de>
4943 PR tree-optimization/82108
4944 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
4945 for gap in the non-permutation SLP case.
4947 2017-09-06 Martin Jambor <mjambor@suse.cz>
4949 PR tree-optimization/82078
4950 * tree-sra.c (sort_and_splice_var_accesses): Move call to
4951 add_access_to_work_queue...
4952 (build_accesses_from_assign): ...here.
4953 (propagate_all_subaccesses): Make sure racc is the group
4954 representative, if there is one.
4956 2017-09-06 Jakub Jelinek <jakub@redhat.com>
4959 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
4962 2017-09-06 Richard Biener <rguenther@suse.de>
4964 * gimple-ssa-strength-reduction.c
4965 (find_candidates_dom_walker::before_doom_children): Use a
4966 type and not a mode check.
4968 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4971 * config/arm/predicates.md (arm_general_adddi_operand): Create new
4973 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
4975 2017-09-05 Jeff Law <law@redhat.com>
4977 PR tree-optimization/64910
4978 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
4979 cases where we have 3 or more operands.
4981 2017-09-05 Jakub Jelinek <jakub@redhat.com>
4984 * omp-low.c (lower_omp_for): Recompute tree invariant if
4985 gimple_omp_for_initial/final is ADDR_EXPR.
4988 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
4989 into gimple val before gimplification fo the COND_EXPR.
4991 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
4993 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
4994 REGION_COPY argument.
4995 (thread_through_all_blocks): Remove unused argument to
4996 duplicate_thread_path.
4998 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4999 Alan Hayward <alan.hayward@arm.com>
5000 David Sherwood <david.sherwood@arm.com>
5002 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
5003 Take a scalar_mode rather than a machine_mode.
5004 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5005 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
5006 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5007 (aarch64_gen_adjusted_ldpstp): Likewise.
5008 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
5010 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5011 Alan Hayward <alan.hayward@arm.com>
5012 David Sherwood <david.sherwood@arm.com>
5014 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
5015 Take a scalar_int_mode instead of a machine_mode.
5016 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5017 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5018 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5019 (aarch64_simd_attr_length_rglist): Delete.
5020 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
5021 a scalar_int_mode instead of a machine_mode.
5022 (aarch64_add_offset): Likewise.
5023 (aarch64_internal_mov_immediate): Likewise
5024 (aarch64_add_constant_internal): Likewise.
5025 (aarch64_add_constant): Likewise.
5026 (aarch64_movw_imm): Likewise.
5027 (aarch64_rtx_arith_op_extract_p): Likewise.
5028 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5029 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5030 Remove assert that the mode isn't a vector.
5031 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5032 (aarch64_expand_mov_immediate): Update calls after above changes.
5033 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
5034 (aarch64_and_bitmask_imm): Check for scalar integer modes.
5035 (aarch64_move_imm): Likewise.
5036 (aarch64_can_const_movi_rtx_p): Likewise.
5037 (aarch64_strip_extend): Likewise.
5038 (aarch64_extr_rtx_p): Likewise.
5039 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
5040 a CONST_INT when the mode parameter is VOIDmode.
5041 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
5043 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5045 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
5046 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5047 (bitwise_type_for_mode): Update accordingly.
5049 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5051 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
5052 * stor-layout.c (mode_for_size_tree): Likewise.
5053 (mode_for_array): Update accordingly.
5054 (layout_decl): Likewise.
5055 (compute_record_mode): Likewise. Only set the mode once.
5057 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5059 * target.def (get_mask_mode): Change return type to opt_mode.
5061 * doc/tm.texi: Regenerate.
5062 * targhooks.h (default_get_mask_mode): Return an opt_mode.
5063 * targhooks.c (default_get_mask_mode): Likewise.
5064 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5065 * optabs-query.c (can_vec_mask_load_store_p): Update use of
5066 targetm.get_mask_mode.
5067 * tree.c (build_truth_vector_type): Likewise.
5069 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5071 * machmode.h (mode_for_vector): Return an opt_mode.
5072 * stor-layout.c (mode_for_vector): Likewise.
5073 (mode_for_int_vector): Update accordingly.
5074 (layout_type): Likewise.
5075 * config/i386/i386.c (emit_memmov): Likewise.
5076 (ix86_expand_set_or_movmem): Likewise.
5077 (ix86_expand_vector_init): Likewise.
5078 (ix86_get_mask_mode): Likewise.
5079 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
5081 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
5082 * expmed.c (extract_bit_field_1): Likewise.
5083 * expr.c (expand_expr_real_2): Likewise.
5084 * optabs-query.c (can_vec_perm_p): Likewise.
5085 (can_vec_mask_load_store_p): Likewise.
5086 * optabs.c (expand_vec_perm): Likewise.
5087 * targhooks.c (default_get_mask_mode): Likewise.
5088 * tree-vect-stmts.c (vectorizable_store): Likewise.
5089 (vectorizable_load): Likewise.
5090 (get_vectype_for_scalar_type_and_size): Likewise.
5092 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5094 * machmode.h (mode_for_int_vector): New function.
5095 * stor-layout.c (mode_for_int_vector): Likewise.
5096 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
5097 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
5098 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5099 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5100 (s390_expand_vcond): Likewise.
5102 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5104 * machmode.h (opt_machine_mode): New type.
5105 (opt_mode<T>): Allow construction from anything that can be
5107 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
5108 (mode_for_size): Return an opt_machine_mode.
5109 * stor-layout.c (mode_for_size): Likewise.
5110 (mode_for_size_tree): Update call accordingly.
5111 (bitwise_mode_for_mode): Likewise.
5112 (make_fract_type): Likewise.
5113 (make_accum_type): Likewise.
5114 * caller-save.c (replace_reg_with_saved_mem): Update call
5116 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5117 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5118 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5119 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5120 * expmed.c (extract_bit_field_1): Likewise.
5121 * reload.c (get_secondary_mem): Likewise.
5122 * varasm.c (assemble_integer): Likewise.
5123 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
5126 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5128 * machmode.h (decimal_float_mode_for_size): New function.
5129 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
5130 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
5131 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
5132 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
5133 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
5134 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
5136 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5138 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
5139 (get_builtin_sync_mode): Likewise.
5140 (expand_ifn_atomic_compare_exchange): Likewise.
5141 (expand_builtin_atomic_clear): Likewise.
5142 (expand_builtin_atomic_test_and_set): Likewise.
5143 (fold_builtin_atomic_always_lock_free): Likewise.
5144 * calls.c (compute_argument_addresses): Likewise.
5145 (emit_library_call_value_1): Likewise.
5146 (store_one_arg): Likewise.
5147 * combine.c (combine_instructions): Likewise.
5148 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
5149 * config/arm/arm.c (arm_function_value): Likewise.
5150 (aapcs_allocate_return_reg): Likewise.
5151 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
5152 * config/i386/i386.c (construct_container): Likewise.
5153 (ix86_gimplify_va_arg): Likewise.
5154 (ix86_expand_sse_cmp): Likewise.
5155 (emit_memmov): Likewise.
5156 (emit_memset): Likewise.
5157 (expand_small_movmem_or_setmem): Likewise.
5158 (ix86_expand_pextr): Likewise.
5159 (ix86_expand_pinsr): Likewise.
5160 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
5161 * config/microblaze/microblaze.c (microblaze_block_move_straight):
5163 * config/mips/mips.c (mips_function_value_1) Likewise.
5164 (mips_block_move_straight): Likewise.
5165 (mips_expand_ins_as_unaligned_store): Likewise.
5166 * config/powerpcspe/powerpcspe.c
5167 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5168 (rs6000_darwin64_record_arg_flush): Likewise.
5169 * config/rs6000/rs6000.c
5170 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5171 (rs6000_darwin64_record_arg_flush): Likewise.
5172 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
5173 (sparc_function_value_1): Likewise.
5174 * config/spu/spu.c (adjust_operand): Likewise.
5175 (spu_emit_branch_or_set): Likewise.
5176 (arith_immediate_p): Likewise.
5177 * emit-rtl.c (gen_lowpart_common): Likewise.
5178 * expr.c (expand_expr_real_1): Likewise.
5179 * function.c (assign_parm_setup_block): Likewise.
5180 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5181 * reload1.c (alter_reg): Likewise.
5182 * stor-layout.c (mode_for_vector): Likewise.
5183 (layout_type): Likewise.
5185 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5187 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
5188 (spu_convert_move): Likewise.
5189 * lower-subreg.c (resolve_simple_move): Likewise.
5191 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5194 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
5195 define_insn to a define_expand.
5196 (altivec_vsum2sws_direct): New define_insn.
5197 (altivec_vsumsws): Convert from a define_insn to a define_expand.
5199 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5201 * config/arm/arm.c (arm_option_params_internal): Improve setting of
5204 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
5209 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
5210 realign stack if stack alignment needed is less than incoming
5213 2017-09-05 Marek Polacek <polacek@redhat.com>
5216 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
5219 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5221 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
5223 2017-09-05 Richard Biener <rguenther@suse.de>
5225 PR tree-optimization/82084
5226 * fold-const.c (can_native_encode_string_p): Handle wide characters.
5228 2017-09-05 Richard Biener <rguenther@suse.de>
5230 PR tree-optimization/82102
5231 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
5233 2017-09-05 Martin Liska <mliska@suse.cz>
5235 PR tree-optimization/82032
5236 * tree-cfg.c (generate_range_test): New function.
5237 * tree-cfg.h (generate_range_test): Declared here.
5238 * tree-cfgcleanup.c (convert_single_case_switch): New function.
5239 (cleanup_control_expr_graph): Use it.
5240 * tree-switch-conversion.c (try_switch_expansion): Remove
5242 (emit_case_nodes): Use generate_range_test.
5244 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5247 * config/i386/i386.md (*<btsc><mode>_mask): Add
5248 TARGET_USE_BT to insn constraint.
5249 (*btr<mode>_mask): Ditto.
5251 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
5253 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
5254 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
5256 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5259 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
5260 TARGET_NEON and TARGET_IWMMXT.
5261 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
5262 TARGET_NEON and TARGET_IWMMXT.
5263 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
5265 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5267 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
5268 (ix86_rewrite_tls_address): Ditto.
5269 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
5270 (ix86_rewrite_tls_address_1): Ditto.
5271 (ix86_rewrite_tls_address): Ditto.
5272 * config/i386/predicates.md (tls_address_pattern): New predicate.
5273 * config/i386/i386.md (TLS address splitter): New splitter.
5275 2017-09-04 Richard Biener <rguenther@suse.de>
5277 PR tree-optimization/82084
5278 * fold-const.h (can_native_encode_string_p): Declare.
5279 * fold-const.c (can_native_encode_string_p): Factor out from ...
5280 (native_encode_string): ... here.
5281 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
5282 vectorizing stores from constants we later cannot handle.
5284 2017-09-04 Marek Polacek <polacek@redhat.com>
5287 * doc/invoke.texi: Update -Wtautological-compare documentation.
5289 2017-09-04 Jeff Law <law@redhat.com>
5291 PR tree-optimization/64910
5292 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
5293 swap the first and last operand if the last is a constant.
5295 2017-09-04 Marek Polacek <polacek@redhat.com>
5298 * convert.c (do_narrow): When sanitizing signed integer overflows,
5299 bail out for signed types.
5300 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
5302 2017-09-04 Richard Biener <rguenther@suse.de>
5304 PR tree-optimization/82060
5305 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
5306 Move devirtualization after stmt folding and before EH/AB/noreturn
5307 cleanup to get the stmt refs canonicalized. Use a bool instead
5308 of gimple_modified_p since that doesn't work for NOPs. Schedule
5309 NOPs generated by folding for removal.
5311 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5312 Alan Hayward <alan.hayward@arm.com>
5313 David Sherwood <david.sherwood@arm.com>
5315 * coretypes.h (pad_direction): New enum.
5316 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
5317 (FUNCTION_ARG_PADDING): Likewise.
5318 * target.def (function_arg_padding): New hook.
5319 * targhooks.h (default_function_arg_padding): Declare.
5320 * targhooks.c (default_function_arg_padding): New function.
5321 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
5322 (TARGET_FUNCTION_ARG_PADDING): ...this.
5323 * doc/tm.texi: Regenerate.
5324 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
5325 instead of direction.
5326 (compute_argument_addresses): Likewise.
5327 (load_register_parameters): Likewise.
5328 (emit_library_call_value_1): Likewise.
5329 (store_one_arg): Use targetm.calls.function_arg_padding instead
5330 of FUNCTION_ARG_PADDING.
5331 (must_pass_in_stack_var_size_or_pad): Likewise.
5332 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
5333 (emit_group_store): Likewise.
5334 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
5335 instead of FUNCTION_ARG_PADDING.
5336 (emit_push_insn): Likewise, and propagate enum change throughout
5338 * function.h (direction): Delete.
5339 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
5341 * function.c (assign_parm_find_stack_rtl): Likewise.
5342 (assign_parm_setup_block_p): Likewise.
5343 (assign_parm_setup_block): Likewise.
5344 (gimplify_parameters): Likewise.
5345 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
5346 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
5348 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
5349 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5350 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
5351 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
5352 (aarch64_function_arg_padding): ...this new function.
5353 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
5354 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5355 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
5356 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5357 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
5358 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5359 (arm_pad_arg_upward): Replace with...
5360 (arm_function_arg_padding): ...this new function.
5361 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
5363 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
5364 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
5365 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5366 (ia64_hpux_function_arg_padding): Replace with...
5367 (ia64_function_arg_padding): ...this new function. Use pad_direction
5368 instead of direction. Check for TARGET_HPUX.
5369 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
5370 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5371 (iq2000_function_arg_padding): New function.
5372 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
5373 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
5374 (mips_function_arg_padding): ...this new function.
5375 (mips_pad_reg_upward): Update accordingly.
5376 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5377 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
5378 targetm.calls.function_arg_padding.
5379 (FUNCTION_ARG_PADDING): Delete.
5380 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5381 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
5382 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5383 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
5384 (nios2_block_reg_padding): Return pad_direction instead of direction.
5385 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
5386 instead of direction.
5387 (nios2_function_arg_padding): Likewise. Make static.
5388 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5389 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
5390 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
5391 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
5392 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5393 (pa_function_arg_padding): Make static. Return pad_direction instead
5395 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
5396 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5397 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
5398 instead of direction. Use targetm.calls.function_arg_padding.
5399 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
5400 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5401 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
5402 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
5403 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
5405 (function_arg_padding): Rename to...
5406 (rs6000_function_arg_padding): ...this. Make static. Return
5407 pad_direction instead of direction.
5408 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5409 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
5410 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5411 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
5412 instead of direction. Use targetm.calls.function_arg_padding.
5413 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
5414 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5415 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
5416 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
5417 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5418 (function_arg_padding): Rename to...
5419 (rs6000_function_arg_padding): ...this. Make static. Return
5420 pad_direction instead of direction.
5421 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5422 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
5423 * config/s390/s390.c (s390_function_arg_padding): New function.
5424 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5425 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
5426 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
5427 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5428 (function_arg_padding): Rename to...
5429 (sparc_function_arg_padding): ...this. Make static. Return
5430 pad_direction instead of direction.
5431 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
5432 * config/spu/spu.c (spu_function_arg_padding): New function.
5433 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5434 * system.h (FUNCTION_ARG_PADDING): Poison.
5436 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5437 Alan Hayward <alan.hayward@arm.com>
5438 David Sherwood <david.sherwood@arm.com>
5440 * target.def (modes_tieable_p): New hook.
5441 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
5442 (TARGET_MODES_TIEABLE_P): ...this.
5443 * doc/tm.texi.in: Regenerate.
5444 * hooks.h (hook_bool_mode_mode_true): Declare.
5445 * hooks.c (hook_bool_mode_mode_true): New function.
5446 * combine.c (subst): Use targetm.modes_tieable_p instead of
5448 * dse.c (find_shift_sequence): Likewise.
5449 * expmed.c (extract_low_bits): Likewise.
5450 * lower-subreg.c: Include target.h.
5451 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
5453 * rtlanal.c (rtx_cost): Likewise.
5454 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
5455 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
5456 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
5457 (TARGET_MODES_TIEABLE_P): Redefine.
5458 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
5459 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
5460 (TARGET_MODES_TIEABLE_P): Redefine.
5461 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
5462 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
5463 (arc_modes_tieable_p): New function.
5464 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
5465 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
5466 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
5467 (arm_modes_tieable_p): Make static.
5468 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
5469 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
5470 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
5471 (TARGET_MODES_TIEABLE_P): Redefine.
5472 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
5473 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
5474 (TARGET_MODES_TIEABLE_P): Redefine.
5475 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
5476 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
5477 (cr16_modes_tieable_p): New function.
5478 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
5479 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
5480 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
5481 (TRULY_NOOP_TRUNCATION): Update comment.
5482 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
5483 (TRULY_NOOP_TRUNCATION): Update comment.
5484 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
5485 (frv_modes_tieable_p): New function.
5486 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
5487 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
5488 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
5489 (TARGET_MODES_TIEABLE_P): Redefine.
5490 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
5491 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
5492 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
5493 (TARGET_MODES_TIEABLE_P): Redefine.
5494 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
5495 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
5496 (ia64_modes_tieable_p): New function.
5497 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
5498 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
5499 (iq2000_modes_tieable_p): New function.
5500 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
5501 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
5502 (lm32_modes_tieable_p): New function.
5503 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
5504 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
5505 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
5506 (TARGET_MODES_TIEABLE_P): Redefine.
5507 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
5508 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
5509 (m32r_modes_tieable_p): New function.
5510 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
5511 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
5512 (m68k_modes_tieable_p): New function.
5513 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
5514 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
5515 (mcore_modes_tieable_p): New function.
5516 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
5517 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
5519 (TARGET_MODES_TIEABLE_P): Redefine.
5520 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
5521 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
5522 * config/mips/mips.c (mips_modes_tieable_p): Make static.
5523 (TARGET_MODES_TIEABLE_P): Redefine.
5524 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
5525 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
5526 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
5527 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
5528 (mn10300_modes_tieable_p): ...this and make static.
5529 (TARGET_MODES_TIEABLE_P): Redefine.
5530 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
5531 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
5532 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
5533 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
5534 (msp430_modes_tieable_p): Make static.
5535 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
5536 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
5537 (TARGET_MODES_TIEABLE_P): Redefine.
5538 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
5539 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
5540 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
5541 (TARGET_MODES_TIEABLE_P): Redefine.
5542 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
5543 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
5544 * config/pa/pa.c (pa_modes_tieable_p): Make static.
5545 (TARGET_MODES_TIEABLE_P): Redefine.
5546 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
5547 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
5548 (pdp11_modes_tieable_p): New function.
5549 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
5550 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
5551 (rs6000_modes_tieable_p): New function.
5552 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
5553 * config/powerpcspe/powerpcspe.md: Update comment.
5554 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
5555 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
5556 (TARGET_MODES_TIEABLE_P): Redefine.
5557 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
5558 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
5559 (rl78_modes_tieable_p): New function.
5560 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
5561 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
5562 (rs6000_modes_tieable_p): New function.
5563 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
5564 * config/rs6000/rs6000.md: Update comment.
5565 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
5566 * config/rx/rx.c (rx_modes_tieable_p): New function.
5567 (TARGET_MODES_TIEABLE_P): Redefine.
5568 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
5569 * config/s390/s390.c (s390_modes_tieable_p): New function.
5570 (TARGET_MODES_TIEABLE_P): Redefine.
5571 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
5572 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
5573 (sh_modes_tieable_p): New function.
5574 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
5575 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
5576 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
5577 (sparc_modes_tieable_p): Make static.
5578 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
5579 * config/spu/spu.c (spu_modes_tieable_p): New function.
5580 (TARGET_MODES_TIEABLE_P): Redefine.
5581 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
5582 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
5583 (TARGET_MODES_TIEABLE_P): Redefine.
5584 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
5585 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
5586 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
5587 * config/v850/v850.c (v850_modes_tieable_p): New function.
5588 (TARGET_MODES_TIEABLE_P): Redefine.
5589 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
5590 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
5591 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
5592 (visium_modes_tieable_p): New function.
5593 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
5594 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
5595 (xtensa_modes_tieable_p): New function.
5596 * system.h (MODES_TIEABLE_P): Poison.
5598 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5599 Alan Hayward <alan.hayward@arm.com>
5600 David Sherwood <david.sherwood@arm.com>
5602 * target.def (hard_regno_mode_ok): New hook.
5603 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
5604 (TARGET_HARD_REGNO_MODE_OK): ...this.
5605 * doc/tm.texi.in: Regenerate.
5606 * hooks.h (hook_bool_uint_mode_true): Declare.
5607 * hooks.c (hook_bool_uint_mode_true): New function.
5608 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
5610 * genpreds.c (write_insn_preds_c): Add an include of target.h.
5611 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
5612 instead of HARD_REGNO_MODE_OK.
5613 * caller-save.c: Include target.h.
5614 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
5616 * combine.c (can_combine_p): Likewise.
5617 (combinable_i3pat): Likewise.
5618 (can_change_dest_mode): Likewise.
5619 * expr.c (init_expr_target): Likewise.
5620 (convert_move): Likewise.
5621 (convert_modes): Likewise.
5622 * ira.c (setup_prohibited_class_mode_regs): Likewise.
5623 (setup_prohibited_mode_move_regs): Likewise.
5624 * ira.h (target_ira): Likewise.
5625 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5626 * lra-constraints.c (process_alt_operands): Likewise.
5627 (split_reg): Likewise.
5628 * recog.c (peep2_find_free_register): Likewise.
5629 * ree.c (combine_reaching_defs): Likewise.
5630 * regcprop.c (maybe_mode_change): Likewise.
5631 * reginfo.c (init_reg_sets_1): Likewise.
5632 (choose_hard_reg_mode): Likewise.
5633 (simplifiable_subregs): Likewise.
5634 * regrename.c (check_new_reg_p): Likewise.
5635 * reload.c (find_valid_class): Likewise.
5636 (find_valid_class_1): Likewise.
5637 (reload_inner_reg_of_subreg): Likewise.
5638 (push_reload): Likewise.
5639 (combine_reloads): Likewise.
5640 (find_dummy_reload): Likewise.
5641 (find_reloads): Likewise.
5642 * reload1.c (find_reg): Likewise.
5643 (set_reload_reg): Likewise.
5644 (allocate_reload_reg): Likewise.
5645 (choose_reload_regs): Likewise.
5646 (reload_adjust_reg_for_temp): Likewise.
5647 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
5648 (simplify_subreg_regno): Likewise.
5649 * sel-sched.c (init_regs_for_mode): Likewise.
5650 * varasm.c (make_decl_rtl): Likewise.
5651 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
5652 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
5654 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
5655 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
5656 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5657 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
5658 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
5659 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5660 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
5661 (arc_mode_class): Delete.
5662 (HARD_REGNO_MODE_OK): Delete.
5663 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5664 (arc_hard_regno_mode_ok): Rename old array to...
5665 (arc_hard_regno_mode_ok_modes): ...this.
5666 (arc_conditional_register_usage): Update accordingly.
5667 (arc_mode_class): Make static.
5668 (arc_hard_regno_mode_ok): New function.
5669 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
5670 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
5671 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5672 (arm_hard_regno_mode_ok): Make static.
5673 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
5675 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
5676 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
5677 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
5679 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5680 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
5681 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
5682 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
5683 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
5684 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5685 * config/bfin/predicates.md (valid_reg_operand): Use
5686 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
5687 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
5688 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
5689 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5690 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
5691 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
5692 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5693 (cr16_hard_regno_mode_ok): Make static and return a bool.
5694 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
5695 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5696 (cris_hard_regno_mode_ok): New function.
5697 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
5698 (epiphany_mode_class): Delete.
5699 (HARD_REGNO_MODE_OK): Delete.
5700 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
5701 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5702 (hard_regno_mode_ok): Rename to...
5703 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
5704 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
5705 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
5707 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
5708 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
5709 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5710 (frv_hard_regno_mode_ok): Make static and return a bool.
5711 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
5713 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
5714 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
5715 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
5716 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
5718 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5719 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
5720 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
5721 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
5723 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5724 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
5725 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5726 (ia64_hard_regno_mode_ok): New function.
5727 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
5728 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5729 (iq2000_hard_regno_mode_ok): New function.
5730 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
5731 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5732 (lm32_hard_regno_mode_ok): New function.
5733 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
5734 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
5735 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
5736 instead of HARD_REGNO_MODE_OK.
5737 (m32c_hard_regno_ok): Rename to...
5738 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
5739 (m32c_cannot_change_mode_class): Update accordingly.
5740 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5741 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
5742 (m32r_mode_class): Delete.
5743 (HARD_REGNO_MODE_OK): Delete.
5744 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5745 (m32r_hard_regno_mode_ok): Rename to...
5746 (m32r_hard_regno_modes): ...this.
5747 (m32r_mode_class): Make static.
5748 (m32r_hard_regno_mode_ok): New function.
5749 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
5750 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
5751 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5752 (m68k_hard_regno_mode_ok): Make static.
5753 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
5754 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5755 (mcore_hard_regno_mode_ok): New function.
5756 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
5757 (HARD_REGNO_MODE_OK): Delete.
5758 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
5760 (microblaze_hard_regno_mode_ok_p): ...this and make static.
5761 (microblaze_hard_regno_mode_ok): New function.
5762 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5763 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
5764 (mips_hard_regno_mode_ok): Delete.
5765 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
5766 (mips_hard_regno_mode_ok_p): ...this and make static.
5767 (mips_hard_regno_mode_ok_p): Rename to...
5768 (mips_hard_regno_mode_ok_uncached): ...this.
5769 (mips_hard_regno_mode_ok): New function.
5770 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
5771 of HARD_REGNO_MODE_OK.
5772 (mips_option_override): Update after above name changes.
5773 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5774 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
5775 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
5776 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
5777 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
5778 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5779 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
5780 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
5781 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
5782 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5783 (msp430_hard_regno_mode_ok): Make static and return a bool.
5784 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
5785 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
5786 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
5788 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5789 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
5790 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
5791 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
5792 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
5793 (PA_HARD_REGNO_MODE_OK): ...this
5794 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
5795 (PA_HARD_REGNO_MODE_OK): ...this.
5796 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5797 (pa_hard_regno_mode_ok): New function.
5798 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
5799 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5800 (pdp11_hard_regno_mode_ok): New function.
5801 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
5802 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
5804 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
5806 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5807 (rs6000_hard_regno_mode_ok): Rename to...
5808 (rs6000_hard_regno_mode_ok_uncached): ...this.
5809 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5810 (rs6000_hard_regno_mode_ok): New function.
5811 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
5812 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
5813 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
5814 (riscv_hard_regno_mode_ok): ...this and make static.
5815 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5816 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
5817 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
5818 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5819 (rl78_hard_regno_mode_ok): Make static and return bool.
5820 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
5821 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
5823 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
5824 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5825 (rs6000_hard_regno_mode_ok): Rename to...
5826 (rs6000_hard_regno_mode_ok_uncached): ...this.
5827 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5828 (rs6000_hard_regno_mode_ok): New function.
5829 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
5830 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
5831 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5832 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
5833 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
5834 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
5835 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5836 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
5837 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
5838 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5839 (sh_hard_regno_mode_ok): Make static.
5840 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
5841 instead of HARD_REGNO_MODE_OK.
5842 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
5843 (sparc_mode_class): Delete.
5844 (HARD_REGNO_MODE_OK): Delete.
5845 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5846 (hard_regno_mode_classes): Make static.
5847 (sparc_mode_class): Likewise.
5848 (sparc_hard_regno_mode_ok): New function.
5849 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
5850 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
5851 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
5853 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5854 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
5855 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
5856 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
5857 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
5858 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5859 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
5860 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
5861 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5862 (visium_hard_regno_mode_ok): New function.
5863 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
5864 instead of HARD_REGNO_MODE_OK.
5865 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
5866 (HARD_REGNO_MODE_OK): Delete.
5867 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
5868 (xtensa_hard_regno_mode_ok_p): ...this and make static.
5869 (xtensa_option_override): Update accordingly.
5870 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5871 (xtensa_hard_regno_mode_ok): New function.
5872 * system.h (HARD_REGNO_MODE_OK): Poison.
5874 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5875 Alan Hayward <alan.hayward@arm.com>
5876 David Sherwood <david.sherwood@arm.com>
5878 * target.def (hard_regno_call_part_clobbered): New hook.
5879 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
5880 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
5881 * doc/tm.texi: Regenerate.
5882 * hooks.h (hook_bool_uint_mode_false): Declare.
5883 * hooks.c (hook_bool_uint_mode_false): New function.
5884 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5885 * cselib.c (cselib_process_insn): Use
5886 targetm.hard_regno_call_part_clobbered instead of
5887 HARD_REGNO_CALL_PART_CLOBBERED.
5888 * ira-conflicts.c (ira_build_conflicts): Likewise.
5889 * ira-costs.c (ira_tune_allocno_costs): Likewise.
5890 * lra-constraints.c (need_for_call_save_p): Likewise.
5891 * lra-lives.c: Include target.h.
5892 (check_pseudos_live_through_calls): Use
5893 targetm.hard_regno_call_part_clobbered instead of
5894 HARD_REGNO_CALL_PART_CLOBBERED.
5895 * regcprop.c: Include target.h.
5896 (copyprop_hardreg_forward_1): Use
5897 targetm.hard_regno_call_part_clobbered instead of
5898 HARD_REGNO_CALL_PART_CLOBBERED.
5899 * reginfo.c (choose_hard_reg_mode): Likewise.
5900 * regrename.c (check_new_reg_p): Likewise.
5901 * reload.c (find_equiv_reg): Likewise.
5902 * reload1.c (emit_reload_insns): Likewise.
5903 * sched-deps.c (deps_analyze_insn): Likewise.
5904 * sel-sched.c (init_regs_for_mode): Likewise.
5905 (mark_unavailable_hard_regs): Likewise.
5906 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
5907 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5908 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
5910 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5911 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5912 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
5914 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
5916 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5917 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5918 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
5920 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5921 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5922 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
5924 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5925 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
5927 * config/powerpcspe/powerpcspe.c
5928 (rs6000_hard_regno_call_part_clobbered): New function.
5929 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5930 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5931 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
5933 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5934 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5935 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
5937 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5938 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5939 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
5941 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5942 Alan Hayward <alan.hayward@arm.com>
5943 David Sherwood <david.sherwood@arm.com>
5945 * rtl.h (subreg_memory_offset): Declare.
5946 * emit-rtl.c (subreg_memory_offset): New function.
5947 * expmed.c (store_bit_field_1): Use it.
5948 * expr.c (undefined_operand_subword_p): Likewise.
5949 * simplify-rtx.c (simplify_subreg): Likewise.
5951 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
5953 PR rtl-optimization/57448
5956 * optabs.c (expand_atomic_load): Place compiler memory barriers if
5957 using atomic_load pattern.
5958 (expand_atomic_store): Likewise.
5960 2017-09-04 Jakub Jelinek <jakub@redhat.com>
5963 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
5964 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
5965 handling. Use replace_call_with_value with NULL instead of
5966 gsi_replace, unlink_stmt_vdef and release_defs.
5968 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
5971 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
5973 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5976 * rtl.h (emit_library_call_value_1): Declare.
5977 (emit_library_call): Replace declaration with a series of overloads.
5978 Remove the parameter count argument.
5979 (emit_library_call_value): Likewise.
5980 * calls.c (emit_library_call_value_1): Make global. Replace varargs
5981 with an "rtx_mode_t *".
5982 (emit_library_call_value): Delete.
5983 (emit_library_call): Likewise.
5984 * asan.c (asan_emit_stack_protection): Update calls accordingly.
5985 (asan_emit_allocas_unpoison): Likewise.
5986 * builtins.c (expand_builtin_powi): Likewise.
5987 (expand_asan_emit_allocas_unpoison): Likewise.
5988 * cfgexpand.c (expand_main_function): Likewise.
5989 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
5990 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
5991 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
5992 * config/arm/arm.c (arm_trampoline_init): Likewise.
5993 (arm_call_tls_get_addr): Likewise.
5994 (arm_expand_divmod_libfunc): Likewise.
5995 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
5996 (smulsi3_highpart): Likewise.
5997 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
5998 (c6x_expand_compare): Likewise.
5999 (c6x_expand_movmem): Likewise.
6000 * config/frv/frv.c (frv_trampoline_init): Likewise.
6001 * config/i386/i386.c (ix86_trampoline_init): Likewise.
6002 (ix86_expand_divmod_libfunc): Likewise.
6003 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
6004 (ia64_expand_compare): Likewise.
6005 (ia64_profile_hook): Likewise.
6006 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
6007 (nonlocal_goto): Likewise.
6008 (restore_stack_nonlocal): Likewise.
6009 * config/m32r/m32r.c (block_move_call): Likewise.
6010 (m32r_trampoline_init): Likewise.
6011 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
6012 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
6013 (m68k_call_m68k_read_tp): Likewise.
6014 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
6015 (microblaze_expand_divide): Likewise.
6016 * config/mips/mips.h (mips_args): Likewise.
6017 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
6018 (MIPS_ICACHE_SYNC): Likewise.
6019 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
6020 (nios2_trampoline_init): Likewise.
6021 * config/pa/pa.c (hppa_tls_call): Likewise.
6022 (pa_trampoline_init): Likewise.
6023 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
6024 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
6025 (expand_strn_compare): Likewise.
6026 (rs6000_generate_compare): Likewise.
6027 (rs6000_expand_float128_convert): Likewise.
6028 (output_profile_hook): Likewise.
6029 (rs6000_trampoline_init): Likewise.
6030 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
6031 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
6032 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
6033 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
6034 (rs6000_generate_compare): Likewise.
6035 (rs6000_expand_float128_convert): Likewise.
6036 (output_profile_hook): Likewise.
6037 (rs6000_trampoline_init): Likewise.
6038 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
6039 * config/sh/sh.c (sh_trampoline_init): Likewise.
6040 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
6041 (sparc_emit_float_lib_cmp): Likewise.
6042 (sparc32_initialize_trampoline): Likewise.
6043 (sparc64_initialize_trampoline): Likewise.
6044 (sparc_profile_hook): Likewise.
6045 * config/spu/spu.c (ea_load_store): Likewise.
6046 * config/spu/spu.md (floatunssidf2): Likewise.
6047 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
6048 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
6049 * config/visium/visium.c (expand_block_move_4): Likewise.
6050 (expand_block_move_2): Likewise.
6051 (expand_block_move_1): Likewise.
6052 (expand_block_set_4): Likewise.
6053 (expand_block_set_2): Likewise.
6054 (expand_block_set_1): Likewise.
6055 (visium_trampoline_init): Likewise.
6056 (visium_profile_hook): Likewise.
6057 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
6058 (xtensa_setup_frame_addresses): Likewise.
6059 (xtensa_trampoline_init): Likewise.
6060 * except.c (sjlj_emit_function_enter): Likewise.
6061 (sjlj_emit_function_exit): Likewise.
6062 * explow.c (allocate_dynamic_stack_space): Likewise.
6063 (probe_stack_range): Likewise.
6064 * expr.c (convert_mode_scalar): Likewise.
6065 * optabs.c (expand_binop): Likewise.
6066 (expand_twoval_binop_libfunc): Likewise.
6067 (expand_unop): Likewise.
6068 (prepare_cmp_insn): Likewise.
6069 (prepare_float_lib_cmp): Likewise.
6070 (expand_float): Likewise.
6071 (expand_fix): Likewise.
6072 (expand_fixed_convert): Likewise.
6073 (maybe_emit_sync_lock_test_and_set): Likewise.
6074 (expand_atomic_compare_and_swap): Likewise.
6075 (expand_mem_thread_fence): Likewise.
6076 (expand_atomic_fetch_op): Likewise.
6078 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
6080 * doc/generic.texi (OpenACC): Adjust URL.
6081 * doc/invoke.texi (C Dialect Options): Ditto.
6083 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
6085 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
6086 predicate for operand 1. Add (m,<S>) constraint.
6087 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
6088 Prevent memory operand 1 with register operand 2.
6090 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
6092 PR rtl-optimization/82024
6093 * combine.c (try_combine): If the combination result is a PARALLEL,
6094 and we only need to retain the SET in there that would be placed
6095 at I2, check that we can place that at I3 instead, before doing so.
6097 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6100 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
6102 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
6103 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
6106 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
6107 Jeff Law <law@redhat.com>
6109 * varasm.c (bss_initializer_p): Do not put constants into .bss
6110 (categorize_decl_for_section): Handle bss_initializer_p returning
6111 false when DECL_INITIAL is NULL.
6113 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6116 * config/s390/s390.c (s390_can_inline_p): New function.
6118 2017-09-01 Jeff Law <law@redhat.com>
6120 PR tree-optimization/82052
6121 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
6122 Always initialize the returned slot after a hash table miss
6123 when INSERT is true.
6125 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
6127 * config/s390/s390.md (mem_signal_fence): Remove.
6128 * doc/md.texi (mem_signal_fence): Remove.
6129 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
6131 * target-insns.def (mem_signal_fence): Remove.
6133 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6136 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
6139 * asan.c (create_odr_indicator): Strip name encoding from assembler
6140 name before appending it after __odr_asan_.
6142 2017-09-01 Martin Liska <mliska@suse.cz>
6144 PR tree-optimization/82059
6145 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
6146 frequency only when an edge is redirected.
6148 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6150 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
6151 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
6152 (arc_conditional_register_usage): Remove ARC600 lp_count
6154 (arc_file_start): Emit Tag_ARC_CPU_variation.
6155 (arc_can_use_doloop_p): New conditions to use ZOLs.
6156 (hwloop_fail): New function.
6157 (hwloop_optimize): Likewise.
6158 (hwloop_pattern_reg): Likewise.
6159 (arc_doloop_hooks): New struct, to be used with reorg_loops.
6160 (arc_reorg_loops): New function, calls reorg_loops.
6161 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
6162 (arc600_corereg_hazard): Remove ZOL checking, case handled by
6164 (arc_loop_hazard): Remove function, functionality moved into
6166 (arc_hazard): Remove arc_loop_hazard call.
6167 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
6168 into hwloop_optimize.
6169 (arc_label_align): Remove ZOL handling.
6170 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
6171 * config/arc/arc.md (doloop_begin): Remove pattern.
6172 (doloop_begin_i): Likewise.
6173 (doloop_end_i): Likewise.
6174 (doloop_fallback): Likewise.
6175 (doloop_fallback_m): Likewise.
6176 (doloop_end): Reimplement expand.
6177 (arc_lp): New pattern for LP instruction.
6178 (loop_end): New pattern.
6179 (loop_fail): Likewise.
6180 (decrement_and_branch_until_zero): Likewise.
6181 * config/arc/arc.opt (mlpc-width): New option.
6182 * doc/invoke.texi (mlpc-width): Document option.
6184 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6186 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
6187 (arc_ccfsm_advance): Fix checking for delay slots.
6188 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
6190 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6192 * config/arc/arc.md (movqi_insn): Add stores to save constant long
6194 (movhi_insn): Update store instruction constraint which are saving
6195 6-bit short immediates.
6196 (movsi_insn): Consider also short scaled load operations.
6197 (zero_extendhisi2_i): Use Usd constraint instead of T.
6198 (extendhisi2_i): Add q constraint.
6199 (arc_clzsi2): Add type and length attributes.
6200 (arc_ctzsi2): Likewise.
6201 * config/arc/constraints.md (Usc): Update constraint, the
6202 assembler can parse two relocations for a single instruction.
6204 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6206 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
6207 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
6209 2017-08-31 Olivier Hainque <hainque@adacore.com>
6211 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
6212 match as powerpc-wrs-vxworks*.
6214 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
6216 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
6217 register constraint for by-element operand.
6218 (aarch64_mls_elt_merge<mode>): Likewise.
6220 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6222 * config/arc/arc.c (arc_can_follow_jump): Check for short
6225 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6227 * config.gcc: Use g.opt for arc.
6228 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
6229 functionality moved to ...
6230 (legitimate_scaled_address_p): New function, ...here.
6231 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
6232 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
6233 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
6235 (arc_override_options): Handle G option.
6236 (arc_output_pic_addr_const): Correct function definition.
6237 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
6238 (arc_decl_anon_ns_mem_p): Delete.
6239 (arc_in_small_data_p): Overhaul this function to take into
6240 consideration the value given via G option.
6241 (arc_rewrite_small_data_1): Renamed and corrected old
6242 arc_rewrite_small_data function.
6243 (arc_rewrite_small_data): New function.
6244 (small_data_pattern): Don't use pic_offset_table_rtx.
6245 * config/arc/arc.h (CC1_SPEC): Recognize G option.
6246 * config/arc/simdext.md (movmisalignv2hi): Use
6247 prepare_move_operands function.
6249 (movmisalign*): Likewise.
6250 * doc/invoke.texi (ARC options): Document -G option.
6252 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6254 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
6256 * config/arc/arc.c (arc_print_operand): Output scalled address for
6257 sdata whenever is possible.
6258 (arc_in_small_data_p): Allow sdata for 64bit datum when double
6259 load/stores are available.
6260 (compact_sda_memory_operand): Check for the alignment required by
6261 code density instructions.
6262 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
6264 * config/arc/constraints.md (Usd): Update constraint.
6265 (Us0): New constraint.
6266 (Usc): Update constraint.
6268 2017-08-31 Richard Biener <rguenther@suse.de>
6271 * dwarf2out.c (dwarf2out_early_global_decl): Process each
6274 2017-08-31 Tamar Christina <tamar.christina@arm.com>
6276 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
6277 Resize type_signature.
6279 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6280 Alan Hayward <alan.hayward@arm.com>
6281 David Sherwood <david.sherwood@arm.com>
6283 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
6284 subregs whose inner modes can be stored in GPRs.
6285 (aarch64_classify_index): Likewise.
6287 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6288 Alan Hayward <alan.hayward@arm.com>
6289 David Sherwood <david.sherwood@arm.com>
6291 * config/aarch64/iterators.md (V_cmp_result): Rename to...
6292 (V_INT_EQUIV): ...this.
6293 (v_cmp_result): Rename to...
6294 (v_int_equiv): ...this.
6295 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
6296 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
6297 (copysign<mode>3): Likewise.
6298 (aarch64_simd_bsl<mode>_internal): Likewise.
6299 (aarch64_simd_bsl<mode>): Likewise.
6300 (vec_cmp<mode><mode>): Likewise.
6301 (vcond<mode><mode>): Likewise.
6302 (vcond<v_cmp_mixed><mode>): Likewise.
6303 (vcondu<mode><v_cmp_mixed>): Likewise.
6304 (aarch64_cm<optab><mode>): Likewise.
6305 (aarch64_cmtst<mode>): Likewise.
6306 (aarch64_fac<optab><mode>): Likewise.
6307 (vec_perm_const<mode>): Likewise.
6308 (vcond_mask_<mode><v_cmp_result>): Rename to...
6309 (vcond_mask_<mode><v_int_equiv>): ...this.
6310 (vec_cmp<mode><v_cmp_result>): Rename to...
6311 (vec_cmp<mode><v_int_equiv>): ...this.
6313 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6314 Alan Hayward <alan.hayward@arm.com>
6315 David Sherwood <david.sherwood@arm.com>
6317 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
6319 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
6320 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
6321 (UNSPEC_LD4_DREG): New unspecs.
6322 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
6323 (aarch64_ld2<mode>_dreg_be): Replace with...
6324 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
6326 (aarch64_ld3<mode>_dreg_le)
6327 (aarch64_ld3<mode>_dreg_be): Replace with...
6328 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
6330 (aarch64_ld4<mode>_dreg_le)
6331 (aarch64_ld4<mode>_dreg_be): Replace with...
6332 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
6335 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6337 PR tree-optimization/81987
6338 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
6339 insert an initializer in a location not dominated by the stride
6342 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6344 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
6345 on the entire header of the finally block in the fallthru case.
6347 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6349 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
6351 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
6353 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
6354 rs6000_emit_move_from_cr and call renamed function.
6355 (rs6000_emit_prologue): Call renamed functions.
6356 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
6357 movesi_from_cr, remove volatile CRs.
6359 2017-08-30 Jon Beniston <jon@beniston.com>
6360 Richard Biener <rguenther@suse.de>
6362 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
6363 of VECTOR_MODE_P check.
6364 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
6365 element vector types.
6367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6369 * df.h (df_read_modify_subreg_p): Remove in favor of...
6370 * rtl.h (read_modify_subreg_p): ...this new function. Take a
6371 const_rtx instead of an rtx.
6372 * cprop.c (local_cprop_find_used_regs): Update accordingly.
6373 * df-problems.c (df_word_lr_mark_ref): Likewise.
6374 * ira-lives.c (mark_pseudo_reg_live): Likewise.
6375 (mark_pseudo_reg_dead): Likewise.
6376 (mark_ref_dead): Likewise.
6377 * reginfo.c (init_subregs_of_mode): Likewise.
6378 * sched-deps.c (sched_analyze_1): Likewise.
6379 * df-scan.c (df_def_record_1): Likewise.
6380 (df_uses_record): Likewise.
6381 (df_read_modify_subreg_p): Remove in favor of...
6382 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
6383 const_rtx instead of an rtx.
6385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6386 Alan Hayward <alan.hayward@arm.com>
6387 David Sherwood <david.sherwood@arm.com>
6389 * rtl.h (partial_subreg_p): New function.
6390 * caller-save.c (save_call_clobbered_regs): Use it.
6391 * calls.c (expand_call): Likewise.
6392 * combine.c (combinable_i3pat): Likewise.
6393 (simplify_set): Likewise.
6394 (make_extraction): Likewise.
6395 (make_compound_operation_int): Likewise.
6396 (gen_lowpart_or_truncate): Likewise.
6397 (force_to_mode): Likewise.
6398 (make_field_assignment): Likewise.
6399 (reg_truncated_to_mode): Likewise.
6400 (record_truncated_value): Likewise.
6401 (move_deaths): Likewise.
6402 * cse.c (record_jump_cond): Likewise.
6403 (cse_insn): Likewise.
6404 * cselib.c (cselib_lookup_1): Likewise.
6405 * expmed.c (extract_bit_field_using_extv): Likewise.
6406 * function.c (assign_parm_setup_reg): Likewise.
6407 * ifcvt.c (noce_convert_multiple_sets): Likewise.
6408 * ira-build.c (create_insn_allocnos): Likewise.
6409 * lra-coalesce.c (merge_pseudos): Likewise.
6410 * lra-constraints.c (match_reload): Likewise.
6411 (simplify_operand_subreg): Likewise.
6412 (curr_insn_transform): Likewise.
6413 * lra-lives.c (process_bb_lives): Likewise.
6414 * lra.c (new_insn_reg): Likewise.
6415 (lra_substitute_pseudo): Likewise.
6416 * regcprop.c (mode_change_ok): Likewise.
6417 (maybe_mode_change): Likewise.
6418 (copyprop_hardreg_forward_1): Likewise.
6419 * reload.c (push_reload): Likewise.
6420 (find_reloads): Likewise.
6421 (find_reloads_subreg_address): Likewise.
6422 * reload1.c (alter_reg): Likewise.
6423 (eliminate_regs_1): Likewise.
6424 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6426 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
6428 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
6429 back to if statements, including unpack.
6431 2017-08-30 Martin Liska <mliska@suse.cz>
6434 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
6436 (func_checker::compare_asm_inputs_outputs): ... this function.
6437 (func_checker::compare_gimple_asm): Use the function to compare
6438 also ASM constrains.
6439 * ipa-icf-gimple.h: Rename the function.
6441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6442 Alan Hayward <alan.hayward@arm.com>
6443 David Sherwood <david.sherwood@arm.com>
6445 * coretypes.h (complex_mode): New type.
6446 * gdbhooks.py (build_pretty_printer): Handle it.
6447 * machmode.h (complex_mode): New class.
6448 (complex_mode::includes_p): New function.
6449 (is_complex_int_mode): Likewise.
6450 (is_complex_float_mode): Likewise.
6451 * genmodes.c (get_mode_class): Handle complex mode classes.
6452 * function.c (expand_function_end): Use is_complex_int_mode.
6454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6455 Alan Hayward <alan.hayward@arm.com>
6456 David Sherwood <david.sherwood@arm.com>
6458 * coretypes.h (scalar_mode_pod): New typedef.
6459 * gdbhooks.py (build_pretty_printer): Handle it.
6460 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
6461 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
6462 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
6463 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
6466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6467 Alan Hayward <alan.hayward@arm.com>
6468 David Sherwood <david.sherwood@arm.com>
6470 * machmode.h (mode_for_vector): Take a scalar_mode instead
6472 * stor-layout.c (mode_for_vector): Likewise.
6473 * explow.c (promote_mode): Use as_a <scalar_mode>.
6474 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
6476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6477 Alan Hayward <alan.hayward@arm.com>
6478 David Sherwood <david.sherwood@arm.com>
6480 * target.def (preferred_simd_mode): Take a scalar_mode
6481 instead of a machine_mode.
6482 * targhooks.h (default_preferred_simd_mode): Likewise.
6483 * targhooks.c (default_preferred_simd_mode): Likewise.
6484 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
6485 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
6486 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
6487 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
6488 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
6489 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
6490 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
6491 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
6493 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
6494 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
6495 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
6496 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
6497 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
6498 * doc/tm.texi: Regenerate.
6499 * optabs-query.c (can_vec_mask_load_store_p): Return false for
6502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6503 Alan Hayward <alan.hayward@arm.com>
6504 David Sherwood <david.sherwood@arm.com>
6506 * target.def (scalar_mode_supported_p): Take a scalar_mode
6507 instead of a machine_mode.
6508 * targhooks.h (default_scalar_mode_supported_p): Likewise.
6509 * targhooks.c (default_scalar_mode_supported_p): Likewise.
6510 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
6511 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6512 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
6513 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
6514 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
6515 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
6516 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
6517 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
6518 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
6519 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
6520 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
6521 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
6523 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
6524 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
6525 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6526 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6527 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
6529 * doc/tm.texi: Regenerate.
6531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6532 Alan Hayward <alan.hayward@arm.com>
6533 David Sherwood <david.sherwood@arm.com>
6535 * coretypes.h (opt_scalar_mode): New typedef.
6536 * gdbhooks.py (build_pretty_printers): Handle it.
6537 * machmode.h (mode_iterator::get_2xwider): Add overload for
6539 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
6541 * expr.c (convert_mode_scalar): Likewise.
6542 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6543 * optabs.c (expand_float): Likewise.
6544 (expand_fix): Likewise.
6545 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6548 Alan Hayward <alan.hayward@arm.com>
6549 David Sherwood <david.sherwood@arm.com>
6551 * optabs.c (expand_float): Explicitly check for scalars before
6552 using a branching expansion.
6553 (expand_fix): Likewise.
6555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6556 Alan Hayward <alan.hayward@arm.com>
6557 David Sherwood <david.sherwood@arm.com>
6559 * expr.c (convert_mode): Split scalar handling out into...
6560 (convert_mode_scalar): ...this new function. Treat the modes
6563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6564 Alan Hayward <alan.hayward@arm.com>
6565 David Sherwood <david.sherwood@arm.com>
6567 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
6569 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
6571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6572 Alan Hayward <alan.hayward@arm.com>
6573 David Sherwood <david.sherwood@arm.com>
6575 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
6576 rather than a machine_mode.
6577 (fixed_from_string): Likewise.
6578 (fixed_convert): Likewise.
6579 (fixed_convert_from_int): Likewise.
6580 (fixed_convert_from_real): Likewise.
6581 (real_convert_from_fixed): Likewise.
6582 * fixed-value.c (fixed_from_double_int): Likewise.
6583 (fixed_from_string): Likewise.
6584 (fixed_convert): Likewise.
6585 (fixed_convert_from_int): Likewise.
6586 (fixed_convert_from_real): Likewise.
6587 (real_convert_from_fixed): Likewise.
6588 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
6590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6591 Alan Hayward <alan.hayward@arm.com>
6592 David Sherwood <david.sherwood@arm.com>
6594 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
6595 of separate mode class checks. Do not allow vector modes here.
6596 (immed_wide_int_const): Use as_a <scalar_mode>.
6597 * explow.c (trunc_int_for_mode): Likewise.
6598 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
6599 (wi::shwi): Likewise.
6600 (wi::min_value): Likewise.
6601 (wi::max_value): Likewise.
6602 * dwarf2out.c (loc_descriptor): Likewise.
6603 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
6606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6607 Alan Hayward <alan.hayward@arm.com>
6608 David Sherwood <david.sherwood@arm.com>
6610 * tree.h (SCALAR_TYPE_MODE): New macro.
6611 * expr.c (expand_expr_addr_expr_1): Use it.
6612 (expand_expr_real_2): Likewise.
6613 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
6614 (fold_convert_const_fixed_from_int): Likewise.
6615 (fold_convert_const_fixed_from_real): Likewise.
6616 (native_encode_fixed): Likewise
6617 (native_encode_complex): Likewise
6618 (native_encode_vector): Likewise.
6619 (native_interpret_fixed): Likewise.
6620 (native_interpret_real): Likewise.
6621 (native_interpret_complex): Likewise.
6622 (native_interpret_vector): Likewise.
6623 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
6624 (simd_clone_adjust_argument_types): Likewise.
6625 (simd_clone_init_simd_arrays): Likewise.
6626 (simd_clone_adjust): Likewise.
6627 * stor-layout.c (layout_type): Likewise.
6628 * tree.c (build_minus_one_cst): Likewise.
6629 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6630 * tree-inline.c (estimate_move_cost): Likewise.
6631 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
6632 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
6633 (vectorizable_reduction): Likewise.
6634 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
6635 (vect_recog_mixed_size_cond_pattern): Likewise.
6636 (check_bool_pattern): Likewise.
6637 (adjust_bool_pattern): Likewise.
6638 (search_type_for_mask_1): Likewise.
6639 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
6640 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6641 (vectorizable_load): Likewise.
6642 (vectorizable_store): Likewise.
6643 * ubsan.c (ubsan_encode_value): Likewise.
6644 * varasm.c (output_constant): Likewise.
6646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6647 Alan Hayward <alan.hayward@arm.com>
6648 David Sherwood <david.sherwood@arm.com>
6650 * coretypes.h (scalar_mode): New class.
6651 * machmode.h (scalar_mode): Likewise.
6652 (scalar_mode::includes_p): New function.
6653 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
6654 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
6655 * genmodes.c (get_mode_class): Handle remaining scalar modes.
6656 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
6657 * expmed.c (store_bit_field_1): Likewise.
6658 (extract_bit_field_1): Likewise.
6659 * expr.c (write_complex_part): Likewise.
6660 (read_complex_part): Likewise.
6661 (emit_move_complex_push): Likewise.
6662 (expand_expr_real_2): Likewise.
6663 * function.c (assign_parm_setup_reg): Likewise.
6664 (assign_parms_unsplit_complex): Likewise.
6665 * optabs.c (expand_binop): Likewise.
6666 * rtlanal.c (subreg_get_info): Likewise.
6667 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6668 * varasm.c (output_constant_pool_2): Likewise.
6670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6671 Alan Hayward <alan.hayward@arm.com>
6672 David Sherwood <david.sherwood@arm.com>
6674 * expmed.c (extract_high_half): Use scalar_int_mode and remove
6676 (expmed_mult_highpart_optab): Likewise.
6677 (expmed_mult_highpart): Likewise.
6679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6680 Alan Hayward <alan.hayward@arm.com>
6681 David Sherwood <david.sherwood@arm.com>
6683 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
6684 instead of a machine_mode.
6685 (builtin_memset_read_str): Likewise.
6686 * builtins.c (c_readstr): Likewise.
6687 (builtin_memcpy_read_str): Likewise.
6688 (builtin_strncpy_read_str): Likewise.
6689 (builtin_memset_read_str): Likewise.
6690 (builtin_memset_gen_str): Likewise.
6691 (expand_builtin_signbit): Use scalar_int_mode for local variables.
6692 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
6693 instead of a machine_mode.
6694 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
6696 (make_extraction): Likewise.
6697 (try_widen_shift_mode): Take and return scalar_int_modes instead
6699 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
6700 a scalar_int_mode instead of a machine_mode.
6701 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
6702 (avr_addr_space_pointer_mode): Likewise.
6703 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
6704 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
6705 (msp430_unwind_word_mode): Likewise.
6706 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
6707 (spu_addr_space_pointer_mode): Likewise.
6708 (spu_addr_space_address_mode): Likewise.
6709 (spu_libgcc_cmp_return_mode): Likewise.
6710 (spu_libgcc_shift_count_mode): Likewise.
6711 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
6712 (rl78_addr_space_pointer_mode): Likewise.
6713 (fl78_unwind_word_mode): Likewise.
6714 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
6716 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
6717 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
6718 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
6719 (mips_valid_pointer_mode): Likewise.
6720 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
6721 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
6722 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
6724 (ft32_addr_space_address_mode): Likewise.
6725 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
6726 scalar_int_mode instead of a machine_mode.
6727 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
6729 (m32c_addr_space_address_mode): Likewise.
6730 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
6731 (rs6000_eh_return_filter_mode): Likewise.
6732 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
6733 (rs6000_eh_return_filter_mode): Likewise.
6734 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
6735 (s390_libgcc_shift_count_mode): Likewise.
6736 (s390_unwind_word_mode): Likewise.
6737 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
6739 * target.def (mode_rep_extended): Likewise.
6740 (valid_pointer_mode): Likewise.
6741 (addr_space.valid_pointer_mode): Likewise.
6742 (eh_return_filter_mode): Return a scalar_int_mode rather than
6744 (libgcc_cmp_return_mode): Likewise.
6745 (libgcc_shift_count_mode): Likewise.
6746 (unwind_word_mode): Likewise.
6747 (addr_space.pointer_mode): Likewise.
6748 (addr_space.address_mode): Likewise.
6749 * doc/tm.texi: Regenerate.
6750 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
6752 (do_jump): Use scalar_int_mode for local variables.
6753 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
6754 rather than a machine_mode.
6755 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
6756 (scompare_loc_descriptor_wide): Likewise.
6757 (scompare_loc_descriptor_narrow): Likewise.
6758 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
6760 * except.c (sjlj_emit_dispatch_table): Likewise.
6761 (expand_builtin_eh_copy_values): Likewise.
6762 * explow.c (convert_memory_address_addr_space_1): Likewise.
6763 Take a scalar_int_mode rather than a machine_mode.
6764 (convert_memory_address_addr_space): Take a scalar_int_mode rather
6765 than a machine_mode.
6766 (memory_address_addr_space): Use scalar_int_mode for local variables.
6767 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
6768 rather than a machine_mode.
6769 * expmed.c (mask_rtx): Likewise.
6770 (init_expmed_one_conv): Likewise.
6771 (expand_mult_highpart_adjust): Likewise.
6772 (extract_high_half): Likewise.
6773 (expmed_mult_highpart_optab): Likewise.
6774 (expmed_mult_highpart): Likewise.
6775 (expand_smod_pow2): Likewise.
6776 (expand_sdiv_pow2): Likewise.
6777 (emit_store_flag_int): Likewise.
6778 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
6780 (extract_low_bits): Likewise.
6781 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
6783 * expr.c (pieces_addr::adjust): Likewise.
6784 (can_store_by_pieces): Likewise.
6785 (store_by_pieces): Likewise.
6786 (clear_by_pieces_1): Likewise.
6787 (expand_expr_addr_expr_1): Likewise.
6788 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
6789 (expand_expr_real_1): Likewise.
6790 (try_casesi): Likewise.
6791 * final.c (shorten_branches): Likewise.
6792 * fold-const.c (fold_convert_const_int_from_fixed): Change the
6793 type of "mode" to machine_mode.
6794 * internal-fn.c (expand_arith_overflow_result_store): Take a
6795 scalar_int_mode rather than a machine_mode.
6796 (expand_mul_overflow): Use scalar_int_mode for local variables.
6797 * loop-doloop.c (doloop_modify): Likewise.
6798 (doloop_optimize): Likewise.
6799 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
6800 than a machine_mode.
6801 (expand_doubleword_shift_condmove): Likewise.
6802 (expand_doubleword_shift): Likewise.
6803 (expand_doubleword_clz): Likewise.
6804 (expand_doubleword_popcount): Likewise.
6805 (expand_doubleword_parity): Likewise.
6806 (expand_absneg_bit): Use scalar_int_mode for local variables.
6807 (prepare_float_lib_cmp): Likewise.
6808 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
6809 rather than a machine_mode.
6810 (convert_memory_address_addr_space): Likewise.
6811 (get_mode_bounds): Likewise.
6812 (get_address_mode): Return a scalar_int_mode rather than a
6814 * rtlanal.c (get_address_mode): Likewise.
6815 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
6816 than a machine_mode.
6817 * targhooks.c (default_mode_rep_extended): Likewise.
6818 (default_valid_pointer_mode): Likewise.
6819 (default_addr_space_valid_pointer_mode): Likewise.
6820 (default_eh_return_filter_mode): Return a scalar_int_mode rather
6821 than a machine_mode.
6822 (default_libgcc_cmp_return_mode): Likewise.
6823 (default_libgcc_shift_count_mode): Likewise.
6824 (default_unwind_word_mode): Likewise.
6825 (default_addr_space_pointer_mode): Likewise.
6826 (default_addr_space_address_mode): Likewise.
6827 * targhooks.h (default_eh_return_filter_mode): Likewise.
6828 (default_libgcc_cmp_return_mode): Likewise.
6829 (default_libgcc_shift_count_mode): Likewise.
6830 (default_unwind_word_mode): Likewise.
6831 (default_addr_space_pointer_mode): Likewise.
6832 (default_addr_space_address_mode): Likewise.
6833 (default_mode_rep_extended): Take a scalar_int_mode rather than
6835 (default_valid_pointer_mode): Likewise.
6836 (default_addr_space_valid_pointer_mode): Likewise.
6837 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
6839 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
6840 rather than a machine_mode.
6841 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
6842 for local variables.
6843 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6844 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
6845 than a machine_mode.
6847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6848 Alan Hayward <alan.hayward@arm.com>
6849 David Sherwood <david.sherwood@arm.com>
6851 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
6852 the mode argument to scalar_int_mode.
6853 (do_jump_by_parts_zero_rtx): Likewise.
6854 (do_jump_by_parts_equality_rtx): Likewise.
6855 (do_jump_by_parts_greater): Take a mode argument.
6856 (do_jump_by_parts_equality): Likewise.
6857 (do_jump_1): Update calls accordingly.
6859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6860 Alan Hayward <alan.hayward@arm.com>
6861 David Sherwood <david.sherwood@arm.com>
6863 * is-a.h (safe_dyn_cast): New function.
6864 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
6865 (jump_table_for_label): Likewise.
6866 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
6867 instead of an rtx_insn *.
6868 (shorten_branches): Use dyn_cast instead of LABEL_P and
6869 JUMP_TABLE_DATA_P. Use jump_table_for_label and
6870 rtx_jump_table_data::get_data_mode.
6871 (final_scan_insn): Likewise.
6873 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6874 Alan Hayward <alan.hayward@arm.com>
6875 David Sherwood <david.sherwood@arm.com>
6877 * combine.c (try_combine): Use is_a <scalar_int_mode> when
6878 trying to combine a full-register integer set with a subreg
6881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6882 Alan Hayward <alan.hayward@arm.com>
6883 David Sherwood <david.sherwood@arm.com>
6885 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
6886 that is always either address_mode or pointer_mode.
6888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6889 Alan Hayward <alan.hayward@arm.com>
6890 David Sherwood <david.sherwood@arm.com>
6892 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
6893 when the two are known to be equal.
6895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6896 Alan Hayward <alan.hayward@arm.com>
6897 David Sherwood <david.sherwood@arm.com>
6899 * simplify-rtx.c (simplify_const_unary_operation): Use
6900 is_a <scalar_int_mode> instead of checking for a nonzero
6901 precision. Forcibly convert op_mode to a scalar_int_mode
6902 in that case. More clearly differentiate the operand and
6903 result modes and use the former when deciding what the value
6904 of a count-bits operation should be. Use is_int_mode instead
6905 of checking for a MODE_INT. Remove redundant check for whether
6906 this mode has a zero precision.
6908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6909 Alan Hayward <alan.hayward@arm.com>
6910 David Sherwood <david.sherwood@arm.com>
6912 * optabs.c (widen_leading): Change the type of the mode argument
6913 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
6914 (widen_bswap): Likewise.
6915 (expand_parity): Likewise.
6916 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
6917 (expand_ffs): Likewise.
6918 (epand_unop): Check for scalar integer modes before calling the
6921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6922 Alan Hayward <alan.hayward@arm.com>
6923 David Sherwood <david.sherwood@arm.com>
6925 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
6927 (expand_expr_real_1): Update call accordingly.
6929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6930 Alan Hayward <alan.hayward@arm.com>
6931 David Sherwood <david.sherwood@arm.com>
6933 * expmed.c (store_bit_field_using_insv): Add op0_mode and
6934 value_mode arguments. Use scalar_int_mode internally.
6935 (store_bit_field_1): Rename the new integer mode from imode
6936 to op0_mode and use it instead of GET_MODE (op0). Update calls
6937 to store_split_bit_field, store_bit_field_using_insv and
6938 store_fixed_bit_field.
6939 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
6940 Use scalar_int_mode internally. Use a bit count rather than a mode
6941 when calculating the largest bit size for get_best_mode.
6942 Update calls to store_split_bit_field and store_fixed_bit_field_1.
6943 (store_fixed_bit_field_1): Add mode and value_mode arguments.
6944 Remove assertion that OP0 has a scalar integer mode.
6945 (store_split_bit_field): Add op0_mode and value_mode arguments.
6946 Update calls to extract_fixed_bit_field.
6947 (extract_bit_field_using_extv): Add an op0_mode argument.
6948 Use scalar_int_mode internally.
6949 (extract_bit_field_1): Rename the new integer mode from imode to
6950 op0_mode and use it instead of GET_MODE (op0). Update calls to
6951 extract_split_bit_field, extract_bit_field_using_extv and
6952 extract_fixed_bit_field.
6953 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
6954 to extract_split_bit_field and extract_fixed_bit_field_1.
6955 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
6956 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
6958 (extract_split_bit_field): Add an op0_mode argument. Update call
6959 to extract_fixed_bit_field.
6961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6962 Alan Hayward <alan.hayward@arm.com>
6963 David Sherwood <david.sherwood@arm.com>
6965 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
6966 * explow.c (hard_function_value): Likewise.
6967 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
6968 convert_to_mode call outside the loop.
6969 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
6970 for the mode iterator. Require the mode specified by max_pieces
6972 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
6974 (copy_blkmode_to_reg): Likewise.
6975 (set_storage_via_setmem): Likewise.
6976 * optabs.c (prepare_cmp_insn): Likewise.
6977 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6978 * stor-layout.c (finish_bitfield_representative): Likewise.
6980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6981 Alan Hayward <alan.hayward@arm.com>
6982 David Sherwood <david.sherwood@arm.com>
6984 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
6985 * expr.c (convert_move): Use them.
6986 (convert_modes): Likewise.
6987 (store_expr_with_bounds): Likewise.
6989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6990 Alan Hayward <alan.hayward@arm.com>
6991 David Sherwood <david.sherwood@arm.com>
6993 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
6994 parameter for the mode of "x". Remove the "known_x", "known_mode"
6995 and "known_ret" arguments. Change the type of the mode argument
6997 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
6998 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
6999 (reg_num_sign_bit_copies_for_combine): Likewise.
7000 * rtlanal.c (nonzero_bits1): Likewise.
7001 (num_sign_bit_copies1): Likewise.
7002 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
7003 (reg_num_sign_bit_copies_general): Likewise.
7004 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
7005 (reg_nonzero_bits_general): Likewise.
7007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7008 Alan Hayward <alan.hayward@arm.com>
7009 David Sherwood <david.sherwood@arm.com>
7011 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
7012 than in subroutines. Return 1 for non-integer modes.
7013 (cached_num_sign_bit_copies): Change the type of the mode parameter
7015 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
7016 classes. Handle CONST_INT_P first and then check whether X also
7017 has a scalar integer mode. Check the same thing for inner registers
7018 of a SUBREG and for values that are being extended or truncated.
7020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7021 Alan Hayward <alan.hayward@arm.com>
7022 David Sherwood <david.sherwood@arm.com>
7024 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
7025 in subroutines. Return the mode mask for non-integer modes.
7026 (cached_nonzero_bits): Change the type of the mode parameter
7028 (nonzero_bits1): Likewise. Remove early exit for other mode
7029 classes. Handle CONST_INT_P first and then check whether X
7030 also has a scalar integer mode.
7032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7033 Alan Hayward <alan.hayward@arm.com>
7034 David Sherwood <david.sherwood@arm.com>
7036 * expr.c (widest_int_mode_for_size): Make the comment match the code.
7037 Return a scalar_int_mode and assert that the size is greater than
7039 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
7040 (op_by_pieces_d::op_by_pieces_d): Likewise.
7041 (op_by_pieces_d::run): Likewise.
7042 (can_store_by_pieces): Likewise.
7044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7045 Alan Hayward <alan.hayward@arm.com>
7046 David Sherwood <david.sherwood@arm.com>
7048 * combine.c (extract_left_shift): Add a mode argument and update
7050 (make_compound_operation_int): Change the type of the mode parameter
7051 to scalar_int_mode and update the call to extract_left_shift.
7053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7054 Alan Hayward <alan.hayward@arm.com>
7055 David Sherwood <david.sherwood@arm.com>
7057 * combine.c (simplify_and_const_int): Change the type of the mode
7058 parameter to scalar_int_mode.
7059 (simplify_and_const_int_1): Likewise. Update recursive call.
7061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7062 Alan Hayward <alan.hayward@arm.com>
7063 David Sherwood <david.sherwood@arm.com>
7065 * combine.c (simplify_compare_const): Check that the mode is a
7066 scalar_int_mode (rather than VOIDmode) before testing its
7068 (simplify_comparison): Move COMPARISON_P handling out of the
7069 loop and restrict the latter part of the loop to scalar_int_modes.
7070 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
7071 and when considering SUBREG_REGs. Use is_int_mode instead of
7072 checking GET_MODE_CLASS against MODE_INT.
7074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7075 Alan Hayward <alan.hayward@arm.com>
7076 David Sherwood <david.sherwood@arm.com>
7078 * combine.c (try_widen_shift_mode): Move check for equal modes to...
7079 (simplify_shift_const_1): ...here. Use scalar_int_mode for
7080 shift_unit_mode and for modes involved in scalar shifts.
7082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7083 Alan Hayward <alan.hayward@arm.com>
7084 David Sherwood <david.sherwood@arm.com>
7086 * combine.c (force_int_to_mode): New function, split out from...
7087 (force_to_mode): ...here. Keep xmode up-to-date and use it
7088 instead of GET_MODE (x).
7090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7091 Alan Hayward <alan.hayward@arm.com>
7092 David Sherwood <david.sherwood@arm.com>
7094 * optabs-query.h (extraction_insn::struct_mode): Change type to
7095 opt_scalar_int_mode and update comment.
7096 (extraction_insn::field_mode): Change type to scalar_int_mode.
7097 (extraction_insn::pos_mode): Likewise.
7098 * combine.c (make_extraction): Update accordingly.
7099 * optabs-query.c (get_traditional_extraction_insn): Likewise.
7100 (get_optab_extraction_insn): Likewise.
7101 * recog.c (simplify_while_replacing): Likewise.
7102 * expmed.c (narrow_bit_field_mem): Change the type of the mode
7103 parameter to opt_scalar_int_mode.
7105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7106 Alan Hayward <alan.hayward@arm.com>
7107 David Sherwood <david.sherwood@arm.com>
7109 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
7110 to a scalar_int_mode instead of a machine_mode.
7111 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
7112 (get_best_mode): Return a boolean and use a pointer argument to store
7113 the selected mode. Replace the limit mode parameter with a bit limit.
7114 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
7115 for the values returned by bit_field_mode_iterator::next_mode.
7116 (store_bit_field): Update call to get_best_mode.
7117 (store_fixed_bit_field): Likewise.
7118 (extract_fixed_bit_field): Likewise.
7119 * expr.c (optimize_bitfield_assignment_op): Likewise.
7120 * fold-const.c (optimize_bit_field_compare): Likewise.
7121 (fold_truth_andor_1): Likewise.
7122 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
7123 Update for new type of m_mode.
7124 (get_best_mode): As above.
7126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7127 Alan Hayward <alan.hayward@arm.com>
7128 David Sherwood <david.sherwood@arm.com>
7130 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
7131 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
7132 (store_bit_field): Check is_a <scalar_int_mode> before calling
7133 strict_volatile_bitfield_p.
7134 (extract_bit_field): Likewise.
7136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7137 Alan Hayward <alan.hayward@arm.com>
7138 David Sherwood <david.sherwood@arm.com>
7140 * target.def (cstore_mode): Return a scalar_int_mode.
7141 * doc/tm.texi: Regenerate.
7142 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
7143 * targhooks.h (default_cstore_mode): Likewise.
7144 * targhooks.c (default_cstore_mode): Likewise, using a forced
7146 * expmed.c (emit_cstore): Expect the target of the cstore to be
7149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7150 Alan Hayward <alan.hayward@arm.com>
7151 David Sherwood <david.sherwood@arm.com>
7153 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
7155 (niter_desc): Likewise mode.
7156 (iv_analyze): Add a mode parameter.
7158 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
7159 and change its type to scalar_int_mode.
7160 * loop-iv.c: Update commentary at head of file.
7161 (iv_constant): Pass the mode paraeter before the rtx it describes
7162 and change its type to scalar_int_mode. Remove VOIDmode handling.
7163 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
7164 (iv_extend): Likewise.
7165 (shorten_into_mode): Likewise.
7166 (iv_add): Use scalar_int_mode.
7167 (iv_mult): Likewise.
7168 (iv_shift): Likewise.
7169 (canonicalize_iv_subregs): Likewise.
7170 (get_biv_step_1): Pass the outer_mode parameter before the rtx
7171 it describes and change its mode to scalar_int_mode. Also change
7172 the type of the returned inner_mode to scalar_int_mode.
7173 (get_biv_step): Likewise, turning outer_mode from a pointer
7174 into a direct parameter. Update call to get_biv_step_1.
7175 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
7176 iv_constant and get_biv_step.
7177 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
7178 and change its type to scalar_int_mode. Don't initialise iv->mode
7179 to VOIDmode and remove later checks for its still being VOIDmode.
7180 Update calls to iv_analyze_op and iv_analyze_expr. Check
7181 is_a <scalar_int_mode> when changing the mode under consideration.
7182 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
7183 Update call to iv_analyze_expr.
7184 (iv_analyze_op): Add a mode parameter. Reject subregs whose
7185 inner register is not also a scalar_int_mode. Update call to
7187 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
7188 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
7189 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
7190 separate mode class checks. Update calls to iv_analyze. Remove
7191 fix-up of VOIDmodes after iv_analyze_biv.
7192 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
7193 don't have a scalar_int_mode. Update call to biv_p.
7195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7196 Alan Hayward <alan.hayward@arm.com>
7197 David Sherwood <david.sherwood@arm.com>
7199 * cfgexpand.c (convert_debug_memory_address): Use
7200 as_a <scalar_int_mode>.
7201 * combine.c (expand_compound_operation): Likewise.
7202 (make_extraction): Likewise.
7203 (change_zero_ext): Likewise.
7204 (simplify_comparison): Likewise.
7205 * cse.c (cse_insn): Likewise.
7206 * dwarf2out.c (minmax_loc_descriptor): Likewise.
7207 (mem_loc_descriptor): Likewise.
7208 (loc_descriptor): Likewise.
7209 * expmed.c (init_expmed_one_mode): Likewise.
7210 (synth_mult): Likewise.
7211 (emit_store_flag_1): Likewise.
7212 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
7213 of a comparison with size.
7214 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
7215 (reduce_to_bit_field_precision): Likewise.
7216 * function.c (expand_function_end): Likewise.
7217 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
7218 * loop-doloop.c (doloop_modify): Likewise.
7219 * optabs.c (expand_binop): Likewise.
7220 (expand_unop): Likewise.
7221 (expand_copysign_absneg): Likewise.
7222 (prepare_cmp_insn): Likewise.
7223 (maybe_legitimize_operand): Likewise.
7224 * recog.c (const_scalar_int_operand): Likewise.
7225 * rtlanal.c (get_address_mode): Likewise.
7226 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7227 (simplify_cond_clz_ctz): Likewise.
7228 * tree-nested.c (get_nl_goto_field): Likewise.
7229 * tree.c (build_vector_type_for_mode): Likewise.
7230 * var-tracking.c (use_narrower_mode): Likewise.
7232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7233 Alan Hayward <alan.hayward@arm.com>
7234 David Sherwood <david.sherwood@arm.com>
7236 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
7237 * builtins.c (expand_builtin_signbit): Use it.
7238 * cfgexpand.c (expand_debug_expr): Likewise.
7239 * dojump.c (do_jump): Likewise.
7240 (do_compare_and_jump): Likewise.
7241 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
7242 * expmed.c (make_tree): Likewise.
7243 * expr.c (expand_expr_real_2): Likewise.
7244 (expand_expr_real_1): Likewise.
7245 (try_casesi): Likewise.
7246 * fold-const-call.c (fold_const_call_ss): Likewise.
7247 * fold-const.c (unextend): Likewise.
7248 (extract_muldiv_1): Likewise.
7249 (fold_single_bit_test): Likewise.
7250 (native_encode_int): Likewise.
7251 (native_encode_string): Likewise.
7252 (native_interpret_int): Likewise.
7253 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
7254 * internal-fn.c (expand_addsub_overflow): Likewise.
7255 (expand_neg_overflow): Likewise.
7256 (expand_mul_overflow): Likewise.
7257 (expand_arith_overflow): Likewise.
7258 * match.pd: Likewise.
7259 * stor-layout.c (layout_type): Likewise.
7260 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7261 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
7262 * tree-ssanames.c (get_range_info): Likewise.
7263 * tree-switch-conversion.c (array_value_type) Likewise.
7264 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
7265 (vect_recog_divmod_pattern): Likewise.
7266 (vect_recog_mixed_size_cond_pattern): Likewise.
7267 * tree-vrp.c (extract_range_basic): Likewise.
7268 (simplify_float_conversion_using_ranges): Likewise.
7269 * tree.c (int_fits_type_p): Likewise.
7270 * ubsan.c (instrument_bool_enum_load): Likewise.
7271 * varasm.c (mergeable_string_section): Likewise.
7272 (narrowing_initializer_constant_valid_p): Likewise.
7273 (output_constant): Likewise.
7275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7276 Alan Hayward <alan.hayward@arm.com>
7277 David Sherwood <david.sherwood@arm.com>
7279 * machmode.h (NARROWEST_INT_MODE): New macro.
7280 * expr.c (alignment_for_piecewise_move): Use it instead of
7281 GET_CLASS_NARROWEST_MODE (MODE_INT).
7282 (push_block): Likewise.
7283 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
7285 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7288 Alan Hayward <alan.hayward@arm.com>
7289 David Sherwood <david.sherwood@arm.com>
7291 * postreload.c (move2add_valid_value_p): Change the type of the
7292 mode parameter to scalar_int_mode.
7293 (move2add_use_add2_insn): Add a mode parameter and use it instead
7295 (move2add_use_add3_insn): Likewise.
7296 (reload_cse_move2add): Update accordingly.
7298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7299 Alan Hayward <alan.hayward@arm.com>
7300 David Sherwood <david.sherwood@arm.com>
7302 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
7304 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
7305 * optabs.c (expand_unop): Likewise.
7307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7308 Alan Hayward <alan.hayward@arm.com>
7309 David Sherwood <david.sherwood@arm.com>
7311 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
7312 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
7313 (popcount_loc_descriptor): Likewise.
7314 (bswap_loc_descriptor): Likewise.
7315 (rotate_loc_descriptor): Likewise.
7316 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
7317 calling the functions above.
7319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7320 Alan Hayward <alan.hayward@arm.com>
7321 David Sherwood <david.sherwood@arm.com>
7323 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
7325 (try_combine): Likewise.
7326 (simplify_if_then_else): Likewise.
7327 * cse.c (cse_insn): Likewise.
7328 * dwarf2out.c (mem_loc_descriptor): Likewise.
7329 * emit-rtl.c (gen_lowpart_common): Likewise.
7330 * simplify-rtx.c (simplify_truncation): Likewise.
7331 (simplify_binary_operation_1): Likewise.
7332 (simplify_const_relational_operation): Likewise.
7333 (simplify_ternary_operation): Likewise.
7334 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
7336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7337 Alan Hayward <alan.hayward@arm.com>
7338 David Sherwood <david.sherwood@arm.com>
7340 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
7341 * reload.c (push_reload): Likewise.
7342 (find_reloads): Likewise.
7344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7345 Alan Hayward <alan.hayward@arm.com>
7346 David Sherwood <david.sherwood@arm.com>
7348 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
7349 (make_compound_operation_int): Likewise.
7350 (change_zero_ext): Likewise.
7351 * expr.c (convert_move): Likewise.
7352 (convert_modes): Likewise.
7353 * fwprop.c (forward_propagate_subreg): Likewise.
7354 * loop-iv.c (get_biv_step_1): Likewise.
7355 * optabs.c (widen_operand): Likewise.
7356 * postreload.c (move2add_valid_value_p): Likewise.
7357 * recog.c (simplify_while_replacing): Likewise.
7358 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7359 (simplify_binary_operation_1): Likewise. Remove redundant
7360 mode equality check.
7362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7363 Alan Hayward <alan.hayward@arm.com>
7364 David Sherwood <david.sherwood@arm.com>
7366 * combine.c (combine_simplify_rtx): Add checks for
7367 is_a <scalar_int_mode>.
7368 (simplify_if_then_else): Likewise.
7369 (make_field_assignment): Likewise.
7370 (simplify_comparison): Likewise.
7371 * ifcvt.c (noce_try_bitop): Likewise.
7372 * loop-invariant.c (canonicalize_address_mult): Likewise.
7373 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7376 Alan Hayward <alan.hayward@arm.com>
7377 David Sherwood <david.sherwood@arm.com>
7379 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
7380 is_a <scalar_int_mode> instead of != BLKmode.
7382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7383 Alan Hayward <alan.hayward@arm.com>
7384 David Sherwood <david.sherwood@arm.com>
7386 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
7387 instead of != VOIDmode.
7388 * combine.c (if_then_else_cond): Likewise.
7389 (change_zero_ext): Likewise.
7390 * dwarf2out.c (mem_loc_descriptor): Likewise.
7391 (loc_descriptor): Likewise.
7392 * rtlanal.c (canonicalize_condition): Likewise.
7393 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
7395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7396 Alan Hayward <alan.hayward@arm.com>
7397 David Sherwood <david.sherwood@arm.com>
7399 * simplify-rtx.c (simplify_binary_operation_1): Use
7400 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
7402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7403 Alan Hayward <alan.hayward@arm.com>
7404 David Sherwood <david.sherwood@arm.com>
7406 * wide-int.h (int_traits<unsigned char>) New class.
7407 (int_traits<unsigned short>) Likewise.
7408 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
7409 Use GET_MODE_UNIT_PRECISION and remove redundant test for
7411 * combine.c (set_nonzero_bits_and_sign_copies): Use
7412 is_a <scalar_int_mode>.
7413 (find_split_point): Likewise.
7414 (combine_simplify_rtx): Likewise.
7415 (simplify_logical): Likewise.
7416 (expand_compound_operation): Likewise.
7417 (expand_field_assignment): Likewise.
7418 (make_compound_operation): Likewise.
7419 (extended_count): Likewise.
7420 (change_zero_ext): Likewise.
7421 (simplify_comparison): Likewise.
7422 * dwarf2out.c (scompare_loc_descriptor): Likewise.
7423 (ucompare_loc_descriptor): Likewise.
7424 (minmax_loc_descriptor): Likewise.
7425 (mem_loc_descriptor): Likewise.
7426 (loc_descriptor): Likewise.
7427 * expmed.c (init_expmed_one_mode): Likewise.
7428 * lra-constraints.c (lra_constraint_offset): Likewise.
7429 * optabs.c (prepare_libcall_arg): Likewise.
7430 * postreload.c (move2add_note_store): Likewise.
7431 * reload.c (operands_match_p): Likewise.
7432 * rtl.h (load_extend_op): Likewise.
7433 * rtlhooks.c (gen_lowpart_general): Likewise.
7434 * simplify-rtx.c (simplify_truncation): Likewise.
7435 (simplify_unary_operation_1): Likewise.
7436 (simplify_binary_operation_1): Likewise.
7437 (simplify_const_binary_operation): Likewise.
7438 (simplify_const_relational_operation): Likewise.
7439 (simplify_subreg): Likewise.
7440 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7441 * var-tracking.c (adjust_mems): Likewise.
7442 (prepare_call_arguments): Likewise.
7444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7445 Alan Hayward <alan.hayward@arm.com>
7446 David Sherwood <david.sherwood@arm.com>
7448 * machmode.h (is_int_mode): New fuction.
7449 * combine.c (find_split_point): Use it.
7450 (combine_simplify_rtx): Likewise.
7451 (simplify_if_then_else): Likewise.
7452 (simplify_set): Likewise.
7453 (simplify_shift_const_1): Likewise.
7454 (simplify_comparison): Likewise.
7455 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
7456 * cse.c (notreg_cost): Likewise.
7457 (cse_insn): Likewise.
7458 * cselib.c (cselib_lookup_1): Likewise.
7459 * dojump.c (do_jump_1): Likewise.
7460 (do_compare_rtx_and_jump): Likewise.
7461 * dse.c (get_call_args): Likewise.
7462 * dwarf2out.c (rtl_for_decl_init): Likewise.
7463 (native_encode_initializer): Likewise.
7464 * expmed.c (emit_store_flag_1): Likewise.
7465 (emit_store_flag): Likewise.
7466 * expr.c (convert_modes): Likewise.
7467 (store_field): Likewise.
7468 (expand_expr_real_1): Likewise.
7469 * fold-const.c (fold_read_from_constant_string): Likewise.
7470 * gimple-ssa-sprintf.c (get_format_string): Likewise.
7471 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
7472 * optabs.c (expand_binop): Likewise.
7473 (expand_unop): Likewise.
7474 (expand_abs_nojump): Likewise.
7475 (expand_one_cmpl_abs_nojump): Likewise.
7476 * simplify-rtx.c (mode_signbit_p): Likewise.
7477 (val_signbit_p): Likewise.
7478 (val_signbit_known_set_p): Likewise.
7479 (val_signbit_known_clear_p): Likewise.
7480 (simplify_relational_operation_1): Likewise.
7481 * tree.c (vector_type_mode): Likewise.
7483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7484 Alan Hayward <alan.hayward@arm.com>
7485 David Sherwood <david.sherwood@arm.com>
7487 * machmode.h (smallest_mode_for_size): Fix formatting.
7488 (smallest_int_mode_for_size): New function.
7489 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
7490 instead of smallest_mode_for_size.
7491 * combine.c (make_extraction): Likewise.
7492 * config/arc/arc.c (arc_expand_movmem): Likewise.
7493 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
7494 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7495 * config/s390/s390.c (s390_expand_insv): Likewise.
7496 * config/sparc/sparc.c (assign_int_registers): Likewise.
7497 * config/spu/spu.c (spu_function_value): Likewise.
7498 (spu_function_arg): Likewise.
7499 * coverage.c (get_gcov_type): Likewise.
7500 (get_gcov_unsigned_t): Likewise.
7501 * dse.c (find_shift_sequence): Likewise.
7502 * expmed.c (store_bit_field_1): Likewise.
7503 * expr.c (convert_move): Likewise.
7504 (store_field): Likewise.
7505 * internal-fn.c (expand_arith_overflow): Likewise.
7506 * optabs-query.c (get_best_extraction_insn): Likewise.
7507 * optabs.c (expand_twoval_binop_libfunc): Likewise.
7508 * stor-layout.c (layout_type): Likewise.
7509 (initialize_sizetypes): Likewise.
7510 * targhooks.c (default_get_mask_mode): Likewise.
7511 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
7513 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7514 Alan Hayward <alan.hayward@arm.com>
7515 David Sherwood <david.sherwood@arm.com>
7517 * machmode.h (opt_mode::else_blk): New function.
7518 (int_mode_for_mode): Declare.
7519 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
7520 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
7522 * cfgexpand.c (expand_debug_expr): Likewise.
7523 * combine.c (gen_lowpart_or_truncate): Likewise.
7524 (gen_lowpart_for_combine): Likewise.
7525 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
7526 * config/avr/avr.c (avr_to_int_mode): Likewise.
7527 (avr_out_plus_1): Likewise.
7528 (avr_out_plus): Likewise.
7529 (avr_out_round): Likewise.
7530 * config/i386/i386.c (ix86_split_to_parts): Likewise.
7531 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
7532 (s390_expand_vcond): Likewise.
7533 * config/spu/spu.c (spu_split_immediate): Likewise.
7534 (spu_expand_mov): Likewise.
7535 * dse.c (get_stored_val): Likewise.
7536 * expmed.c (store_bit_field_1): Likewise.
7537 (convert_extracted_bit_field): Use int_mode_for_mode instead of
7539 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
7540 (extract_low_bits): Likewise.
7541 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
7542 handling rather than repeating the check.
7543 (emit_group_store): Likewise.
7544 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
7545 * optabs.c (expand_absneg_bit): Likewise.
7546 (expand_copysign_absneg): Likewise.
7547 (expand_copysign_bit): Likewise.
7548 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
7549 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
7550 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
7551 * var-tracking.c (prepare_call_arguments): Likewise.
7552 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
7553 int_mode_for_mode instead of mode_for_size.
7554 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
7556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7557 Alan Hayward <alan.hayward@arm.com>
7558 David Sherwood <david.sherwood@arm.com>
7560 * machmode.h (int_mode_for_size): New function.
7561 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
7562 instead of mode_for_size.
7563 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
7565 * combine.c (expand_field_assignment): Use int_mode_for_size
7566 instead of mode_for_size.
7567 (make_extraction): Likewise.
7568 (simplify_shift_const_1): Likewise.
7569 (simplify_comparison): Likewise.
7570 * dojump.c (do_jump): Likewise.
7571 * dwarf2out.c (mem_loc_descriptor): Likewise.
7572 * emit-rtl.c (init_derived_machine_modes): Likewise.
7573 * expmed.c (flip_storage_order): Likewise.
7574 (convert_extracted_bit_field): Likewise.
7575 * expr.c (copy_blkmode_from_reg): Likewise.
7576 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
7577 * internal-fn.c (expand_mul_overflow): Likewise.
7578 * lower-subreg.c (simple_move): Likewise.
7579 * optabs-libfuncs.c (init_optabs): Likewise.
7580 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7581 * tree.c (vector_type_mode): Likewise.
7582 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
7583 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
7584 * tree-vect-generic.c (expand_vector_parallel): Likewise.
7585 * tree-vect-stmts.c (vectorizable_load): Likewise.
7586 (vectorizable_store): Likewise.
7588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7589 Alan Hayward <alan.hayward@arm.com>
7590 David Sherwood <david.sherwood@arm.com>
7592 * coretypes.h (pod_mode): New type.
7593 (scalar_int_mode_pod): New typedef.
7594 * machmode.h (pod_mode): New class.
7595 (int_n_data_t::m): Change type to scalar_int_mode_pod.
7596 * genmodes.c (emit_mode_int_n): Update accordingly.
7597 * lower-subreg.h (target_lower_subreg): Change type to
7598 scalar_int_mode_pod.
7599 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
7600 scalar_int_mode_pod.
7602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7603 Alan Hayward <alan.hayward@arm.com>
7604 David Sherwood <david.sherwood@arm.com>
7606 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
7607 machine_mode to scalar_int_mode.
7608 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
7609 (rs6000_option_override_internal): Remove cast to int.
7610 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
7611 machine_mode to scalar_int_mode.
7612 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
7613 (rs6000_option_override_internal): Remove cast to int.
7614 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
7615 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
7617 * config/s390/s390.c (s390_expand_builtin): Likewise.
7618 * coretypes.h (scalar_int_mode): New type.
7619 (opt_scalar_int_mode): New typedef.
7620 * machmode.h (scalar_int_mode): New class.
7621 (scalar_int_mode::includes_p): New function.
7622 (byte_mode): Change type to scalar_int_mode.
7623 (word_mode): Likewise.
7624 (ptr_mode): Likewise.
7625 * emit-rtl.c (byte_mode): Likewise.
7626 (word_mode): Likewise.
7627 (ptr_mode): Likewise.
7628 (init_derived_machine_modes): Update accordingly.
7629 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
7630 and MODE_PARTIAL_INT.
7631 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
7632 opt_scalar_int_mode.
7634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7635 Alan Hayward <alan.hayward@arm.com>
7636 David Sherwood <david.sherwood@arm.com>
7638 * target.def (libgcc_floating_mode_supported_p): Take a
7640 * doc/tm.texi: Regenerate.
7641 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
7643 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7644 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
7647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7648 Alan Hayward <alan.hayward@arm.com>
7649 David Sherwood <david.sherwood@arm.com>
7651 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
7652 * doc/tm.texi: Regenerate.
7653 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
7654 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
7655 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
7656 * targhooks.h (default_floatn_mode): Likewise.
7657 * targhooks.c (default_floatn_mode): Likewise.
7658 * tree.c (build_common_tree_nodes): Update accordingly.
7660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7661 Alan Hayward <alan.hayward@arm.com>
7662 David Sherwood <david.sherwood@arm.com>
7664 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
7665 (mode_iterator::iterate_p): Likewise.
7666 (mode_iterator::get_wider): Likewise.
7667 * expr.c (init_expr_target): Use opt_scalar_float_mode.
7669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7670 Alan Hayward <alan.hayward@arm.com>
7671 David Sherwood <david.sherwood@arm.com>
7673 * coretypes.h (opt_scalar_float_mode): New typedef.
7674 * machmode.h (float_mode_for_size): New function.
7675 * emit-rtl.c (double_mode): Delete.
7676 (init_emit_once): Use float_mode_for_size.
7677 * stor-layout.c (layout_type): Likewise.
7678 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
7680 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7681 Alan Hayward <alan.hayward@arm.com>
7682 David Sherwood <david.sherwood@arm.com>
7684 * output.h (assemble_real): Take a scalar_float_mode.
7685 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
7686 * config/arm/arm.md (consttable_4): Likewise.
7687 (consttable_8): Likewise.
7688 (consttable_16): Likewise.
7689 * config/mips/mips.md (consttable_float): Likewise.
7690 * config/s390/s390.c (s390_output_pool_entry): Likewise.
7691 * varasm.c (assemble_real): Take a scalar_float_mode.
7692 (output_constant_pool_2): Update accordingly.
7693 (output_constant): Likewise.
7695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7696 Alan Hayward <alan.hayward@arm.com>
7697 David Sherwood <david.sherwood@arm.com>
7699 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
7700 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
7701 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
7702 (native_encode_real): Likewise.
7703 (native_interpret_real): Likewise.
7704 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
7705 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7708 Alan Hayward <alan.hayward@arm.com>
7709 David Sherwood <david.sherwood@arm.com>
7711 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
7712 <scalar_float_mode>. Simplify.
7713 (gen_extend_conv_libfunc): Likewise.
7715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7716 Alan Hayward <alan.hayward@arm.com>
7717 David Sherwood <david.sherwood@arm.com>
7719 * coretypes.h (scalar_float_mode): New type.
7720 * machmode.h (mode_traits::from_int): Use machine_mode if
7721 USE_ENUM_MODES is defined.
7722 (is_a): New function.
7724 (dyn_cast): Likewise.
7725 (scalar_float_mode): New class.
7726 (scalar_float_mode::includes_p): New function.
7727 (is_float_mode): Likewise.
7728 * gdbhooks.py (MachineModePrinter): New class.
7729 (build_pretty_printer): Use it for scalar_float_mode.
7730 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
7731 (format_helper::format_helper): Turn into a template.
7732 * genmodes.c (get_mode_class): New function.
7733 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
7734 or machine_mode if none.
7735 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
7736 as_a <scalar_float_mode>.
7737 * dwarf2out.c (mem_loc_descriptor): Likewise.
7738 (insert_float): Likewise.
7739 (add_const_value_attribute): Likewise.
7740 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7741 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
7742 (expand_unop): Update accordingly.
7743 (expand_abs_nojump): Likewise.
7744 (expand_copysign_absneg): Take a scalar_float_mode.
7745 (expand_copysign_bit): Likewise.
7746 (expand_copysign): Update accordingly.
7748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7749 Alan Hayward <alan.hayward@arm.com>
7750 David Sherwood <david.sherwood@arm.com>
7752 * coretypes.h (opt_mode): New class.
7753 * machmode.h (opt_mode): Likewise.
7754 (opt_mode::else_void): New function.
7755 (opt_mode::require): Likewise.
7756 (opt_mode::exists): Likewise.
7757 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
7758 (GET_MODE_2XWIDER_MODE): Likewise.
7759 (mode_iterator::get_wider): Update accordingly.
7760 (mode_iterator::get_2xwider): Likewise.
7761 (mode_iterator::get_known_wider): Likewise, turning into a template.
7762 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
7763 forcing a wider mode to exist.
7764 * config/cr16/cr16.h (LONG_REG_P): Likewise.
7765 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7766 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
7767 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
7768 * lower-subreg.c (init_lower_subreg): Likewise.
7769 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
7770 on the final iteration.
7771 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
7772 a wider mode exists before asking for a move pattern.
7773 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
7774 forcing a wider mode to exist.
7775 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
7776 returning false if no such mode exists.
7777 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
7778 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
7779 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
7780 Avoid checking for a MODE_INT if we already know the mode is not a
7782 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
7783 forcing a wider mode to exist.
7784 (expmed_mult_highpart_optab): Likewise.
7785 (expmed_mult_highpart): Likewise.
7786 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
7788 * lto-streamer-in.c (lto_input_mode_table): Likewise.
7789 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
7790 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7791 * internal-fn.c (expand_mul_overflow): Update use of
7792 GET_MODE_2XWIDER_MODE.
7793 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7794 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
7795 GET_MODE_WIDER_MODE.
7796 (convert_plusminus_to_widen): Likewise.
7797 * tree-switch-conversion.c (array_value_type): Likewise.
7798 * var-tracking.c (emit_note_insn_var_location): Likewise.
7799 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7800 Return false inside rather than outside the loop if no wider mode
7802 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
7803 and GET_MODE_2XWIDER_MODE
7804 (can_compare_p): Use else_void.
7805 * gdbhooks.py (OptMachineModePrinter): New class.
7806 (build_pretty_printer): Use it for opt_mode.
7808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7809 Alan Hayward <alan.hayward@arm.com>
7810 David Sherwood <david.sherwood@arm.com>
7812 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
7813 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
7815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7816 Alan Hayward <alan.hayward@arm.com>
7817 David Sherwood <david.sherwood@arm.com>
7819 * machmode.h (mode_traits): New structure.
7820 (get_narrowest_mode): New function.
7821 (mode_iterator::start): Likewise.
7822 (mode_iterator::iterate_p): Likewise.
7823 (mode_iterator::get_wider): Likewise.
7824 (mode_iterator::get_known_wider): Likewise.
7825 (mode_iterator::get_2xwider): Likewise.
7826 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
7827 (FOR_EACH_MODE): Likewise.
7828 (FOR_EACH_MODE_FROM): Likewise.
7829 (FOR_EACH_MODE_UNTIL): Likewise.
7830 (FOR_EACH_WIDER_MODE): Likewise.
7831 (FOR_EACH_2XWIDER_MODE): Likewise.
7832 * builtins.c (expand_builtin_strlen): Use new mode iterators.
7833 * combine.c (simplify_comparison): Likewise
7834 * config/i386/i386.c (type_natural_mode): Likewise.
7835 * cse.c (cse_insn): Likewise.
7836 * dse.c (find_shift_sequence): Likewise.
7837 * emit-rtl.c (init_derived_machine_modes): Likewise.
7838 (init_emit_once): Likewise.
7839 * explow.c (hard_function_value): Likewise.
7840 * expmed.c (extract_fixed_bit_field_1): Likewise.
7841 (extract_bit_field_1): Likewise.
7842 (expand_divmod): Likewise.
7843 (emit_store_flag_1): Likewise.
7844 * expr.c (init_expr_target): Likewise.
7845 (convert_move): Likewise.
7846 (alignment_for_piecewise_move): Likewise.
7847 (widest_int_mode_for_size): Likewise.
7848 (emit_block_move_via_movmem): Likewise.
7849 (copy_blkmode_to_reg): Likewise.
7850 (set_storage_via_setmem): Likewise.
7851 (compress_float_constant): Likewise.
7852 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7853 * optabs-query.c (get_best_extraction_insn): Likewise.
7854 * optabs.c (expand_binop): Likewise.
7855 (expand_twoval_unop): Likewise.
7856 (expand_twoval_binop): Likewise.
7857 (widen_leading): Likewise.
7858 (widen_bswap): Likewise.
7859 (expand_parity): Likewise.
7860 (expand_unop): Likewise.
7861 (prepare_cmp_insn): Likewise.
7862 (prepare_float_lib_cmp): Likewise.
7863 (expand_float): Likewise.
7864 (expand_fix): Likewise.
7865 (expand_sfix_optab): Likewise.
7866 * postreload.c (move2add_use_add2_insn): Likewise.
7867 * reg-stack.c (reg_to_stack): Likewise.
7868 * reginfo.c (choose_hard_reg_mode): Likewise.
7869 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7870 * stor-layout.c (mode_for_size): Likewise.
7871 (smallest_mode_for_size): Likewise.
7872 (mode_for_vector): Likewise.
7873 (finish_bitfield_representative): Likewise.
7874 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
7875 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
7876 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7877 * var-tracking.c (prepare_call_arguments): Likewise.
7879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7880 Alan Hayward <alan.hayward@arm.com>
7881 David Sherwood <david.sherwood@arm.com>
7883 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
7884 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
7885 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
7886 * machmode.h (mode_size): Move earlier in file.
7887 (mode_precision): Likewise.
7888 (mode_inner): Likewise.
7889 (mode_nunits): Likewise.
7890 (mode_unit_size): Likewise.
7891 (unit_unit_precision): Likewise.
7892 (mode_wider): Likewise.
7893 (mode_2xwider): Likewise.
7894 (machine_mode): New class.
7895 (mode_to_bytes): New function.
7896 (mode_to_bits): Likewise.
7897 (mode_to_precision): Likewise.
7898 (mode_to_inner): Likewise.
7899 (mode_to_unit_size): Likewise.
7900 (mode_to_unit_precision): Likewise.
7901 (mode_to_nunits): Likewise.
7902 (GET_MODE_SIZE): Use mode_to_bytes.
7903 (GET_MODE_BITSIZE): Use mode_to_bits.
7904 (GET_MODE_PRECISION): Use mode_to_precision.
7905 (GET_MODE_INNER): Use mode_to_inner.
7906 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
7907 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
7908 (GET_MODE_NUNITS): Use mode_to_nunits.
7909 * system.h (ALWAYS_INLINE): New macro.
7910 * config/powerpcspe/powerpcspe-c.c
7911 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
7912 int for arg1_mode and arg2_mode.
7914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7915 Alan Hayward <alan.hayward@arm.com>
7916 David Sherwood <david.sherwood@arm.com>
7918 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7919 Prefix mode names with E_ in case statements.
7920 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7921 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
7922 (aarch64_split_simd_move): Likewise.
7923 (aarch64_gen_storewb_pair): Likewise.
7924 (aarch64_gen_loadwb_pair): Likewise.
7925 (aarch64_gen_store_pair): Likewise.
7926 (aarch64_gen_load_pair): Likewise.
7927 (aarch64_get_condition_code_1): Likewise.
7928 (aarch64_constant_pool_reload_icode): Likewise.
7929 (get_rsqrte_type): Likewise.
7930 (get_rsqrts_type): Likewise.
7931 (get_recpe_type): Likewise.
7932 (get_recps_type): Likewise.
7933 (aarch64_gimplify_va_arg_expr): Likewise.
7934 (aarch64_simd_container_mode): Likewise.
7935 (aarch64_emit_load_exclusive): Likewise.
7936 (aarch64_emit_store_exclusive): Likewise.
7937 (aarch64_expand_compare_and_swap): Likewise.
7938 (aarch64_gen_atomic_cas): Likewise.
7939 (aarch64_emit_bic): Likewise.
7940 (aarch64_emit_atomic_swap): Likewise.
7941 (aarch64_emit_atomic_load_op): Likewise.
7942 (aarch64_evpc_trn): Likewise.
7943 (aarch64_evpc_uzp): Likewise.
7944 (aarch64_evpc_zip): Likewise.
7945 (aarch64_evpc_ext): Likewise.
7946 (aarch64_evpc_rev): Likewise.
7947 (aarch64_evpc_dup): Likewise.
7948 (aarch64_gen_ccmp_first): Likewise.
7949 (aarch64_gen_ccmp_next): Likewise.
7950 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7951 (alpha_emit_xfloating_libcall): Likewise.
7952 (emit_insxl): Likewise.
7953 (alpha_arg_type): Likewise.
7954 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
7955 (arc_preferred_simd_mode): Likewise.
7956 (arc_secondary_reload): Likewise.
7957 (get_arc_condition_code): Likewise.
7958 (arc_print_operand): Likewise.
7959 (arc_legitimate_constant_p): Likewise.
7960 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7961 * config/arc/arc.md (casesi_load): Likewise.
7962 (casesi_compact_jump): Likewise.
7963 * config/arc/predicates.md (proper_comparison_operator): Likewise.
7964 (cc_use_register): Likewise.
7965 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7966 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7967 (arm_init_iwmmxt_builtins): Likewise.
7968 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
7969 (neon_expand_vector_init): Likewise.
7970 (arm_attr_length_move_neon): Likewise.
7971 (maybe_get_arm_condition_code): Likewise.
7972 (arm_emit_vector_const): Likewise.
7973 (arm_preferred_simd_mode): Likewise.
7974 (arm_output_iwmmxt_tinsr): Likewise.
7975 (thumb1_output_casesi): Likewise.
7976 (thumb2_output_casesi): Likewise.
7977 (arm_emit_load_exclusive): Likewise.
7978 (arm_emit_store_exclusive): Likewise.
7979 (arm_expand_compare_and_swap): Likewise.
7980 (arm_evpc_neon_vuzp): Likewise.
7981 (arm_evpc_neon_vzip): Likewise.
7982 (arm_evpc_neon_vrev): Likewise.
7983 (arm_evpc_neon_vtrn): Likewise.
7984 (arm_evpc_neon_vext): Likewise.
7985 (arm_validize_comparison): Likewise.
7986 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
7987 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
7988 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
7989 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
7990 (c6x_preferred_simd_mode): Likewise.
7991 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
7992 (epiphany_rtx_costs): Likewise.
7993 * config/epiphany/predicates.md (proper_comparison_operator):
7995 * config/frv/frv.c (condexec_memory_operand): Likewise.
7996 (frv_emit_move): Likewise.
7997 (output_move_single): Likewise.
7998 (output_condmove_single): Likewise.
7999 (frv_hard_regno_mode_ok): Likewise.
8000 (frv_matching_accg_mode): Likewise.
8001 * config/h8300/h8300.c (split_adds_subs): Likewise.
8002 (h8300_rtx_costs): Likewise.
8003 (h8300_print_operand): Likewise.
8004 (compute_mov_length): Likewise.
8005 (output_logical_op): Likewise.
8006 (compute_logical_op_length): Likewise.
8007 (compute_logical_op_cc): Likewise.
8008 (h8300_shift_needs_scratch_p): Likewise.
8009 (output_a_shift): Likewise.
8010 (compute_a_shift_length): Likewise.
8011 (compute_a_shift_cc): Likewise.
8012 (expand_a_rotate): Likewise.
8013 (output_a_rotate): Likewise.
8014 * config/i386/i386.c (classify_argument): Likewise.
8015 (function_arg_advance_32): Likewise.
8016 (function_arg_32): Likewise.
8017 (function_arg_64): Likewise.
8018 (function_value_64): Likewise.
8019 (ix86_gimplify_va_arg): Likewise.
8020 (ix86_legitimate_constant_p): Likewise.
8021 (put_condition_code): Likewise.
8022 (split_double_mode): Likewise.
8023 (ix86_avx256_split_vector_move_misalign): Likewise.
8024 (ix86_expand_vector_logical_operator): Likewise.
8025 (ix86_split_idivmod): Likewise.
8026 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
8027 (ix86_build_const_vector): Likewise.
8028 (ix86_build_signbit_mask): Likewise.
8029 (ix86_match_ccmode): Likewise.
8030 (ix86_cc_modes_compatible): Likewise.
8031 (ix86_expand_branch): Likewise.
8032 (ix86_expand_sse_cmp): Likewise.
8033 (ix86_expand_sse_movcc): Likewise.
8034 (ix86_expand_int_sse_cmp): Likewise.
8035 (ix86_expand_vec_perm_vpermi2): Likewise.
8036 (ix86_expand_vec_perm): Likewise.
8037 (ix86_expand_sse_unpack): Likewise.
8038 (ix86_expand_int_addcc): Likewise.
8039 (ix86_split_to_parts): Likewise.
8040 (ix86_vectorize_builtin_gather): Likewise.
8041 (ix86_vectorize_builtin_scatter): Likewise.
8042 (avx_vpermilp_parallel): Likewise.
8043 (inline_memory_move_cost): Likewise.
8044 (ix86_tieable_integer_mode_p): Likewise.
8045 (x86_maybe_negate_const_int): Likewise.
8046 (ix86_expand_vector_init_duplicate): Likewise.
8047 (ix86_expand_vector_init_one_nonzero): Likewise.
8048 (ix86_expand_vector_init_one_var): Likewise.
8049 (ix86_expand_vector_init_concat): Likewise.
8050 (ix86_expand_vector_init_interleave): Likewise.
8051 (ix86_expand_vector_init_general): Likewise.
8052 (ix86_expand_vector_set): Likewise.
8053 (ix86_expand_vector_extract): Likewise.
8054 (emit_reduc_half): Likewise.
8055 (ix86_emit_i387_round): Likewise.
8056 (ix86_mangle_type): Likewise.
8057 (ix86_expand_round_sse4): Likewise.
8058 (expand_vec_perm_blend): Likewise.
8059 (canonicalize_vector_int_perm): Likewise.
8060 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
8061 (expand_vec_perm_1): Likewise.
8062 (expand_vec_perm_interleave3): Likewise.
8063 (expand_vec_perm_even_odd_pack): Likewise.
8064 (expand_vec_perm_even_odd_1): Likewise.
8065 (expand_vec_perm_broadcast_1): Likewise.
8066 (ix86_vectorize_vec_perm_const_ok): Likewise.
8067 (ix86_expand_vecop_qihi): Likewise.
8068 (ix86_expand_mul_widen_hilo): Likewise.
8069 (ix86_expand_sse2_abs): Likewise.
8070 (ix86_expand_pextr): Likewise.
8071 (ix86_expand_pinsr): Likewise.
8072 (ix86_preferred_simd_mode): Likewise.
8073 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
8074 * config/i386/sse.md (*andnot<mode>3): Likewise.
8075 (<mask_codefor><code><mode>3<mask_name>): Likewise.
8076 (*<code><mode>3): Likewise.
8077 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
8078 (ia64_expand_atomic_op): Likewise.
8079 (ia64_arg_type): Likewise.
8080 (ia64_mode_to_int): Likewise.
8081 (ia64_scalar_mode_supported_p): Likewise.
8082 (ia64_vector_mode_supported_p): Likewise.
8083 (expand_vec_perm_broadcast): Likewise.
8084 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
8085 (iq2000_function_arg_advance): Likewise.
8086 (iq2000_function_arg): Likewise.
8087 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
8088 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
8089 (m68k_libcall_value): Likewise.
8090 (m68k_function_value): Likewise.
8091 (sched_attr_op_type): Likewise.
8092 * config/mcore/mcore.c (mcore_output_move): Likewise.
8093 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
8095 (microblaze_function_arg): Likewise.
8096 * config/mips/mips.c (mips16_build_call_stub): Likewise.
8097 (mips_print_operand): Likewise.
8098 (mips_mode_ok_for_mov_fmt_p): Likewise.
8099 (mips_vector_mode_supported_p): Likewise.
8100 (mips_preferred_simd_mode): Likewise.
8101 (mips_expand_vpc_loongson_even_odd): Likewise.
8102 (mips_expand_vec_unpack): Likewise.
8103 (mips_expand_vi_broadcast): Likewise.
8104 (mips_expand_vector_init): Likewise.
8105 (mips_expand_vec_reduc): Likewise.
8106 (mips_expand_msa_cmp): Likewise.
8107 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
8108 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
8109 (cc_flags_for_mode): Likewise.
8110 * config/msp430/msp430.c (msp430_print_operand): Likewise.
8111 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
8112 (nds32_output_casesi_pc_relative): Likewise.
8113 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8114 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
8115 (nvptx_gen_unpack): Likewise.
8116 (nvptx_gen_pack): Likewise.
8117 (nvptx_gen_shuffle): Likewise.
8118 (nvptx_gen_wcast): Likewise.
8119 (nvptx_preferred_simd_mode): Likewise.
8120 * config/pa/pa.c (pa_secondary_reload): Likewise.
8121 * config/pa/predicates.md (base14_operand): Likewise.
8122 * config/powerpcspe/powerpcspe-c.c
8123 (altivec_resolve_overloaded_builtin): Likewise.
8124 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
8126 (rs6000_preferred_simd_mode): Likewise.
8127 (output_vec_const_move): Likewise.
8128 (rs6000_expand_vector_extract): Likewise.
8129 (rs6000_split_vec_extract_var): Likewise.
8130 (reg_offset_addressing_ok_p): Likewise.
8131 (rs6000_legitimate_offset_address_p): Likewise.
8132 (rs6000_legitimize_address): Likewise.
8133 (rs6000_emit_set_const): Likewise.
8134 (rs6000_const_vec): Likewise.
8135 (rs6000_emit_move): Likewise.
8136 (spe_build_register_parallel): Likewise.
8137 (rs6000_darwin64_record_arg_recurse): Likewise.
8138 (swap_selector_for_mode): Likewise.
8139 (spe_init_builtins): Likewise.
8140 (paired_init_builtins): Likewise.
8141 (altivec_init_builtins): Likewise.
8142 (do_load_for_compare): Likewise.
8143 (rs6000_generate_compare): Likewise.
8144 (rs6000_expand_float128_convert): Likewise.
8145 (emit_load_locked): Likewise.
8146 (emit_store_conditional): Likewise.
8147 (rs6000_output_function_epilogue): Likewise.
8148 (rs6000_handle_altivec_attribute): Likewise.
8149 (rs6000_function_value): Likewise.
8150 (emit_fusion_gpr_load): Likewise.
8151 (emit_fusion_p9_load): Likewise.
8152 (emit_fusion_p9_store): Likewise.
8153 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
8154 (fusion_gpr_mem_load): Likewise.
8155 (fusion_addis_mem_combo_load): Likewise.
8156 (fusion_addis_mem_combo_store): Likewise.
8157 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
8158 (fusion_gpr_mem_load): Likewise.
8159 (fusion_addis_mem_combo_load): Likewise.
8160 (fusion_addis_mem_combo_store): Likewise.
8161 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8163 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
8164 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8165 (rs6000_preferred_simd_mode): Likewise.
8166 (output_vec_const_move): Likewise.
8167 (rs6000_expand_vector_extract): Likewise.
8168 (rs6000_split_vec_extract_var): Likewise.
8169 (reg_offset_addressing_ok_p): Likewise.
8170 (rs6000_legitimate_offset_address_p): Likewise.
8171 (rs6000_legitimize_address): Likewise.
8172 (rs6000_emit_set_const): Likewise.
8173 (rs6000_const_vec): Likewise.
8174 (rs6000_emit_move): Likewise.
8175 (rs6000_darwin64_record_arg_recurse): Likewise.
8176 (swap_selector_for_mode): Likewise.
8177 (paired_init_builtins): Likewise.
8178 (altivec_init_builtins): Likewise.
8179 (rs6000_expand_float128_convert): Likewise.
8180 (emit_load_locked): Likewise.
8181 (emit_store_conditional): Likewise.
8182 (rs6000_output_function_epilogue): Likewise.
8183 (rs6000_handle_altivec_attribute): Likewise.
8184 (rs6000_function_value): Likewise.
8185 (emit_fusion_gpr_load): Likewise.
8186 (emit_fusion_p9_load): Likewise.
8187 (emit_fusion_p9_store): Likewise.
8188 * config/rx/rx.c (rx_gen_move_template): Likewise.
8189 (flags_from_mode): Likewise.
8190 * config/s390/predicates.md (s390_alc_comparison): Likewise.
8191 (s390_slb_comparison): Likewise.
8192 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
8193 (s390_vector_mode_supported_p): Likewise.
8194 (s390_cc_modes_compatible): Likewise.
8195 (s390_match_ccmode_set): Likewise.
8196 (s390_canonicalize_comparison): Likewise.
8197 (s390_emit_compare_and_swap): Likewise.
8198 (s390_branch_condition_mask): Likewise.
8199 (s390_rtx_costs): Likewise.
8200 (s390_secondary_reload): Likewise.
8201 (__SECONDARY_RELOAD_CASE): Likewise.
8202 (s390_expand_cs): Likewise.
8203 (s390_preferred_simd_mode): Likewise.
8204 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
8205 * config/sh/sh.c (sh_print_operand): Likewise.
8206 (dump_table): Likewise.
8207 (sh_secondary_reload): Likewise.
8208 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8209 * config/sh/sh.md (casesi_worker_1): Likewise.
8210 (casesi_worker_2): Likewise.
8211 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
8212 (fcc_comparison_operator): Likewise.
8213 * config/sparc/sparc.c (sparc_expand_move): Likewise.
8214 (emit_soft_tfmode_cvt): Likewise.
8215 (sparc_preferred_simd_mode): Likewise.
8216 (output_cbranch): Likewise.
8217 (sparc_print_operand): Likewise.
8218 (sparc_expand_vec_perm_bmask): Likewise.
8219 (vector_init_bshuffle): Likewise.
8220 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8221 (spu_vector_mode_supported_p): Likewise.
8222 (spu_expand_insv): Likewise.
8223 (spu_emit_branch_or_set): Likewise.
8224 (spu_handle_vector_attribute): Likewise.
8225 (spu_builtin_splats): Likewise.
8226 (spu_builtin_extract): Likewise.
8227 (spu_builtin_promote): Likewise.
8228 (spu_expand_sign_extend): Likewise.
8229 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8230 (tilegx_simd_int): Likewise.
8231 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
8232 (tilepro_simd_int): Likewise.
8233 * config/v850/v850.c (const_double_split): Likewise.
8234 (v850_print_operand): Likewise.
8235 (ep_memory_offset): Likewise.
8236 * config/vax/vax.c (vax_rtx_costs): Likewise.
8237 (vax_output_int_move): Likewise.
8238 (vax_output_int_add): Likewise.
8239 (vax_output_int_subtract): Likewise.
8240 * config/visium/predicates.md (visium_branch_operator): Likewise.
8241 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
8242 (visium_print_operand_address): Likewise.
8243 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8244 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
8245 (xtensa_expand_conditional_branch): Likewise.
8246 (xtensa_copy_incoming_a7): Likewise.
8247 (xtensa_output_literal): Likewise.
8248 * dfp.c (decimal_real_maxval): Likewise.
8249 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8252 Alan Hayward <alan.hayward@arm.com>
8253 David Sherwood <david.sherwood@arm.com>
8255 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
8256 (mode_nunits_inline): Likewise.
8257 (mode_inner_inline): Likewise.
8258 (mode_unit_size_inline): Likewise.
8259 (mode_unit_precision_inline): Likewise.
8260 (emit_insn_modes_h): Likewise. Also emit a #define of the
8262 (emit_mode_wider): Add an E_ prefix to mode names.
8263 (emit_mode_complex): Likewise.
8264 (emit_mode_inner): Likewise.
8265 (emit_mode_adjustments): Likewise.
8266 (emit_mode_int_n): Likewise.
8267 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
8268 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
8269 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
8270 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
8271 (CRC32_BUILTIN, ENTRY): Likewise.
8272 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
8273 (aarch64_pop_regs): Likewise.
8274 (aarch64_process_components): Likewise.
8275 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
8276 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
8277 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
8278 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
8279 * config/arm/arm.c (arm_init_libfuncs): Likewise.
8280 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
8282 * config/i386/i386-builtin.def (pcmpestr): Likewise.
8283 (pcmpistr): Likewise.
8284 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
8285 * config/mmix/mmix.c (mmix_output_condition): Likewise.
8286 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8288 * config/rl78/rl78.c (mduc_regs): Likewise.
8289 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8290 (htm_expand_builtin): Likewise.
8291 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
8292 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
8293 * config/xtensa/xtensa.c (print_operand): Likewise.
8294 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
8295 (NUM_MODE_VECTOR_INT): Likewise.
8296 * genoutput.c (null_operand): Likewise.
8297 (output_operand_data): Likewise.
8298 * genrecog.c (print_parameter_value): Likewise.
8299 * lra.c (debug_operand_data): Likewise.
8301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8303 * dwarf2out.c (scompare_loc_descriptor_wide)
8304 (scompare_loc_descriptor_narrow): New functions, split out from...
8305 (scompare_loc_descriptor): ...here.
8306 * expmed.c (emit_store_flag_int): New function, split out from...
8307 (emit_store_flag): ...here.
8309 2017-08-30 Richard Biener <rguenther@suse.de>
8311 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
8312 (dwarf2out_early_finish): Move setting of AT_pubnames from
8313 early debug output to early finish.
8315 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8317 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
8318 and -mdata-region to the assembler.
8320 2017-08-30 Richard Biener <rguenther@suse.de>
8322 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
8324 (gen_subprogram_die): Add DW_AT_object_pointer only early.
8325 (dwarf2out_early_global_decl): Only generate a DIE for the
8326 abstract origin if it doesn't already exist or is a declaration DIE.
8327 (resolve_addr): Do not add the linkage name twice when
8328 generating a stub DIE for the DW_TAG_GNU_call_site target.
8330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8332 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8333 Use machine_mode rather than int for arg1_mode.
8335 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8338 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
8339 that the second argument of the built-in functions to unpack
8340 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
8341 switch statement instead a lot of if statements.
8342 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
8343 Allow 64-bit values to be in Altivec registers as well as
8344 traditional floating point registers.
8345 (pack<mode>, FMOVE128_VSX iterator): Likewise.
8347 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
8349 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
8350 MAX_REGS_PER_ADDRESS == 1.
8352 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
8354 * config/i386/i386.opt (flag_fentry): Do not init to -1.
8355 * config/i386/i386.c (ix86_option_override_internal): Simplify
8356 setting of opts->x_flag_entry.
8358 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8359 Jakub Jelinek <jakub@redhat.com>
8360 Richard Biener <rguenther@suse.de>
8362 PR tree-optimization/81503
8363 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
8364 folded constant fits in the target type; reorder tests for clarity.
8366 2017-08-29 Martin Liska <mliska@suse.cz>
8368 * passes.def: Include pass_lower_switch.
8369 * stmt.c (dump_case_nodes): Remove and move to
8370 tree-switch-conversion.
8371 (case_values_threshold): Likewise.
8372 (expand_switch_as_decision_tree_p): Likewise.
8373 (emit_case_decision_tree): Likewise.
8374 (expand_case): Likewise.
8375 (balance_case_nodes): Likewise.
8376 (node_has_low_bound): Likewise.
8377 (node_has_high_bound): Likewise.
8378 (node_is_bounded): Likewise.
8379 (emit_case_nodes): Likewise.
8380 (struct simple_case_node): New struct.
8381 (add_case_node): Remove.
8382 (emit_case_dispatch_table): Use vector instead of case_list.
8383 (reset_out_edges_aux): Remove.
8384 (compute_cases_per_edge): Likewise.
8385 (expand_case): Build list of simple_case_node.
8386 (expand_sjlj_dispatch_table): Use it.
8387 * tree-switch-conversion.c (struct case_node): Moved from
8388 stmt.c and adjusted.
8389 (emit_case_nodes): Likewise.
8390 (node_has_low_bound): Likewise.
8391 (node_has_high_bound): Likewise.
8392 (node_is_bounded): Likewise.
8393 (case_values_threshold): Likewise.
8394 (reset_out_edges_aux): Likewise.
8395 (compute_cases_per_edge): Likewise.
8396 (add_case_node): Likewise.
8397 (dump_case_nodes): Likewise.
8398 (balance_case_nodes): Likewise.
8399 (expand_switch_as_decision_tree_p): Likewise.
8400 (emit_jump): Likewise.
8401 (emit_case_decision_tree): Likewise.
8402 (try_switch_expansion): Likewise.
8403 (do_jump_if_equal): Likewise.
8404 (emit_cmp_and_jump_insns): Likewise.
8405 (fix_phi_operands_for_edge): New function.
8406 (record_phi_operand_mapping): Likewise.
8407 (class pass_lower_switch): New pass.
8408 (pass_lower_switch::execute): New function.
8409 (make_pass_lower_switch): Likewise.
8410 (conditional_probability):
8411 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
8412 * tree-pass.h: Add make_pass_lower_switch.
8414 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8417 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
8420 2017-08-29 Richard Biener <rguenther@suse.de>
8422 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
8423 we do not add a DW_AT_inline attribute twice.
8424 (gen_subprogram_die): Remove code setting DW_AT_inline on
8425 DECL_ABSTRACT_P nodes.
8427 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
8429 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
8430 calls to internal functions.
8431 (gimplify_modify_expr): Likewise.
8432 * tree-call-cdce.c (use_internal_fn): Likewise.
8433 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
8434 (convert_to_divmod): Set the nothrow flag.
8435 * tree-if-conv.c (predicate_mem_writes): Likewise.
8436 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8437 (vectorizable_call): Likewise.
8438 (vectorizable_store): Likewise.
8439 (vectorizable_load): Likewise.
8440 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
8441 (vect_recog_mask_conversion_pattern): Likewise.
8443 2017-08-29 Martin Liska <mliska@suse.cz>
8446 * gcc.c (driver_handle_option): Add new argument.
8447 * opts-common.c (handle_option): Pass
8448 target_option_override_hook.
8449 * opts-global.c (lang_handle_option): Add new option.
8450 (set_default_handlers): Add new argument.
8451 (decode_options): Likewise.
8452 * opts.c (target_handle_option): Likewise.
8453 (common_handle_option): Call target_option_override_hook.
8454 * opts.h (struct cl_option_handler_func): Add hook for
8455 target option override.
8456 (struct cl_option_handlers): Likewise.
8457 (set_default_handlers): Add new argument.
8458 (decode_options): Likewise.
8459 (common_handle_option): Likewise.
8460 (target_handle_option): Likewise.
8461 * toplev.c (toplev::main): Pass targetm.target_option.override
8464 2017-08-29 Richard Biener <rguenther@suse.de>
8465 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8467 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
8468 life to the active subtree.
8470 2017-08-28 Jeff Law <law@redhat.com>
8472 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
8473 derive_equivalences.
8474 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
8476 (edge_info::derive_equivalences): New private member function
8478 * tree-ssa-dom.c (class edge_info): Changed from a struct
8479 to a class. Add ctor/dtor, methods and data members.
8480 (edge_info::edge_info): Renamed from allocate_edge_info.
8481 Initialize additional members.
8482 (edge_info::~edge_info): New.
8483 (free_dom_edge_info): Delete the edge info.
8484 (record_edge_info): Use new class & associated member functions.
8485 Tighten forms for testing for edge equivalences.
8486 (record_temporary_equivalences): Iterate over the simple
8487 equivalences rather than assuming there's only one per edge.
8488 (cprop_into_successor_phis): Iterate over the simple
8489 equivalences rather than assuming there's only one per edge.
8490 (optimize_stmt): Use operand_equal_p rather than pointer
8491 equality for mini-DSE code.
8493 2017-08-28 Nathan Sidwell <nathan@acm.org>
8495 * gcc.c (execute): Fold SIGPIPE handling into switch
8496 statement. Adjust internal error message.
8498 2017-08-28 Richard Biener <rguenther@suse.de>
8501 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
8502 Do nothing for removed DIEs.
8504 2017-08-28 Richard Biener <rguenther@suse.de>
8506 PR tree-optimization/81977
8507 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
8510 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
8513 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
8514 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
8515 using targetm.gen_mem_thread_fence.
8517 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
8520 * config/i386/i386.md (*<btsc><mode>): Change operand 2
8521 predicate to register_operand. Reorder operands.
8522 (*btr<mode>): Ditto.
8523 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
8524 (*btr<mode>_mask): Ditto.
8526 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
8528 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
8529 * config/rs6000/xmmintrin.h: New file.
8530 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
8532 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8535 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
8536 parameter and_insn and return it.
8537 (recombine_lvx_pattern): Insert a copy to ensure availability of
8538 the base register of the copied masking operation at the point of
8539 the instruction replacement.
8540 (recombine_stvx_pattern): Likewise.
8542 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
8544 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
8545 undocumented switches.
8546 (-mpower9-dform-vector): Likewise.
8547 (-mpower9-dform): Likewise.
8548 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
8549 comments to delete references to -mpower9-dform* switches.
8550 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
8551 Delete reference to -mpower9-dform* switches, test for
8552 -mpower9-vector instead.
8553 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
8554 (OTHER_P9_VECTOR_MASKS): Likewise.
8555 (POWERPC_MASKS): Likewise.
8556 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
8557 tests against -mpower9-dform* to -mpower9-vector. Delete code
8558 that checked for -mpower9-dform* consistancy with other options.
8559 Add test for -mpower9-misc to enable other power9 switches.
8560 (rs6000_init_hard_regno_mode_ok): Likewise.
8561 (rs6000_option_override_internal): Likewise.
8562 (rs6000_emit_prologue): Likewise.
8563 (rs6000_emit_epilogue): Likewise.
8564 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
8565 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
8566 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
8568 (emit_fusion_p9_store): Likewise.
8569 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
8570 resetting these macros if the assembler does not support ISA 3.0
8572 (TARGET_P9_DFORM_VECTOR): Likewise.
8573 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
8574 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
8576 2017-08-25 Alan Modra <amodra@gmail.com>
8578 PR rtl-optimization/81747
8579 * cse.c (cse_extended_basic_block): Don't attempt to record
8580 equivalences for degenerate conditional branches that branch
8581 to their fall-through.
8583 2017-08-24 Martin Sebor <msebor@redhat.com>
8586 * gimple-fold.c (size_must_be_zero_p): New function.
8587 (gimple_fold_builtin_memory_op): Call it.
8589 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
8591 * config/rs6000/mm_malloc.h: New file.
8593 2017-08-24 Bin Cheng <bin.cheng@arm.com>
8595 PR tree-optimization/81913
8596 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
8597 analysis when either IVs in condition can wrap.
8599 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
8601 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
8602 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
8604 2017-08-24 Richard Biener <rguenther@suse.de>
8607 * targhooks.c (default_target_can_inline_p): Properly
8608 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
8609 is present and always compare.
8610 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
8611 infer -mfpmath=sse from TARGET_SSE_P.
8612 (ix86_can_inline_p): Properly use target_option_default_node when
8613 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
8615 2017-08-24 Richard Biener <rguenther@suse.de>
8618 * dwarf2out.c (output_die): Handle flag_generate_offload like
8620 (output_comp_unit): Likewise.
8621 (gen_array_type_die): Likewise.
8622 (dwarf2out_early_finish): Likewise.
8623 (note_variable_value_in_expr): Likewise.
8624 (dwarf2out_finish): Likewise. Adjust assert.
8625 * cgraphunit.c (symbol_table::compile): Move setting of
8626 flag_generate_offload earlier ...
8627 (symbol_table::finalize_compilation_unit): ... here, before
8628 early debug finalization.
8630 2017-08-24 Richard Biener <rguenther@suse.de>
8632 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
8633 and ipa-fnsummary.h.
8634 (ix86_can_inline_p): When ix86_fpmath flags do not match
8635 check whether the callee uses FP math at all.
8637 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
8640 * tree-ssanames.c (get_nonzero_bits): Use element_precision
8641 instead of TYPE_PRECISION.
8643 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
8644 Alan Hayward <alan.hayward@arm.com>
8645 David Sherwood <david.sherwood@arm.com>
8647 * combine.c (make_extraction): Use subreg_offset_from_lsb.
8649 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
8651 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
8653 (ix86_frame::stack_realign_allocate): New field.
8654 (struct machine_frame_state): Modify comments.
8655 (machine_frame_state::sp_realigned_fp_end): New field.
8656 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
8658 (sp_valid_at): Add assertion to assure no attempt to access invalid
8659 offset of a realigned stack.
8660 (fp_valid_at): Likewise.
8661 (choose_baseaddr): Modify comments.
8662 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
8663 ix86_expand_prologue.
8664 (ix86_expand_prologue): Modify stack realignment and allocation.
8665 (ix86_expand_epilogue): Modify comments.
8666 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
8667 avx2_runtime, avx512f, and avx512f_runtime.
8669 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
8671 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
8672 (mstackrealign): Do not init to -1.
8673 * config/i386/i386.c (ix86_option_override_internal):
8674 Check opts_set, not opts when setting default value of
8675 opts->x_ix86_force_align_arg_pointer.
8677 2017-08-23 Richard Biener <rguenther@suse.de>
8679 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
8680 lang_hooks.decl_printable_name.
8681 * print-rtl-function.c (print_rtx_function): Likewise.
8682 * tree-pretty-print.c (dump_function_header): Likewise.
8684 2017-08-23 Richard Biener <rguenther@suse.de>
8687 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
8688 -g0 at compile-time.
8690 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8693 * doc/sourcebuild.texi (Other hardware attributes):
8696 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8699 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
8700 Fix single-use check.
8702 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
8704 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
8706 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8708 * doc/install.texi: Modify to add more details on running selected
8711 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8713 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
8714 is combined with -mabi=ms.
8715 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
8718 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8720 PR tree-optimization/81488
8721 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
8722 and cached_basis fields.
8723 (MAX_SPREAD): New constant.
8724 (alloc_cand_and_find_basis): Initialize new fields.
8725 (clear_visited): New function.
8726 (create_phi_basis_1): Rename from create_phi_basis, set visited
8727 and cached_basis fields.
8728 (create_phi_basis): New wrapper function.
8729 (phi_add_costs_1): Rename from phi_add_costs, add spread
8730 parameter, set visited field, short-circuit when limits reached.
8731 (phi_add_costs): New wrapper function.
8732 (record_phi_increments_1): Rename from record_phi_increments, set
8734 (record_phi_increments): New wrapper function.
8735 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
8736 (phi_incr_cost): New wrapper function.
8737 (all_phi_incrs_profitable_1): Rename from
8738 all_phi_incrs_profitable, set visited field.
8739 (all_phi_incrs_profitable): New wrapper function.
8741 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8742 Alan Hayward <alan.hayward@arm.com>
8743 David Sherwood <david.sherwood@arm.com>
8745 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
8746 that takes the outer and inner modes.
8747 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
8748 comparison as the canonical test for a paradoxical subreg.
8749 * combine.c (simplify_set): Use paradoxical_subreg_p.
8750 (make_extraction): Likewise.
8751 (force_to_mode): Likewise.
8752 (rtx_equal_for_field_assignment_p): Likewise.
8753 (gen_lowpart_for_combine): Likewise.
8754 (simplify_comparison): Likewise.
8755 * cse.c (equiv_constant): Likewise.
8756 * expmed.c (store_bit_field_1): Likewise.
8757 * final.c (alter_subreg): Likewise.
8758 * fwprop.c (propagate_rtx): Likewise.
8759 (forward_propagate_subreg): Likewise.
8760 * ira-conflicts.c (ira_build_conflicts): Likewise.
8761 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
8762 * lra-constraints.c (curr_insn_transform): Likewise.
8763 (split_reg): Likewise.
8764 * lra-eliminations.c (move_plus_up): Likewise.
8765 (lra_eliminate_regs_1): Likewise.
8766 * recog.c (general_operand): Likewise.
8767 * ree.c (combine_reaching_defs): Likewise.
8768 * reload.c (push_reload): Likewise.
8769 (find_reloads): Likewise.
8770 * reload1.c (elimination_effects): Likewise.
8771 (compute_reload_subreg_offset): Likewise.
8772 (choose_reload_regs): Likewise.
8773 * rtlanal.c (subreg_lsb_1): Likewise.
8774 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8775 (simplify_subreg): Likewise.
8776 * var-tracking.c (track_loc_p): Likewise.
8777 * emit-rtl.c (byte_lowpart_offset): Likewise.
8778 (paradoxical_subreg_p): Delete out-of-line definition.
8780 2017-08-22 Jeff Law <law@redhat.com>
8782 PR tree-optimization/81741
8783 PR tree-optimization/71947
8784 * tree-ssa-dom.c: Include tree-inline.h.
8785 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
8786 equivalences if one is more expensive to compute than the other.
8787 * tree-ssa-scopedtables.h (class const_or_copies): Make
8788 record_const_or_copy_raw method private.
8789 (class avail_exprs_stack): New method simplify_binary_operation.
8790 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
8791 avail_exprs_stack::simplify_binary_operation as needed.
8792 (avail_exprs_stack::simplify_binary_operation): New function.
8794 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
8796 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
8797 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
8798 (DOT_SYMBOLS): Likewise.
8799 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
8800 (RELOCATABLE_NEEDS_FIXUP): Likewise.
8801 (RS6000_ABI_NAME): Likewise.
8802 (TARGET_CMODEL): Likewise.
8803 (TOC_SECTION_ASM_OP): Likewise.
8804 (SET_CMODEL): New macro.
8805 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
8807 2017-08-22 Richard Biener <rguenther@suse.de>
8809 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
8810 to allow for free-lang-data replacements similar to verify_type_variant.
8812 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8813 Alan Hayward <alan.hayward@arm.com>
8814 David Sherwood <david.sherwood@arm.com>
8816 * config/aarch64/aarch64.md (casesi): Use DImode rather than
8817 VOIDmode for the LABEL_REF.
8819 2017-08-22 Richard Biener <rguenther@suse.de>
8821 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
8823 2017-08-22 Richard Biener <rguenther@suse.de>
8825 * common.opt (feliminate-dwarf2-dups): Ignore.
8826 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
8827 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
8828 same_die_p_wrap, compute_section_prefix,
8829 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
8830 (comdat_symbol_id, comdat_symbol_number): Likewise.
8831 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
8833 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
8834 (output_die): Mark unreachable path unreachable.
8835 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
8836 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
8837 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
8838 (dwarf2out_early_finish): Likewise.
8840 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
8842 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
8844 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
8847 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
8848 not VAR_P. Filter attribute warnings with OPT_Wattributes.
8849 (avr_attribute_table) <io, io_low, address>: Initialize
8850 .decl_required with true.
8852 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
8854 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
8855 undocumented debugging options.
8856 (-mvsx-scalar-double): Likewise.
8857 (-mallow-df-permute): Likewise.
8858 (-mvectorize-builtins): Likewise.
8859 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8860 (rs6000_builtin_vectorized_function): Likewise.
8861 (rs6000_builtin_md_vectorized_function): Likewise.
8862 (rs6000_opt_vars): Likewise.
8864 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
8867 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
8868 (*btrq_imm): Rename from *btrq.
8869 (*btcq_imm): Rename from *btcq.
8870 (btsc): New code attribute.
8871 (*<btsc><mode>): New insn pattern.
8872 (*btr<mode>): Ditto.
8873 (*<btsc><mode>_mask): New insn_and_split pattern.
8874 (*btr<mode>_mask): Ditto.
8876 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8877 Alan Hayward <alan.hayward@arm.com>
8878 David Sherwood <david.sherwood@arm.com>
8880 * function.c (pad_below): Simplify padding calculation.
8882 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8883 Alan Hayward <alan.hayward@arm.com>
8884 David Sherwood <david.sherwood@arm.com>
8886 * target.def (function_prologue): Remove frame size argument.
8887 (function_epilogue): Likewise.
8888 * doc/tm.texi: Regenerate.
8889 * final.c (final_start_function): Update call to function_prologue.
8890 (final_end_function): Update call to function_epilogue.
8891 (default_function_pro_epilogue): Remove frame size argument.
8892 * output.h (default_function_pro_epilogue): Likewise.
8893 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
8894 (arm_output_function_prologue): Likewise.
8895 * config/frv/frv.c (frv_function_prologue): Likewise.
8896 (frv_function_epilogue): Likewise.
8897 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
8898 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
8899 (ia64_output_function_epilogue): Likewise.
8900 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
8901 (m32r_output_function_epilogue): Likewise.
8902 * config/microblaze/microblaze.c (microblaze_function_prologue)
8903 (microblaze_function_epilogue): Likewise.
8904 * config/mips/mips.c (mips_output_function_prologue): Likewise.
8905 (mips_output_function_epilogue): Likewise.
8906 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
8907 (mmix_target_asm_function_epilogue): Likewise.
8908 * config/msp430/msp430.c (msp430_start_function): Likewise.
8909 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
8910 (nds32_asm_function_epilogue): Likewise.
8911 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
8912 * config/pa/pa.c (pa_output_function_prologue): Likewise.
8913 (pa_output_function_epilogue): Likewise.
8914 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
8915 (rs6000_output_function_epilogue): Likewise.
8916 * config/rl78/rl78.c (rl78_start_function): Likewise.
8917 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
8918 (rs6000_output_function_epilogue): Likewise.
8919 * config/rx/rx.c (rx_output_function_prologue): Likewise.
8920 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
8921 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
8922 (sparc_asm_function_epilogue): Likewise.
8924 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8926 * tree.h (type_has_mode_precision_p): New function.
8927 * convert.c (convert_to_integer_1): Use it.
8928 * expr.c (expand_expr_real_2): Likewise.
8929 (expand_expr_real_1): Likewise.
8930 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
8931 * match.pd: Likewise.
8932 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
8933 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
8934 * tree-tailcall.c (process_assignment): Likewise.
8935 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8936 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
8937 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
8938 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8939 (vectorizable_assignment): Likewise.
8940 (vectorizable_shift): Likewise.
8941 (vectorizable_operation): Likewise.
8942 * tree-vrp.c (register_edge_assert_for_2): Likewise.
8944 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
8946 * match.pd: Add pow (C, x) simplification.
8948 2017-08-21 Richard Biener <rguenther@suse.de>
8950 PR tree-optimization/81900
8951 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
8952 for blocks with abnormal predecessors.
8953 (compute_antic): Do not set visited flag prematurely.
8955 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
8958 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
8960 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8962 * stor-layout.h (vector_type_mode): Move to...
8963 * tree.h (vector_type_mode): ...here.
8964 * stor-layout.c (vector_type_mode): Move to...
8965 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
8967 2017-08-21 Richard Biener <rguenther@suse.de>
8969 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
8970 register_external_die hooks.
8971 (debug_false_tree_charstarstar_uhwistar): Declare.
8972 (debug_nothing_tree_charstar_uhwi): Likewise.
8973 * debug.c (do_nothing_debug_hooks): Adjust.
8974 (debug_false_tree_charstarstar_uhwistar): New do nothing.
8975 (debug_nothing_tree_charstar_uhwi): Likewise.
8976 * dbxout.c (dbx_debug_hooks): Adjust.
8977 (xcoff_debug_hooks): Likewise.
8978 * sdbout.c (sdb_debug_hooks): Likewise.
8979 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8980 * dwarf2out.c (macinfo_label_base): New global.
8981 (dwarf2out_register_external_die): New function for the
8982 register_external_die hook.
8983 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
8984 (dwarf2_debug_hooks): Use them.
8985 (dwarf2_lineno_debug_hooks): Adjust.
8986 (struct die_struct): Add with_offset flag.
8987 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
8988 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
8989 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
8990 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
8991 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
8992 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
8993 defining section names for the early LTO debug variants.
8994 (reset_indirect_string): New helper.
8995 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
8996 (print_dw_val): Add support for offsetted symbol references.
8997 (get_ultimate_context): Split out from is_cxx.
8998 (is_cxx): Use get_ultimate_context.
8999 (is_fortran): Add decl overload.
9000 (compute_comp_unit_symbol): Split out worker from
9001 compute_section_prefix.
9002 (compute_section_prefix): Call compute_comp_unit_symbol and
9003 set comdat_type_p here.
9004 (output_die): Skip DIE symbol output for the LTO added one.
9005 Handle DIE symbol references with offset.
9006 (output_comp_unit): Guard section name mangling properly.
9007 For LTO debug sections emit a symbol at the section beginning
9008 which we use to refer to its DIEs.
9009 (add_abstract_origin_attribute): For DIEs registered via
9010 dwarf2out_register_external_die directly refer to the early
9011 DIE rather than indirectly through the shadow one we created.
9012 Remove obsolete call to dwarf2out_abstract_function for
9013 non-function/block origins.
9014 (gen_array_type_die): When generating early LTO debug do
9015 not emit DW_AT_string_length.
9016 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
9017 late when in LTO. As suggested place a gcc_unreachable for
9018 the DECL_ABSTRACT_P case.
9019 (gen_subprogram_die): Avoid another specification DIE
9020 for early built declarations/definitions for the late LTO case.
9021 (gen_variable_die): Add type references for late duplicated VLA dies
9023 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
9024 we have the abstract instance already.
9025 (process_scope_var): Adjust decl DIE contexts in LTO which
9026 first puts them in limbo.
9027 (gen_decl_die): Do not generate type DIEs late apart from
9028 types for VLAs or for decls we do not yet have a DIE. Do not
9029 call dwarf2out_abstract_function late.
9030 (dwarf2out_early_global_decl): Make sure to create DIEs
9031 for abstract instances of a decl first.
9032 (dwarf2out_late_global_decl): Adjust comment.
9033 (output_macinfo_op): With multiple macro sections use
9034 macinfo_label_base to distinguish labels.
9035 (output_macinfo): Likewise. Update macinfo_label_base.
9036 Pass in the line info label.
9037 (note_variable_value_in_expr): When generating LTO resolve
9038 all variable values here by generating DIEs as needed.
9039 (init_sections_and_labels): Add early LTO debug flag parameter
9040 and generate different sections and names if set. Add generation
9041 counter for the labels so we can have multiple of them.
9042 (reset_dies): Helper to allow DIEs to be output multiple times.
9043 (dwarf2out_finish): When outputting DIEs to the fat part of an
9044 LTO object first reset DIEs.
9045 (dwarf2out_early_finish): Output early DIEs when generating LTO.
9046 (modified_type_die): Check for decl_ultimate_origin being self
9048 (gen_type_die_with_usage): Likewise.
9049 (gen_typedef_die): Allow decl_ultimate_origin being self.
9050 (set_decl_abstract_flags): Remove.
9051 (set_block_abstract_flags): Likewise.
9052 (dwarf2out_abstract_function): Treat the early generated DIEs
9053 as the abstract copy and only add DW_AT_inline and
9054 DW_AT_artificial here and call set_decl_origin_self.
9055 If the DIE has an abstract origin don't do anything.
9056 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
9057 if we have none yet (Go fails to build one, PR78628).
9058 (variably_modified_type_p): Prevent endless recursion for Ada
9059 cyclic pointer types.
9060 * lto-streamer-in.c: Include debug.h.
9061 (dref_queue): New global.
9062 (lto_read_tree_1): Stream in DIE references.
9063 (lto_input_tree): Register DIE references.
9064 (input_function): Stream DECL_DEBUG_ARGS.
9065 * lto-streamer-out.c: Include debug.h.
9066 (lto_write_tree_1): Output DIE references.
9067 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
9068 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
9069 (output_function): Stream DECL_DEBUG_ARGS.
9070 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
9071 Stream DECL_ABSTRACT_ORIGIN.
9072 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
9073 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
9074 DECL_CONTEXT for file-scope decls.
9075 * lto-streamer.h (struct dref_entry): Declare.
9076 (dref_queue): Likewise.
9077 * cfgexpand.c (pass_expand::execute): Do not call the
9078 outlining_inline_function hook here.
9079 * lto-wrapper.c (debug_obj): New global.
9080 (tool_cleanup): Unlink it if required.
9081 (debug_objcopy): New function.
9082 (run_gcc): Handle early debug sections in the IL files by
9083 extracting them to separate files, partially linkin them and
9084 feeding the result back as result to the linker.
9085 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
9086 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
9087 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
9088 sections into a separate segment.
9089 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
9091 (darwin_asm_dwarf_section): Likewise.
9092 (darwin_asm_output_dwarf_offset): Likewise.
9093 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
9095 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9096 Alan Hayward <alan.hayward@arm.com>
9097 David Sherwood <david.sherwood@arm.com>
9099 * read-md.h (md_reader::record_potential_iterator_use): Replace
9100 pointer argument with an rtx and an index.
9101 * read-rtl.c (iterator_group::apply_iterator): Likewise.
9102 (apply_mode_iterator): Likewise.
9103 (apply_code_iterator): Likewise.
9104 (apply_int_iterator): Likewise.
9105 (apply_subst_iterator): Likewise.
9106 (record_iterator_use): Likewise.
9107 (record_attribute_use): Likewise.
9108 (md_reader::record_potential_iterator_use): Likewise. Update calls
9109 to record_iterator_use and apply_iterator.
9110 (iterator_use): Replace ptr with x and index.
9111 (attribute_use): Likewise.
9112 (apply_attribute_uses): Update calls to apply_iterator.
9113 (apply_iterators): Likewise. Update initialization of iterator_use.
9114 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
9115 and record_potential_iterator_use.
9116 (rtx_reader::read_rtx_operand): Likewise.
9118 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9119 Alan Hayward <alan.hayward@arm.com>
9120 David Sherwood <david.sherwood@arm.com>
9122 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
9125 2017-08-21 Richard Biener <rguenther@suse.de>
9128 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
9129 at struct end conservatively when comparing common bases.
9131 2017-08-21 Richard Biener <rguenther@suse.de>
9133 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
9134 (mem_ref_in_stmt): Remove.
9135 (determine_max_movement): Use ref index to get at the reference.
9136 (invariantness_dom_walker::before_dom_children): Deal with
9137 lim data already initialized.
9138 (gather_mem_refs_stmt): Initialize lim data and record ref index.
9140 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9142 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
9143 (TARGET_ISA_ROUND): Ditto.
9144 (TARGET_ROUND): Ditto.
9145 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
9146 * config/i386/i386.md: Ditto.
9147 * config/i386/sse.md: Ditto.
9148 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
9149 with OPTION_MASK_ISA_SSE4_1.
9151 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9154 * doc/extend.texi (x86 Built-in Functions): Correct the name of
9155 __builtin_ia32_lzcnt_u16.
9157 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9160 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
9161 (rs6000_set_current_function): Rewrite function to use it.
9163 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
9166 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
9167 and TYPE_WARN_IF_NOT_ALIGN.
9168 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
9169 (handle_warn_if_not_align): New.
9170 (place_union_field): Call handle_warn_if_not_align.
9171 (place_field): Call handle_warn_if_not_align.
9172 Copy TYPE_WARN_IF_NOT_ALIGN.
9173 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
9174 (layout_type): Likewise.
9175 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
9177 (tree_decl_common): Add warn_if_not_align.
9178 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
9179 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
9180 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
9181 (DECL_WARN_IF_NOT_ALIGN): Likewise.
9182 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
9183 * doc/extend.texi: Document warn_if_not_aligned attribute.
9184 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
9186 2017-08-17 Martin Liska <mliska@suse.cz>
9189 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
9190 (get_data_dependence): Use it as pointer type.
9191 (distribute_loop): Likewise.
9193 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9195 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
9196 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
9197 (p8_vmrgow_<mode>_direct): New define_insn.
9198 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
9199 handle endianness for vmrgew and vmrgow permute patterns.
9201 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9203 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
9204 * config/rs6000/rs6000-cpus.def: Remove comment.
9205 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
9206 (POWERPC_MASKS): Likewise.
9207 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
9208 use of TARGET_VSX_TIMODE.
9209 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
9210 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
9211 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
9212 (rs6000_option_override_internal): Remove dead code.
9213 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
9214 (rs6000_legitimize_reload_address): Likewise.
9215 (rs6000_legitimate_address_p): Likewise.
9216 (rs6000_opt_masks): Delete "vsx-timode".
9217 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
9218 from function comment.
9219 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
9220 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
9221 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
9223 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
9224 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
9225 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
9228 (define_peephole2): Likewise.
9230 2017-08-17 Martin Sebor <msebor@redhat.com>
9233 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
9234 past the end of an array.
9235 (test_pp_format): Add test cases.
9237 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
9239 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
9240 missing ECF_NOTHROW flags.
9242 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9245 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
9246 operands residing in integer registers.
9247 (*vsx_le_perm_load_<mode>): Likewise.
9248 (*vsx_le_perm_store_<mode>): Likewise.
9249 (define_peephole2): Add peepholes to optimize the above.
9251 2017-08-17 Marek Polacek <polacek@redhat.com>
9254 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
9255 to mimic what shorten_compare did. Change the return type to bool.
9256 (fold_cond_expr_with_comparison): Update call to
9257 operand_equal_for_comparison_p.
9258 (fold_ternary_loc): Likewise.
9260 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
9262 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
9264 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
9266 2017-08-17 Richard Biener <rguenther@suse.de>
9268 * tree-ssa-structalias.c (solve_graph): When propagating
9269 to successors update the graphs succ edges and avoid duplicate work.
9271 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
9274 * config/i386/i386.c (ix86_option_override_internal): Save target
9275 specific options after ix86_stack_protector_guard_reg was changed.
9277 2017-08-17 Richard Biener <rguenther@suse.de>
9279 PR tree-optimization/81827
9280 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
9281 (new_var_info): Initialize it conservatively.
9282 (get_call_vi): Mark register vars.
9283 (new_scalar_tmp_constraint_exp): Likewise.
9284 (handle_rhs_call): Likewise.
9285 (handle_const_call): Likewise.
9286 (create_function_info_for): Likewise.
9287 (solve_constraints): Sort varinfos to separate register from
9288 non-register vars to pack points-to solution bitmaps during
9291 2017-08-17 Marek Polacek <polacek@redhat.com>
9293 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
9295 2017-08-17 Richard Biener <rguenther@suse.de>
9297 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
9298 to true when overflow is undefined and we saturated the result.
9300 2017-08-17 Alan Modra <amodra@gmail.com>
9303 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
9304 Don't use store multiple if only one reg needs saving.
9305 (interesting_frame_related_regno): New function.
9306 (rs6000_frame_related): Don't emit frame info for regs that
9308 (rs6000_emit_epilogue): Likewise.
9310 2017-08-16 Nathan Sidwell <nathan@acm.org>
9312 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
9313 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
9314 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
9315 * tree.c (free_lang_data_in_type): Use else-if chain. Always
9316 clear TYPE_LANG_1. Remove obsolete member-function stripping.
9317 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
9318 (verify_type): Adjust for TYPE_BINFO move.
9319 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
9320 process TYPE_BINFO directly.
9321 (hash_tree): Likewise.
9322 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9324 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9327 2017-08-16 David Malcolm <dmalcolm@redhat.com>
9329 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
9331 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
9334 * config/i386/i386.md (*anddi_1_btr): Change predicates of
9335 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
9336 Add ix86_binary_operator_ok to insn constraint.
9337 (*iordi_1_bts): Ditto.
9338 (*xordi_1_btc): Ditto.
9339 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
9340 Update corresponding peephole2 pattern.
9344 2017-08-16 Bin Cheng <bin.cheng@arm.com>
9346 PR tree-optimization/81832
9347 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
9348 copy loop header which has IFN_LOOP_DIST_ALIAS call.
9350 2017-08-16 Marek Polacek <polacek@redhat.com>
9353 * fold-const.c (fold_indirect_ref_1): Restore original behavior
9354 regarding size_zero_node.
9356 2017-08-16 Martin Liska <mliska@suse.cz>
9359 * config.gcc: Respect previously set extra_objs in case
9362 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
9364 PR tree-optimization/81835
9365 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
9366 the phi SSA_NAME. Check that the condition in a COND_EXPR does
9367 not depend on the phi.
9369 2017-08-16 Alan Modra <amodra@gmail.com>
9371 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
9374 2017-08-16 Alan Modra <amodra@gmail.com>
9376 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
9377 (save_reg_p): ..into this. Update all callers.
9378 (first_reg_to_save): Simplify.
9380 2017-08-16 Alan Modra <amodra@gmail.com>
9382 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
9385 2017-08-15 Joseph Myers <joseph@codesourcery.com>
9389 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
9390 constant count if that count is less than 32.
9392 2017-08-15 Nathan Sidwell <nathan@acm.org>
9394 * gcc.c (execute): Emit friendlier message if inferior is killed
9395 by an external cause.
9397 2017-08-15 Richard Biener <rguenther@suse.de>
9399 PR tree-optimization/81790
9400 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
9401 CONSTRUCTORs from simplifying and VN.
9403 2017-08-14 Martin Sebor <msebor@redhat.com>
9405 * builtin-attrs.def: Add comments.
9407 2017-08-14 Martin Sebor <msebor@redhat.com>
9410 * doc/extend.texi (attribute nonstring): Document new attribute.
9412 2017-08-14 Martin Sebor <msebor@redhat.com>
9415 * tree-diagnostic.c (default_tree_printer): Handle %G.
9416 * gimple-pretty-print.h (percent_G_format): Declare new function.
9417 * gimple-pretty-print.c (percent_G_format): Define.
9418 * tree-pretty-print.c (percent_K_format): Add argument.
9420 2017-08-14 Martin Sebor <msebor@redhat.com>
9422 PR translation/79998
9423 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
9424 Remove a stray space.
9426 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
9429 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
9430 (*iordi_1_bts): Ditto.
9431 (*xordi_1_btc): Ditto.
9433 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9436 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
9437 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9439 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
9440 quoted strings, and make more translator-friendly.
9441 (darwin_rs6000_override_options): Likewise.
9442 (rs6000_option_override_internal): Likewise.
9443 (rs6000_return_in_memory): Fix overlong line.
9444 (init_cmulative_args): Use quoted strings, and make more
9445 translator-friendly.
9446 (rs6000_pass_by_reference): Fix overlong line.
9447 (def_builtin): Use quoted strings.
9448 (altivec_expand_predicate_builtin): Use quoted strings, and make
9449 more translator-friendly.
9450 (htm_expand_builtin): Use quoted strings.
9451 (cpu_expand_builtin): Use quoted strings, and make more
9452 translator-friendly.
9453 (altivec_expand_builtin): Likewise.
9454 (paired_expand_predicate_builtin): Likewise.
9455 (rs6000_invalid_builtin): Likewise.
9456 (builtin_function_type): Use quoted strings.
9457 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
9458 more translator-friendly.
9459 (rs6000_trampoline_init): Likewise.
9460 (rs6000_handle_altivec_attribute): Likewise.
9461 (rs6000_inner_target_options): Use quoted strings.
9462 (rs6000_disable_incompatible_switches): Likewise.
9463 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
9464 strings, and make more translator-friendly.
9465 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
9467 2017-08-14 Bin Cheng <bin.cheng@arm.com>
9469 PR tree-optimization/81799
9470 * tree-loop-distribution.c (version_loop_by_alias_check): Force
9471 cond_expr to simple gimple operand.
9473 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
9476 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
9478 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
9482 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
9483 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
9484 TARGET_GASISR_PROLOGUES.
9485 * config/avr/avr.c (avr_option_override): Same.
9486 (avr_pass_pre_proep::execute): Same.
9488 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
9491 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
9492 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
9494 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
9496 * config/i386/i386.md (*load_tp_<mode>): Redefine as
9497 define_insn_and_split. Split to a memory load from 0 in
9498 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
9499 using PTR mode iterator.
9500 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
9501 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
9502 (*add_tp_<mode>): Redefine as define_insn_and_split.
9503 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
9504 address space. Merge with *add_tp_x32 using PTR mode iterator.
9505 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
9506 Split to an add with a memory load from 0 in
9507 DEFAULT_TLS_SEG_REG address space.
9509 2017-08-12 Andrew Pinski <apinski@cavium.com>
9511 * config/aarch64/aarch64-option-extensions.def (rdma):
9512 Fix feature string to what Linux prints out in /proc/cpuinfo.
9514 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
9517 * dwarf2out.c (modified_type_die): For C typedef types that have
9518 an ultimate origin, process the ultimate origin instead of the
9520 (gen_typedef_die): Assert that input DECLs have no ultimate
9522 (gen_type_die_with_usage): For typedef variants that have an
9523 ultimate origin, just call gen_decl_die on the original DECL.
9524 (process_scope_var): Avoid creating DIEs for local typedefs and
9525 concrete static variables.
9527 2017-08-12 Alan Modra <amodra@gmail.com>
9531 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
9532 match gnu-user.h startfile.
9533 (ENDFILE_LINUX_SPEC): Similarly.
9535 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
9538 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
9540 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
9542 2017-08-11 Tamar Christina <tamar.christina@arm.com>
9543 * config/aarch64/aarch64.md (mov<mode>): Change.
9544 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9545 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
9546 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
9548 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
9550 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
9551 for storage order barriers.
9553 2017-08-11 Martin Liska <mliska@suse.cz>
9556 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
9557 variables of void type.
9559 2017-08-11 Martin Liska <mliska@suse.cz>
9561 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
9562 TARGET_SUPPORTS_ALIASES.
9563 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
9564 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
9565 (optimize_weakref): Likewise.
9566 * symtab.c (symtab_node::noninterposable_alias): Likewise.
9567 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
9568 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
9570 2017-08-11 Martin Liska <mliska@suse.cz>
9573 * config/i386/i386.c (make_resolver_func): Do complete
9574 refactoring of the function.
9576 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9579 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
9580 * config/i386/i386.c (ix86_stack_protect_guard): Use
9581 ix86_stack_protect_guard_symbol_str to generate varible declaration.
9582 * doc/invoke.texi (x86 Options): Document
9583 -mstack-protector-guard-symbol= option.
9585 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9587 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
9588 * config/i386/i386.c (ix86_split_stack_guard): New function.
9589 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
9590 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
9591 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
9592 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
9593 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
9594 (split_stack_space_check): Call ix86_split_stack_guard.
9596 2017-08-10 Martin Sebor <msebor@redhat.com>
9598 * print-tree.c (print_node): Print location using the established
9600 Replace spaces with colons.
9601 (debug_raw, debug): Ditto.
9603 2017-08-10 Martin Sebor <msebor@redhat.com>
9606 * pretty-print.c (pp_format): Correct the handling of %s precision.
9608 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
9611 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9613 (ix86_finalize_stack_frame_flags): This. Also clear
9614 frame_pointer_needed if -fno-omit-frame-pointer is used without
9616 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9617 with ix86_finalize_stack_frame_flags.
9618 (ix86_expand_epilogue): Likewise.
9619 (ix86_expand_split_stack_prologue): Likewise.
9620 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
9622 2017-08-10 Martin Liska <mliska@suse.cz>
9625 * c-attribs.c (handle_target_attribute):
9626 Report warning for an empty string argument of target attribute.
9628 2017-08-09 Jakub Jelinek <jakub@redhat.com>
9631 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
9633 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
9634 or DECL_NONLOCAL labels.
9635 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
9636 or DECL_NONLOCAL labels here.
9638 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
9640 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
9641 to indicate when early gimple folding has been disabled.
9642 (rs6000_gimple_fold_builtin): Add debug content.
9643 (rs6000_invalid_builtin): Fix whitespace.
9644 (rs6000_expand_builtin): Fix whitespace.
9645 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
9647 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
9650 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
9651 SAVE_MULTIPLE if not all the registers that saves, should be saved.
9653 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
9655 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
9656 (qdf24xx): Likewise.
9657 * config/aarch64/aarch64.md: Include falkor.md.
9658 * config/aarch64/falkor.md: New.
9660 2017-08-09 Marek Polacek <polacek@redhat.com>
9663 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
9664 * diagnostic.c (emit_diagnostic): Add a comment.
9665 (emit_diagnostic_valist): New function.
9667 2017-08-09 Marek Polacek <polacek@redhat.com>
9670 * input.c (make_location): New overload.
9671 * input.h (make_location): Declare.
9673 2017-08-08 Alan Modra <amodra@gmail.com>
9674 H.J. Lu <hongjiu.lu@intel.com>
9677 * gcc.c (NO_PIE_SPEC): Delete.
9678 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
9680 (LINK_PIE_SPEC): ..to here.
9681 (LINK_COMMAND_SPEC): Support -no-pie.
9682 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
9683 chain of crtbegin*.o selection, update for PIE_SPEC changes and
9685 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
9686 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
9687 (ENDFILE_CRTEND_SPEC): Similarly.
9689 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
9692 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
9693 (mstack-protector-guard-offset=): Ditto.
9694 * config/i386/i386.c (ix86_option_override): Handle
9695 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
9697 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
9698 ix86_stack_protect_guard_offset variables.
9699 (TARGET_STACK_PROTECT_GUARD): Always define.
9700 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
9701 and -mstack-protector-guard-offset= options.
9703 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9705 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
9706 boundary case for the last candidate.
9708 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9710 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
9711 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
9713 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9716 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
9717 * config/aarch64/aarch64-builtins.c
9718 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
9719 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
9720 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
9722 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9723 Andrew Pinski <pinskia@gmail.com>
9726 * internal-fn.def (XORSIGN): New.
9727 * optabs.def (xorsign_optab): New.
9728 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
9729 (convert_expand_mult_copysign): New.
9730 (pass_optimize_widening_mul::execute): Call
9731 convert_expand_mult_copysign.
9733 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9735 PR tree-optimization/81354
9736 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
9737 Insert on edges rather than explicitly creating landing pads.
9738 (analyze_candidates_and_replace): Commit edge inserts.
9740 2017-08-08 Richard Biener <rguenther@suse.de>
9743 * tree-ssa-loop-niter.c: Include tree-dfa.h.
9744 (expand_simple_operations): Also look through ADDR_EXPRs with
9745 MEM_REF bases treating them as POINTER_PLUS_EXPR.
9747 2017-08-08 Richard Biener <rguenther@suse.de>
9749 PR tree-optimization/81723
9750 * tree-vect-slp.c (struct bst_traits): New hash traits.
9751 (bst_fail): New global.
9752 (vect_build_slp_tree_2): New worker, split out from ...
9753 (vect_build_slp_tree): ... this now wrapping it with using
9754 bst_fail set to cache SLP tree build fails. Properly handle
9756 (vect_analyze_slp_instance): Allocate and free bst_fail.
9758 2017-08-08 Martin Liska <mliska@suse.cz>
9761 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
9762 LABEL_DECLs that can be from a different function.
9764 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9766 PR tree-optimization/81744
9767 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
9768 loop's number of iterations.
9770 2017-08-08 Martin Liska <mliska@suse.cz>
9772 * asan.c: Include header files.
9773 * attribs.c (build_decl_attribute_variant): New function moved
9775 (build_type_attribute_qual_variant): Likewise.
9776 (cmp_attrib_identifiers): Likewise.
9777 (simple_cst_list_equal): Likewise.
9778 (omp_declare_simd_clauses_equal): Likewise.
9779 (attribute_value_equal): Likewise.
9780 (comp_type_attributes): Likewise.
9781 (build_type_attribute_variant): Likewise.
9782 (lookup_ident_attribute): Likewise.
9783 (remove_attribute): Likewise.
9784 (merge_attributes): Likewise.
9785 (merge_type_attributes): Likewise.
9786 (merge_decl_attributes): Likewise.
9787 (merge_dllimport_decl_attributes): Likewise.
9788 (handle_dll_attribute): Likewise.
9789 (attribute_list_equal): Likewise.
9790 (attribute_list_contained): Likewise.
9791 * attribs.h (lookup_attribute): New function moved from tree.[ch].
9792 (lookup_attribute_by_prefix): Likewise.
9793 * bb-reorder.c: Include header files.
9794 * builtins.c: Likewise.
9795 * calls.c: Likewise.
9796 * cfgexpand.c: Likewise.
9797 * cgraph.c: Likewise.
9798 * cgraphunit.c: Likewise.
9799 * convert.c: Likewise.
9800 * dwarf2out.c: Likewise.
9801 * final.c: Likewise.
9802 * fold-const.c: Likewise.
9803 * function.c: Likewise.
9804 * gimple-expr.c: Likewise.
9805 * gimple-fold.c: Likewise.
9806 * gimple-pretty-print.c: Likewise.
9807 * gimple.c: Likewise.
9808 * gimplify.c: Likewise.
9809 * hsa-common.c: Likewise.
9810 * hsa-gen.c: Likewise.
9811 * internal-fn.c: Likewise.
9812 * ipa-chkp.c: Likewise.
9813 * ipa-cp.c: Likewise.
9814 * ipa-devirt.c: Likewise.
9815 * ipa-fnsummary.c: Likewise.
9816 * ipa-inline.c: Likewise.
9817 * ipa-visibility.c: Likewise.
9819 * lto-cgraph.c: Likewise.
9820 * omp-expand.c: Likewise.
9821 * omp-general.c: Likewise.
9822 * omp-low.c: Likewise.
9823 * omp-offload.c: Likewise.
9824 * omp-simd-clone.c: Likewise.
9825 * opts-global.c: Likewise.
9826 * passes.c: Likewise.
9827 * predict.c: Likewise.
9828 * sancov.c: Likewise.
9829 * sanopt.c: Likewise.
9830 * symtab.c: Likewise.
9831 * toplev.c: Likewise.
9832 * trans-mem.c: Likewise.
9833 * tree-chkp.c: Likewise.
9834 * tree-eh.c: Likewise.
9835 * tree-into-ssa.c: Likewise.
9836 * tree-object-size.c: Likewise.
9837 * tree-parloops.c: Likewise.
9838 * tree-profile.c: Likewise.
9839 * tree-ssa-ccp.c: Likewise.
9840 * tree-ssa-live.c: Likewise.
9841 * tree-ssa-loop.c: Likewise.
9842 * tree-ssa-sccvn.c: Likewise.
9843 * tree-ssa-structalias.c: Likewise.
9844 * tree-ssa.c: Likewise.
9845 * tree-streamer-in.c: Likewise.
9846 * tree-vectorizer.c: Likewise.
9847 * tree-vrp.c: Likewise.
9849 * ubsan.c: Likewise.
9850 * varasm.c: Likewise.
9851 * varpool.c: Likewise.
9852 * tree.c: Remove functions moved to attribs.[ch].
9854 * config/aarch64/aarch64.c: Add attrs.h header file.
9855 * config/alpha/alpha.c: Likewise.
9856 * config/arc/arc.c: Likewise.
9857 * config/arm/arm.c: Likewise.
9858 * config/avr/avr.c: Likewise.
9859 * config/bfin/bfin.c: Likewise.
9860 * config/c6x/c6x.c: Likewise.
9861 * config/cr16/cr16.c: Likewise.
9862 * config/cris/cris.c: Likewise.
9863 * config/darwin.c: Likewise.
9864 * config/epiphany/epiphany.c: Likewise.
9865 * config/fr30/fr30.c: Likewise.
9866 * config/frv/frv.c: Likewise.
9867 * config/ft32/ft32.c: Likewise.
9868 * config/h8300/h8300.c: Likewise.
9869 * config/i386/winnt.c: Likewise.
9870 * config/ia64/ia64.c: Likewise.
9871 * config/iq2000/iq2000.c: Likewise.
9872 * config/lm32/lm32.c: Likewise.
9873 * config/m32c/m32c.c: Likewise.
9874 * config/m32r/m32r.c: Likewise.
9875 * config/m68k/m68k.c: Likewise.
9876 * config/mcore/mcore.c: Likewise.
9877 * config/microblaze/microblaze.c: Likewise.
9878 * config/mips/mips.c: Likewise.
9879 * config/mmix/mmix.c: Likewise.
9880 * config/mn10300/mn10300.c: Likewise.
9881 * config/moxie/moxie.c: Likewise.
9882 * config/msp430/msp430.c: Likewise.
9883 * config/nds32/nds32-isr.c: Likewise.
9884 * config/nds32/nds32.c: Likewise.
9885 * config/nios2/nios2.c: Likewise.
9886 * config/nvptx/nvptx.c: Likewise.
9887 * config/pa/pa.c: Likewise.
9888 * config/pdp11/pdp11.c: Likewise.
9889 * config/powerpcspe/powerpcspe.c: Likewise.
9890 * config/riscv/riscv.c: Likewise.
9891 * config/rl78/rl78.c: Likewise.
9892 * config/rx/rx.c: Likewise.
9893 * config/s390/s390.c: Likewise.
9894 * config/sh/sh.c: Likewise.
9895 * config/sol2.c: Likewise.
9896 * config/sparc/sparc.c: Likewise.
9897 * config/spu/spu.c: Likewise.
9898 * config/stormy16/stormy16.c: Likewise.
9899 * config/tilegx/tilegx.c: Likewise.
9900 * config/tilepro/tilepro.c: Likewise.
9901 * config/v850/v850.c: Likewise.
9902 * config/vax/vax.c: Likewise.
9903 * config/visium/visium.c: Likewise.
9904 * config/xtensa/xtensa.c: Likewise.
9906 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9909 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
9910 constraints since the -mupper-regs-* switches have been
9912 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
9913 into a vector from a double word element that was extracted from
9914 another vector, and eliminate extra XXPERMDI instructions.
9915 (vsx_concat_<mode>_2): Likewise.
9916 (vsx_concat_<mode>_3): Likewise.
9917 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
9918 concat to allow optimizing inserts from previous extracts.
9920 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
9922 * config/i386/i386.c (ix86_stack_protect_guard): Generate
9923 memory reference to a SSP offset in TLS address space.
9924 (ix86_print_operand) <case '@'>: Remove.
9925 (ix86_print_operand_punct_valid_p): Remove '@' code.
9926 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
9928 (stack_tls_protect_set_<mode>): Remove.
9929 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
9930 (stack_tls_protect_test_<mode>): Remove.
9931 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
9933 2017-08-07 Olivier Hainque <hainque@adacore.com>
9936 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
9938 2017-08-07 Douglas Rupp <rupp@adacore.com>
9940 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
9941 variable was referenced as multidir in command.
9943 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9946 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
9949 2017-08-07 Martin Liska <mliska@suse.cz>
9951 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
9952 * config/rl78/rl78.c: Add include of attribs.h.
9953 * config/sh/sh.c: Likewise.
9954 * config/v850/v850.c: Likewise.
9956 2017-08-07 Tom de Vries <tom@codesourcery.com>
9959 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
9961 2017-08-07 Martin Liska <mliska@suse.cz>
9963 * config/mips/mips.c: Include attribs.h.
9965 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
9968 * doc/invoke.texi: Document change in behvaior for -Ofast for
9971 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
9973 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
9975 (aarch64_pop_regs): Likewise.
9976 (aarch64_gen_load_pair): Likewise.
9977 (aarch64_save_callee_saves): Likewise.
9978 (aarch64_restore_callee_saves): Likewise.
9980 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9982 * config/i386/i386.c: Revert the last change.
9984 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9987 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9989 (ix86_finalize_stack_frame_flags): This. Also clear
9990 frame_pointer_needed if -fno-omit-frame-pointer is used without
9992 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9993 with ix86_finalize_stack_frame_flags.
9994 (ix86_expand_epilogue): Likewise.
9995 (ix86_expand_split_stack_prologue): Likewise.
9997 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10000 * config/i386/i386.c (get_builtin_code_for_version): Set priority
10001 to P_AES for Westmere.
10003 2017-08-07 Jonathan Yong <10walls@gmail.com>
10005 * config/i386/mingw.opt (fset-stack-executable): Removed.
10006 * config/i386/cygming.opt (fset-stack-executable): Moved
10008 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
10010 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
10012 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
10014 2017-08-07 Marek Polacek <polacek@redhat.com>
10016 PR middle-end/81737
10017 * fold-const.c (fold_indirect_ref_1): Check type_domain.
10019 2017-08-07 Martin Liska <mliska@suse.cz>
10021 * attribs.h (canonicalize_attr_name): New function.
10022 (cmp_attribs): Move from c-format.c and adjusted.
10023 (is_attribute_p): Moved from tree.h.
10024 * tree-inline.c: Add new includes.
10025 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
10026 (private_is_attribute_p): Remove.
10027 (private_lookup_attribute): Likewise.
10028 (private_lookup_attribute_by_prefix): Simplify.
10029 (remove_attribute): Use is_attribute_p.
10030 * tree.h: Remove removed declarations.
10032 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10034 PR middle-end/81698
10035 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
10036 instead of computing it in the function. Formatting fix.
10037 (expand_case): Don't rely on default_edge being the first edge,
10038 clear it if removing it, pass default_edge to
10039 emit_case_dispatch_table.
10040 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
10043 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
10045 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
10046 insn in the function, emit NOP after the insn.
10048 2017-08-06 Tom de Vries <tom@codesourcery.com>
10050 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
10053 2017-08-06 Tom de Vries <tom@codesourcery.com>
10055 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
10058 2017-08-04 Yury Gribov <tetra2005@gmail.com>
10060 PR tree-optimization/57371
10061 * match.pd: New pattern.
10063 2017-08-04 Marek Polacek <polacek@redhat.com>
10065 PR middle-end/81695
10066 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
10067 perform the computation in offset_int.
10069 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10071 PR tree-optimization/81136
10072 * tree-vectorizer.h: Include tree-hash-traits.h.
10073 (vec_base_alignments): New typedef.
10074 (vec_info): Add a base_alignments field.
10075 (vect_record_base_alignments): Declare.
10076 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
10078 (DR_IS_CONDITIONAL_IN_STMT): New macro.
10079 (create_data_ref): Add an is_conditional_in_stmt argument.
10080 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
10081 the is_conditional_in_stmt field.
10082 (data_ref_loc): Add an is_conditional_in_stmt field.
10083 (get_references_in_stmt): Set the is_conditional_in_stmt field.
10084 (find_data_references_in_stmt): Update call to create_data_ref.
10085 (graphite_find_data_references_in_stmt): Likewise.
10086 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
10087 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
10088 (vect_record_base_alignment): New function.
10089 (vect_record_base_alignments): Likewise.
10090 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
10091 for nested statements even if we fail to compute a misalignment.
10092 Use pooled base alignments for unconditional references.
10093 (vect_find_same_alignment_drs): Compare base addresses instead
10095 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
10096 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
10098 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10100 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
10101 Add an explicit name for the enum. Use auto_vec for slp_instances
10102 and grouped_stores.
10103 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
10105 (_bb_vec_info): Add a constructor and destructor.
10106 (vinfo_for_stmt): Return NULL for uids of -1 as well.
10107 (destroy_loop_vec_info): Delete.
10108 (vect_destroy_datarefs): Likewise.
10109 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
10110 (vec_info::vec_info): New function.
10111 (vec_info::~vec_info): Likewise.
10112 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
10113 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
10114 destroy_loop_vec_info.
10115 * tree-vect-loop.c (new_loop_vec_info): Replace with...
10116 (_loop_vec_info::_loop_vec_info): ...this.
10117 (destroy_loop_vec_info): Replace with...
10118 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
10119 the stmt_vec_infos. Leave handling of vec_info information to its
10120 destructor. Remove explicit vector releases.
10121 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
10122 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
10123 * tree-vect-slp.c (new_bb_vec_info): Replace with...
10124 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
10125 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
10126 (destroy_bb_vec_info): Replace with...
10127 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
10128 information to its destructor.
10129 (vect_slp_analyze_bb_1): Use new and delete instead of
10130 new_bb_vec_info and destroy_bb_vec_info.
10131 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
10134 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10136 * tree-data-ref.h (subscript): Add access_fn field.
10137 (data_dependence_relation): Add could_be_independent_p.
10138 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
10139 (same_access_functions): Move to tree-data-ref.c.
10140 * tree-data-ref.c (ref_contains_union_access_p): New function.
10141 (access_fn_component_p): Likewise.
10142 (access_fn_components_comparable_p): Likewise.
10143 (dr_analyze_indices): Add a reference to access_fn_component_p.
10144 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
10146 (constant_access_functions): Likewise.
10147 (add_other_self_distances): Likewise.
10148 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
10149 (initialize_data_dependence_relation): Use XCNEW and remove
10150 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
10151 of access functions that have the same type. Allow the
10152 subsequence to end with different bases in some circumstances.
10153 Record the chosen access functions in SUB_ACCESS_FN.
10154 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
10155 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
10156 (subscript_dependence_tester_1): Likewise dra and drb.
10157 (build_classic_dist_vector): Update calls accordingly.
10158 (subscript_dependence_tester): Likewise.
10159 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
10160 DDR_COULD_BE_INDEPENDENT_P.
10161 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
10162 comp_alias_ddrs instead of may_alias_ddrs.
10163 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
10165 (vect_analyze_data_ref_dependence): Use it if
10166 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
10167 distance vectors if that fails.
10168 (dependence_distance_ge_vf): New function.
10169 (vect_prune_runtime_alias_test_list): Use it. Don't clear
10170 LOOP_VINFO_MAY_ALIAS_DDRS.
10172 2017-08-04 Richard Biener <rguenther@suse.de>
10174 PR middle-end/81705
10175 * fold-const.c (fold_binary_loc): Properly restrict
10176 minus_var0 && minus_var1 case when associating undefined overflow
10179 2017-08-04 Tom de Vries <tom@codesourcery.com>
10181 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
10183 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10185 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10186 Don't start diagnostic messages with a capital letter.
10187 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10189 (rs6000_invalid_builtin): Likewise.
10190 (rs6000_trampoline_init): Likewise.
10192 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10195 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
10196 after setting changeable df flags.
10198 2017-08-03 Richard Biener <rguenther@suse.de>
10200 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
10201 up if the use is in USE - X.
10203 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
10205 * toplev.c (dumpfile.h): New include.
10206 (internal_error_reentered): New static function. Use it...
10207 (internal_error_function): ...here to handle reentered internal_error.
10209 2017-08-03 Richard Biener <rguenther@suse.de>
10211 PR middle-end/81148
10212 * fold-const.c (split_tree): Add minus_var and minus_con
10213 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
10214 here but always use minus_*.
10215 (associate_trees): Assert we never associate with MINUS_EXPR
10216 and NULL first operand. Do not recurse for PLUS_EXPR operands
10217 when associating as MINUS_EXPR either.
10218 (fold_binary_loc): Track minus_var and minus_con.
10220 2017-08-03 Tom de Vries <tom@codesourcery.com>
10223 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
10224 ACCEL_COMPILER, apply finish_options on
10225 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
10227 2017-08-03 Tom de Vries <tom@codesourcery.com>
10230 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
10231 function_entry_patch_area_size > 0.
10233 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10236 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
10237 instead of 10??LU, perform unit multiplication in wide_int,
10238 don't change alloc_object_size_limit if the limit is larger
10241 PR tree-optimization/81655
10242 PR tree-optimization/81588
10243 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
10244 the case when ranges[i].low and high are 1 for unsigned type with
10247 PR middle-end/81052
10248 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
10249 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
10251 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10253 * tree-vrp.h: Add include guard.
10255 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
10258 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
10259 (ud2): New insn pattern.
10260 * config/i386/i386.c (ix86_expand_epilogue):
10261 For naked functions, generate ud2 instead of trap insn.
10263 2017-08-02 Marek Polacek <polacek@redhat.com>
10266 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
10268 2017-08-02 Tom de Vries <tom@codesourcery.com>
10269 Cesar Philippidis <cesar@codesourcery.com>
10271 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
10272 Add missing edge probabilities.
10274 2017-08-02 Tamar Christina <tamar.christina@arm.com>
10276 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
10277 Correct endianness.
10279 2017-08-02 Jakub Jelinek <jakub@redhat.com>
10281 PR middle-end/79499
10282 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
10283 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
10284 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
10285 prologue_seq sequences - if any.
10287 2017-08-02 Richard Biener <rguenther@suse.de>
10289 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
10290 via vectors if supported, integer extracts via punning if supported
10291 or otherwise vector extracts.
10293 2017-08-02 Richard Biener <rguenther@suse.de>
10295 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
10297 (bitmap_insert_into_set): ... this.
10299 2017-08-02 Richard Biener <rguenther@suse.de>
10301 PR tree-optimization/81633
10303 2015-08-17 Alan Hayward <alan.hayward@arm.com>
10305 PR tree-optimization/71752
10306 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
10308 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
10310 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
10311 (machine_function::call_ms2sysv_pad_out): Remove field.
10312 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
10313 (ix86_compute_frame_layout): Likewise.
10315 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
10318 * config/i386/i386.c (ix86_set_func_type): Disallow naked
10319 attribute with interrupt attribute.
10321 2017-08-01 Andrew Pinski <apinski@cavium.com>
10323 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
10324 BIT_INSERT_EXPR's operand 1
10325 to see if the types precision matches.
10327 2017-08-01 Martin Liska <mliska@suse.cz>
10329 PR middle-end/70140
10330 * builtins.c (expand_builtin_memcpy_args): Remove.
10331 (expand_builtin_memcpy): Call newly added function
10332 expand_builtin_memory_copy_args.
10333 (expand_builtin_memcpy_with_bounds): Likewise.
10334 (expand_builtin_mempcpy): Remove last argument.
10335 (expand_builtin_mempcpy_with_bounds): Likewise.
10336 (expand_builtin_memory_copy_args): New function created from
10337 expand_builtin_mempcpy_args with small modifications.
10338 (expand_builtin_mempcpy_args): Remove.
10339 (expand_builtin_stpcpy): Remove unused argument.
10340 (expand_builtin): Likewise.
10341 (expand_builtin_with_bounds): Likewise.
10343 2017-08-01 Martin Liska <mliska@suse.cz>
10346 Make mempcpy more optimal (PR middle-end/70140).
10348 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10351 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10352 __builtin_vec_cmpne verify both arguments are compatible vectors
10353 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
10354 verify arg1_type is a pointer or array type. For __builtin_vec_st,
10355 move computation of aligned to after checking the argument types.
10359 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10360 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
10363 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
10364 Doug Rupp <rupp@adacore.com>
10365 Olivier Hainque <hainque@adacore.com>
10367 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
10368 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
10370 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
10371 for TARGET_OS_CPP_BUILTIN.
10372 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
10373 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
10375 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
10376 passing required abi options to the assembler for EABI configurations.
10377 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
10378 of .text.hot and .text.unlikely sections for kernel modules when
10379 using ARM style exceptions.
10380 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
10381 options. Add EXTRA_CC1_SPEC.
10382 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
10384 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
10386 (ARM_TARGET2_DWARF_FORMAT): Define.
10387 * config/arm/t-vxworks: Adjust multilib control to removal of the
10388 Diab command line options.
10390 2017-08-01 Martin Liska <mliska@suse.cz>
10392 PR gcov-profile/81561
10393 * gcov.c (unblock): Make unblocking safe as we need to preserve
10394 index correspondence of blocks and block_lists.
10396 2017-08-01 Richard Biener <rguenther@suse.de>
10398 PR tree-optimization/81181
10399 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10400 (compute_antic): ... end of iteration here.
10402 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
10404 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
10405 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
10406 (ftree-slp-vectorize): Likewise.
10407 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
10408 can no longer be set independent of flag_tree_loop_vectorize.
10409 * omp-general.c (emp_max_vf): Likewise.
10410 * opts.c (enable_fdo_optimizations): Remove references to
10411 flag_tree_vectorize, these are now implicit.
10412 (common_handle_option): Remove handling for OPT_ftree_vectorize,
10413 and leave it for the options machinery.
10415 2017-08-01 Martin Liska <mliska@suse.cz>
10417 PR middle-end/70140
10418 * builtins.c (expand_builtin_memcpy_args): Remove.
10419 (expand_builtin_memcpy): Call newly added function
10420 expand_builtin_memory_copy_args.
10421 (expand_builtin_memcpy_with_bounds): Likewise.
10422 (expand_builtin_mempcpy): Remove last argument.
10423 (expand_builtin_mempcpy_with_bounds): Likewise.
10424 (expand_builtin_memory_copy_args): New function created from
10425 expand_builtin_mempcpy_args with small modifications.
10426 (expand_builtin_mempcpy_args): Remove.
10427 (expand_builtin_stpcpy): Remove unused argument.
10428 (expand_builtin): Likewise.
10429 (expand_builtin_with_bounds): Likewise.
10431 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10434 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
10435 print "ds:" only for immediates in generic address space.
10437 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10440 * config/i386/i386.c (ix86_funciton_naked): New prototype.
10441 (ix86_function_ok_for_sibcall): Return false for naked functions.
10443 2017-08-01 Richard Biener <rguenther@suse.de>
10445 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
10446 (compute_antic): Seed worklist with exit block predecessors.
10447 * cfganal.c (dfs_find_deadend): For a cycle return the source
10448 of the edge closing it.
10450 2017-08-01 Tamar Christina <tamar.christina@arm.com>
10452 * config/aarch64/aarch64.c
10453 (aarch64_can_const_movi_rtx_p): Move 0 check.
10455 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10457 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
10458 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
10460 * match.pd: Ditto in address comparison pattern.
10462 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10464 PR tree-optimization/81627
10465 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
10466 closed ssa form for store-store chain.
10468 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10470 PR tree-optimization/81620
10471 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
10472 for store-store chain.
10474 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10476 PR tree-optimization/81588
10477 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
10478 ranges[i].in_p, invert comparison code ccode. For >/>=,
10479 swap rhs1 and rhs2 and comparison code unconditionally,
10480 for </<= don't do that. Don't swap rhs1/rhs2 again if
10481 ranges[i].in_p, instead invert comparison code ccode if
10482 opcode or oe->rank is BIT_IOR_EXPR.
10485 * optabs.def (vec_extract_optab, vec_init_optab): Change from
10486 a direct optab to conversion optab.
10487 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
10488 with GET_MODE_INNER as last argument instead of optab_handler.
10489 * expmed.c (extract_bit_field_1): Likewise. Use vector from
10490 vector extraction if possible and optab is available.
10491 * expr.c (store_constructor): Use convert_optab_handler instead
10492 of optab_handler. Use vector initialization from smaller
10493 vectors if possible and optab is available.
10494 * tree-vect-stmts.c (vectorizable_load): Likewise.
10495 * doc/md.texi (vec_extract, vec_init): Document that the optabs
10496 now have two modes.
10497 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
10498 of vec_init from half-sized vectors with the same element mode.
10499 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
10500 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
10501 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
10502 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
10503 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
10504 after mode in gen_vec_extract* calls.
10505 (vec_extract<mode>): Renamed to ...
10506 (vec_extract<mode><ssescalarmodelower>): ... this.
10507 (vec_extract<mode><ssehalfvecmodelower>): New expander.
10508 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
10509 element mode after mode in gen_vec_init* calls.
10510 (VEC_INIT_HALF_MODE): New mode iterator.
10511 (vec_init<mode>): Renamed to ...
10512 (vec_init<mode><ssescalarmodelower>): ... this.
10513 (vec_init<mode><ssehalfvecmodelower>): New expander.
10514 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
10515 (vec_extractv2sfsf): ... this.
10516 (vec_initv2sf): Renamed to ...
10517 (vec_initv2sfsf): ... this.
10518 (vec_extractv2si): Renamed to ...
10519 (vec_extractv2sisi): ... this.
10520 (vec_initv2si): Renamed to ...
10521 (vec_initv2sisi): ... this.
10522 (vec_extractv4hi): Renamed to ...
10523 (vec_extractv4hihi): ... this.
10524 (vec_initv4hi): Renamed to ...
10525 (vec_initv4hihi): ... this.
10526 (vec_extractv8qi): Renamed to ...
10527 (vec_extractv8qiqi): ... this.
10528 (vec_initv8qi): Renamed to ...
10529 (vec_initv8qiqi): ... this.
10530 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
10531 (vec_init<mode>): Renamed to ...
10532 (vec_init<mode><VEC_base_l>): ... this.
10533 (vec_extract<mode>): Renamed to ...
10534 (vec_extract<mode><VEC_base_l>): ... this.
10535 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
10536 (vec_initv2sfsf): ... this.
10537 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
10538 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10539 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
10540 element mode after mode in gen_vec_init* calls.
10541 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
10542 (vec_init<mode><Vel>): ... this.
10543 (vec_extract<mode>): Renamed to ...
10544 (vec_extract<mode><Vel>): ... this.
10545 * config/aarch64/iterators.md (Vel): New mode attribute.
10546 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
10547 Add element mode after mode in gen_vec_extract* calls.
10548 * config/s390/vector.md (non_vec_l): New mode attribute.
10549 (vec_extract<mode>): Renamed to ...
10550 (vec_extract<mode><non_vec_l>): ... this.
10551 (vec_init<mode>): Renamed to ...
10552 (vec_init<mode><non_vec_l>): ... this.
10553 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
10554 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
10556 * config/arm/iterators.md (V_elem_l): New mode attribute.
10557 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
10558 (vec_extract<mode><V_elem_l>): ... this.
10559 (vec_extractv2di): Renamed to ...
10560 (vec_extractv2didi): ... this.
10561 (vec_init<mode>): Renamed to ...
10562 (vec_init<mode><V_elem_l>): ... this.
10563 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
10564 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
10565 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
10566 Add element mode after gen_vec_extract* calls.
10567 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
10568 (vec_init<mode><unitmode>): ... this.
10569 (vec_extract<mode>): Renamed to ...
10570 (vec_extract<mode><unitmode>): ... this.
10571 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
10572 (vec_init<mode><unitmode>): ... this.
10573 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
10574 (vec_initv2sfsf): ... this.
10575 (vec_extractv2sf): Renamed to ...
10576 (vec_extractv2sfsf): ... this.
10577 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
10578 Add element mode after gen_vec_extract* calls.
10579 * config/mips/mips.md (unitmode): New mode iterator.
10580 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
10581 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
10582 * config/spu/spu.md (inner_l): New mode attribute.
10583 (vec_init<mode>): Renamed to ...
10584 (vec_init<mode><inner_l>): ... this.
10585 (vec_extract<mode>): Renamed to ...
10586 (vec_extract<mode><inner_l>): ... this.
10587 * config/sparc/sparc.md (veltmode): New mode iterator.
10588 (vec_init<VMALL:mode>): Renamed to ...
10589 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
10590 * config/ia64/vect.md (vec_initv2si): Renamed to ...
10591 (vec_initv2sisi): ... this.
10592 (vec_initv2sf): Renamed to ...
10593 (vec_initv2sfsf): ... this.
10594 (vec_extractv2sf): Renamed to ...
10595 (vec_extractv2sfsf): ... this.
10596 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
10597 (vec_init<mode>): Renamed to ...
10598 (vec_init<mode><VEC_base_l>): ... this.
10599 (vec_extract<mode>): Renamed to ...
10600 (vec_extract<mode><VEC_base_l>): ... this.
10601 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
10602 (vec_initv2sfsf): ... this.
10603 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
10604 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10605 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
10606 gen_vec_init* calls.
10608 2017-08-01 Richard Biener <rguenther@suse.de>
10610 PR tree-optimization/81297
10611 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
10612 TREE_OVERFLOW from INTEGER_CSTs.
10614 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
10616 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
10618 2017-07-31 Carl Love <cel@us.ibm.com>
10620 * config/rs6000/rs6000-c: Add support for built-in functions
10621 vector signed char vec_xl_be (signed long long, signed char *);
10622 vector unsigned char vec_xl_be (signed long long, unsigned char *);
10623 vector signed int vec_xl_be (signed long long, signed int *);
10624 vector unsigned int vec_xl_be (signed long long, unsigned int *);
10625 vector signed long long vec_xl_be (signed long long, signed long long *);
10626 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
10627 vector signed short vec_xl_be (signed long long, signed short *);
10628 vector unsigned short vec_xl_be (signed long long, unsigned short *);
10629 vector double vec_xl_be (signed long long, double *);
10630 vector float vec_xl_be (signed long long, float *);
10631 * config/rs6000/altivec.h (vec_xl_be): Add #define.
10632 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
10633 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
10635 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
10636 (altivec_expand_builtin): Add switch statement to call
10637 altivec_expand_xl_be for each builtin.
10638 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
10639 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
10640 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
10641 __builtin_vsx_le_be_v16qi.
10642 * doc/extend.texi: Update the built-in documentation file for the
10643 new built-in functions.
10645 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10648 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
10650 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10652 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10654 * config.gcc: Add z14.
10655 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
10656 CPU model numbers for z13s and z14.
10657 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
10659 * config/s390/s390-opts.h (enum processor_type): Rename
10660 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10661 * config/s390/s390.c (processor_table): Add field for CPU name to
10662 be passed to Binutils.
10663 (s390_asm_output_machine_for_arch): Use the new field in
10664 processor_table for Binutils.
10665 (s390_expand_builtin): Replace arch12 with z14.
10666 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10667 (s390_get_sched_attrmask): Likewise.
10668 (s390_get_unit_mask): Likewise.
10669 * config/s390/s390.opt: Add z14 to processor_type enum.
10671 2017-07-31 Martin Jambor <mjambor@suse.cz>
10674 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
10675 regardless of optimization level.
10677 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
10678 Martin Liska <mliska@suse.cz>
10680 * predict.def: Remove old comment and adjust probability.
10681 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
10682 PREDICT statements.
10684 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10687 * config/i386/i386.c (ix86_function_naked): New function.
10688 (ix86_can_use_return_insn_p): Return false for naked functions.
10689 (ix86_expand_prologue): Skip prologue for naked functions.
10690 (ix86_expand_epilogue): Skip epilogue for naked functions
10691 and emit trap instruction.
10692 (ix86_warn_func_return): New function.
10693 (ix86_attribute_table): Add "naked" attribute specification.
10694 (TARGET_WARN_FUNC_RETURN): Define.
10695 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
10697 2017-07-31 Martin Liska <mliska@suse.cz>
10699 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
10700 (dump_gimple_bb_header): Always dump BB info.
10701 (pp_cfg_jump): Do not append info about BB when dumping a jump.
10703 2017-07-31 Martin Liska <mliska@suse.cz>
10706 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
10707 also with current_function_decl non-null equality.
10709 2017-07-31 Jakub Jelinek <jakub@redhat.com>
10712 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
10713 change type to the element type, instead add eltype variable and
10714 use it where we are interested in the element type.
10716 PR tree-optimization/81603
10717 * ipa-polymorphic-call.c
10718 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
10719 offset arithmetic in offset_int, bail out if the resulting bit offset
10720 doesn't fit into shwi.
10722 2017-07-31 Martin Liska <mliska@suse.cz>
10724 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
10725 (gimplify_save_expr): Fix comment.
10727 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
10730 * config/i386/i386.c (ix86_function_arg): Update arguments for
10732 (ix86_compute_frame_layout): Set the initial stack offset to
10733 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
10734 INCOMING_FRAME_SP_OFFSET.
10735 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
10736 stack before exception handler returns.
10737 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
10738 the 'ERROR_CODE' for exception handler.
10740 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
10742 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
10743 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
10744 (ASM_OUTPUT_REG_POP): Ditto.
10745 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
10746 instead of asm_fprintf to output pure string.
10748 2017-07-29 Jakub Jelinek <jakub@redhat.com>
10750 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
10751 to imported_module_or_decl hook.
10752 (debug_nothing_tree_tree_tree_bool): Remove.
10753 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
10754 * debug.c (do_nothing_debug_hooks): Use
10755 debug_nothing_tree_tree_tree_bool_bool instead of
10756 debug_nothing_tree_tree_tree_bool.
10757 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10758 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
10759 * sdbout.c (sdb_debug_hooks): Likewise.
10760 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
10761 (gen_namespace_die): Add DW_AT_export_symbols attribute if
10763 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
10764 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
10765 attribute, don't add anything.
10767 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10769 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
10770 (fold_build2_stat_loc): Likewise.
10771 (fold_build3_stat_loc): Likewise.
10772 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
10773 (fold_build1_loc): Remove macro.
10774 (fold_build2_loc): Likewise.
10775 (fold_build3_loc): Likewise.
10777 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10779 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
10780 (gimple_build_debug_bind_source_stat): Likewise.
10781 * gimple.h (gimple_build_debug_bind): Remove macro.
10782 (gimple_build_debug_bind_source): Likewise.
10784 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10786 * bitmap.c (bitmap_alloc): Adjust.
10787 (bitmap_gc_alloc): Likewise.
10788 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
10790 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10792 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
10793 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
10794 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
10795 (bitmap_gc_alloc_stat): Likewise.
10796 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
10798 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10800 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
10801 * rtl.h (shallow_copy_rtx): Remove macro.
10803 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10805 * emit-rtl.c (gen_raw_REG): Adjust.
10806 * gengenrtl.c (gendef): Likewise.
10807 * rtl.c (rtx_alloc_stat): Remove _stat from name.
10808 * rtl.h (rtx_alloc): Remove macro.
10810 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10812 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
10813 (build_tree_list_stat): Likewise.
10814 * tree.h (build_tree_list): Remove macro.
10815 (build_tree_list_vec): Likewise.
10817 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10819 * tree.c (make_vector_stat): Remove _stat from name.
10820 (build_vector_stat): Likewise.
10821 * tree.h (make_vector_stat): Remove macro.
10822 (build_vector_stat): Likewise.
10824 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10826 * tree.h (build_var_debug_value): Remove prototype.
10828 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10830 * tree.c (tree_cons_stat): Remove _stat from name.
10831 * tree.h (tree_cons): Remove macro.
10833 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10835 * tree.c (build_vl_exp_stat): Remove _stat from name.
10836 * tree.h (build_vl_exp): Remove macro.
10838 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10840 * tree.c (build_decl_stat): Remove _stat from name.
10841 * tree.h (build_decl): Remove macro.
10843 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10845 * gimple.c (gimple_build_with_ops_stat): Adjust.
10846 (gimple_alloc_stat): Remove _stat from name.
10847 * gimple.h (gimple_alloc): Remove macro.
10849 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10851 * tree.c (make_tree_vec_stat): Remove _stat from name.
10852 (grow_tree_vec_stat): Likewise.
10853 * tree.h (make_tree_vec_stat): Adjust prototype.
10854 (grow_tree_vec_stat): Likewise.
10855 (make_tree_vec): Remove macro.
10856 (grow_tree_vec): Likewise.
10858 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10860 * fold-const.c (fold_build1_stat_loc): Adjust.
10861 (fold_build2_stat_loc): Likewise.
10862 (fold_build3_stat_loc): Likewise.
10863 * tree.c (build0_stat): Remove _stat from name.
10864 (build1_stat): Likewise.
10865 (build2_stat): Likewise.
10866 (build3_stat): Likewise.
10867 (build4_stat): Likewise.
10868 (build5_stat): Likewise.
10869 * tree.h (build1_loc): Remove macro, and rename _stat function
10871 (build2_loc): Likewise.
10872 (build3_loc): Likewise.
10873 (build4_loc): Likewise.
10874 (build5_loc): Likewise.
10876 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10878 * tree.c (make_int_cst_stat): Remove _stat from name.
10879 * tree.h (make_int_cst_stat): Adjust prototype.
10880 (make_int_cst): Remove macro.
10882 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10884 * tree.c (make_tre_binfo_stat): Remove _stat from name.
10885 * tree.h (make_tree_binfo_stat): Adjust prototype.
10886 (make_tree_binfo): Remove.
10888 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10890 * tree.c (copy_node_stat): Rename to copy_node.
10891 (build_distinct_type_copy): Adjust.
10892 * tree.h (copy_node_stat): Adjust prototype.
10893 (copy_node): Remove macro.
10895 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10897 * tree.c (make_node_stat): rename to make_node.
10898 (build_tree_list_stat): Adjust.
10899 (build0_stat): Likewise.
10900 (build2_stat): Likewise.
10901 (build3_stat): Likewise.
10902 (build4_stat): Likewise.
10903 (build5_stat): Likewise.
10904 (build_decl_stat): Likewise.
10905 * tree.h (make_node_stat): Adjust prototype.
10906 (make_node): remove macro.
10908 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
10910 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
10911 (PPC_FEATURE2_SCV): Likewise.
10912 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
10914 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10916 * config/aarch64/aarch64.c
10917 (aarch64_internal_mov_immediate): Add new special pattern.
10918 * config/aarch64/aarch64.md (*movdi_aarch64):
10919 Add reg/32bit const mov case.
10921 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10922 Richard Sandiford <richard.sandiford@linaro.org>
10924 * config/aarch64/aarch64.md (mov<mode>): Generalize.
10925 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10926 Add integer and movi cases.
10927 (movi-split-hf-df-sf split, fp16): New.
10928 (enabled): Added TARGET_FP_F16INST.
10929 * config/aarch64/iterators.md (GPF_HF): New.
10930 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
10932 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10934 * config/aarch64/aarch64.c
10935 (aarch64_simd_container_mode): Add prototype.
10936 (aarch64_expand_mov_immediate): Add HI support.
10937 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
10938 (aarch64_can_const_movi_rtx_p): New.
10939 (aarch64_preferred_reload_class):
10940 Remove restrictions of using FP registers for certain SIMD operations.
10941 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
10942 (aarch64_valid_floating_const): Add integer move validation.
10943 (aarch64_simd_imm_scalar_p): Remove.
10944 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
10945 (aarch64_legitimate_constant_p): Expand list of supported cases.
10946 * config/aarch64/aarch64-protos.h
10947 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
10948 (aarch64_reinterpret_float_as_int): New.
10949 (aarch64_simd_imm_scalar_p): Remove.
10950 * config/aarch64/constraints.md (Uvi): New.
10951 (Dd): Split into Ds and new Dd.
10952 * config/aarch64/aarch64.md (*movsi_aarch64):
10954 (*movdi_aarch64): Add SIMD mov case.
10956 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10958 * tree-predcom.c: (struct chain): Handle store-store chain in which
10959 stores for elimination only store loop invariant values.
10960 (execute_pred_commoning_chain): Ditto.
10961 (prepare_initializers_chain_store_elim): Ditto.
10962 (prepare_finalizers): Ditto.
10963 (is_inv_store_elimination_chain): New function.
10964 (initialize_root_vars_store_elim_1): New function.
10966 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10968 * tree-predcom.c: Revise general description of the pass.
10969 (enum chain_type): New enum type for store elimination.
10970 (struct chain): New field supporting store elimination.
10971 (struct component): Ditto.
10972 (dump_chain): Dump store-stores chain.
10973 (release_chain): Release resources.
10974 (split_data_refs_to_components): Compute and create component
10975 contains only stores for elimination.
10976 (get_chain_last_ref_at): New function.
10977 (make_invariant_chain): Initialization.
10978 (make_rooted_chain): Specify chain type in parameter and record it.
10979 (add_looparound_copies): Skip for store-stores chain.
10980 (determine_roots_comp): Compute type of chain and pass it to
10982 (initialize_root_vars_store_elim_2): New function.
10983 (finalize_eliminated_stores): New function.
10984 (remove_stmt): Handle store for elimination.
10985 (execute_pred_commoning_chain): Execute predictive commoning on
10986 store-store chains.
10987 (determine_unroll_factor): Skip unroll for store-stores chain.
10988 (prepare_initializers_chain_store_elim): New function.
10989 (prepare_initializers_chain): Hanlde store-store chain.
10990 (prepare_finalizers_chain, prepare_finalizers): New function.
10991 (tree_predictive_commoning_loop): Return integer value indicating
10992 if loop is unrolled or lcssa form is corrupted.
10993 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
10995 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10997 * tree-predcom.c (initialize_root): Delete.
10998 (execute_pred_commoning_chain): Initialize root vars and replace
10999 reference of non-combined chain directly, rather than call above
11002 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11004 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
11005 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
11007 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11009 * tree-predcom.c (struct chain): New field init_seq.
11010 (release_chain): Release init_seq.
11011 (prepare_initializers_chain): Record intialization stmts in above
11013 (insert_init_seqs): New function.
11014 (tree_predictive_commoning_loop): Call insert_init_seqs.
11016 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11018 * tree-predcom.c (determine_roots_comp): Skip trivial components.
11020 2017-07-28 Richard Biener <rguenther@suse.de>
11022 * match.pd: Remove superfluous :c.
11023 * genmatch.c (simplify::id): Add member.
11024 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
11026 (current_id): New global.
11027 (dt_node::parent): Move from ...
11028 (dt_operand::parent): ... here. Add for_id member.
11029 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
11030 (decision_tree::find_node): Relax order requirement when
11031 merging DT_TRUE nodes to ones inbetween the current simplify
11032 and the one we try to merge with. Add diagnostic whenever
11033 we need to enforce pattern order by not merging.
11034 (decision_tree::insert): Set current_id.
11035 (decision_tree::print_node): Dump parent node and for_id.
11036 (parser::last_id): Add member.
11037 (parser::push_simplify): Assign unique id.
11038 (parser::parser): Initialize last_id.
11040 2017-07-28 Martin Liska <mliska@suse.cz>
11043 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
11044 gimple_build_debug_bind.
11046 2017-07-28 Richard Biener <rguenther@suse.de>
11048 PR tree-optimization/81502
11049 * match.pd: Add pattern combining BIT_INSERT_EXPR with
11051 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
11053 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
11054 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
11055 for BIT_FIELD_REF args.
11056 * fold-const.c (make_bit_field_ref): Likewise.
11057 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
11059 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11062 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
11063 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
11064 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
11065 Or it into SANITIZER_UNDEFINED.
11066 * ubsan.c: Include gimple-fold.h and varasm.h.
11067 (ubsan_expand_ptr_ifn): New function.
11068 (instrument_pointer_overflow): New function.
11069 (maybe_instrument_pointer_overflow): New function.
11070 (instrument_object_size): Formatting fix.
11071 (pass_ubsan::execute): Call instrument_pointer_overflow
11072 and maybe_instrument_pointer_overflow.
11073 * internal-fn.c (expand_UBSAN_PTR): New function.
11074 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
11075 * sanitizer.def (__ubsan_handle_pointer_overflow,
11076 __ubsan_handle_pointer_overflow_abort): New builtins.
11077 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
11078 * internal-fn.def (UBSAN_PTR): New internal function.
11079 * opts.c (sanitizer_opts): Add pointer-overflow.
11080 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
11081 * fold-const.c (build_range_check): Compute pointer range check in
11082 integral type if pointer arithmetics would be needed. Formatting
11085 2017-07-28 Martin Liska <mliska@suse.cz>
11088 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
11089 parameters that are of a variable-length.
11091 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
11093 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
11095 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
11096 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
11097 (CRT_CALL_STATIC_FUNCTION): Likewise.
11098 (ASM_DEFAULT_SPEC): New define.
11099 (ASM_SPEC32): Likewise.
11100 (ASM_SPEC64): Likewise.
11101 (ASM_SPEC_COMMON): Likewise.
11102 (ASM_SPEC): Likewise.
11103 (INVALID_64BIT): Likewise.
11104 (LINK_OS_DEFAULT_SPEC): Likewise.
11105 (LINK_OS_SPEC32): Likewise.
11106 (LINK_OS_SPEC64): Likewise.
11107 (POWERPC_LINUX): Likewise.
11108 (PTRDIFF_TYPE): Likewise.
11109 (RESTORE_FP_PREFIX): Likewise.
11110 (RESTORE_FP_SUFFIX): Likewise.
11111 (SAVE_FP_PREFIX): Likewise.
11112 (SAVE_FP_SUFFIX): Likewise.
11113 (SIZE_TYPE): Likewise.
11114 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
11115 (TARGET_64BIT): Likewise.
11116 (TARGET_64BIT): Likewise.
11117 (TARGET_AIX): Likewise.
11118 (WCHAR_TYPE_SIZE): Likewise.
11119 (WCHAR_TYPE): Undefine.
11120 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
11121 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
11122 (CPP_OS_RTEMS_SPEC): Delete.
11123 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
11124 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
11126 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
11128 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11130 PR tree-optimization/81578
11131 * tree-parloops.c (build_new_reduction): Bail out if
11132 reduction_code isn't one of the standard OpenMP reductions.
11133 Move the details printing after that decision.
11135 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11137 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
11138 related to reload_in_progress.
11139 (splat_input_operand): Likewise.
11140 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
11142 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
11144 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
11145 (TARGET_INSTANTIATE_DECLS): Likewise.
11146 (legitimate_indexed_address_p): Delete reload_in_progress code.
11147 (rs6000_debug_legitimate_address_p): Likewise.
11148 (rs6000_eliminate_indexed_memrefs): Likewise.
11149 (rs6000_emit_le_vsx_store): Likewise.
11150 (rs6000_emit_move_si_sf_subreg): Likewise.
11151 (rs6000_emit_move): Likewise.
11152 (register_to_reg_type): Likewise.
11153 (rs6000_pre_atomic_barrier): Likewise.
11154 (rs6000_machopic_legitimize_pic_address): Likewise.
11155 (rs6000_allocate_stack_temp): Likewise.
11156 (rs6000_address_for_fpconvert): Likewise.
11157 (rs6000_address_for_altivec): Likewise.
11158 (rs6000_secondary_memory_needed_rtx): Delete function.
11159 (rs6000_check_sdmode): Likewise.
11160 (rs6000_alloc_sdmode_stack_slot): Likewise.
11161 (rs6000_instantiate_decls): Likewise.
11162 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
11163 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
11164 Delete reload_in_progress.
11165 (*vec_reload_and_plus_<mptrsize>): Likewise.
11166 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
11167 (vsx_div_v2di): Likewise.
11168 (vsx_udiv_v2di): Likewise.
11170 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11172 * config/rs6000/rs6000.opt (mlra): Replace with stub.
11173 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
11174 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
11175 (rs6000_debug_reg_global): Delete print of LRA status.
11176 (rs6000_option_override_internal): Delete dead LRA related code.
11177 (rs6000_lra_p): Delete function.
11178 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
11180 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
11182 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
11183 * config/riscv/rtems.h: New file.
11185 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11186 Sudakshina Das <sudi.das@arm.com>
11188 * config/aarch64/aarch64.md
11189 (define_split for and<mode>3nr_compare): Move
11190 non aarch64_logical_operand to a register.
11191 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
11192 register immediate operand to a register.
11193 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
11195 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11197 PR middle-end/81564
11198 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
11200 2017-07-27 Richard Biener <rguenther@suse.de>
11202 PR tree-optimization/81573
11203 PR tree-optimization/81494
11204 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
11205 multi defuse cycle case.
11207 2017-07-27 Richard Biener <rguenther@suse.de>
11209 PR tree-optimization/81571
11210 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
11213 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
11215 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
11216 earlier and only if MASK_FPU is set. Adjust formatting.
11218 2017-07-27 Martin Liska <mliska@suse.cz>
11220 * opt-functions.awk: Add validation of value of Init.
11221 * optc-gen.awk: Pass new argument.
11223 2017-07-27 Martin Liska <mliska@suse.cz>
11225 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
11226 Fix wrong condition.
11228 2017-07-27 Martin Liska <mliska@suse.cz>
11230 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
11231 BBs and edges seen by autoFDO.
11233 2017-07-27 Richard Biener <rguenther@suse.de>
11235 PR tree-optimization/81502
11236 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
11237 with incompatible but same sized type.
11238 (execute_update_addresses_taken): Likewise.
11240 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
11242 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
11243 flag_tree_loop_vectorize rather than flag_tree_vectorize.
11245 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11248 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
11249 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
11250 Change s_operand to memory_operand.
11252 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
11254 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
11255 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
11256 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
11257 Emit instructions rather than returning an expression. Handle TFmode
11258 and KFmode by casting to TImode.
11259 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
11260 (rs6000_emit_le_vsx_store): Likewise.
11261 * config/rs6000/vsx.md (VSX_TI): New iterator.
11262 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
11263 (*vsx_le_undo_permute_<mode>): Likewise.
11264 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
11265 emit the split sequence.
11266 (*vsx_le_perm_store_<mode>): Likewise.
11268 2017-07-27 Jakub Jelinek <jakub@redhat.com>
11270 PR tree-optimization/81555
11271 PR tree-optimization/81556
11272 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
11273 if true, force CHANGED for the recursive invocation.
11274 (reassociate_bb): Remember original length of ops array, pass
11275 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
11277 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
11278 attributes for noipa attribute. For naked attribute use
11279 lookup_attribute first before lookup_attribute_spec.
11280 * final.c (rest_of_handle_final): Disable IPA RA for functions with
11282 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
11283 for functions with noipa attribute.
11284 (cgraph_externally_visible_p): Return true for functions with noipa
11286 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
11287 for functions with noipa attribute.
11288 * doc/extend.texi: Document noipa function attribute.
11289 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
11290 also for functions with noipa attribute.
11291 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
11293 2017-07-26 Andrew Pinski <apinski@cavium.com>
11295 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
11296 vec_unalign_load_cost and vec_unalign_store_cost.
11298 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11300 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
11301 -mvsx-small-integer option.
11302 (ISA_3_0_MASKS_IEEE): Likewise.
11303 (OTHER_VSX_VECTOR_MASKS): Likewise.
11304 (POWERPC_MASKS): Likewise.
11305 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
11306 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
11307 code, only testing for DImode being allowed in non-VSX floating
11309 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
11310 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
11312 (rs6000_option_override_internal): Delete -mvsx-small-integer.
11313 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
11314 TARGET_P8_VECTOR test.
11315 (rs6000_secondary_reload_simple_move): Likewise.
11316 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
11317 since TARGET_P9_VECTOR was already tested.
11318 (rs6000_opt_masks): Remove -mvsx-small-integer.
11319 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
11320 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11322 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
11323 test for TARGET_VEXTRACTUB was used, and that uses
11325 (p9 extract splitter): Likewise.
11326 (vsx_extract_<mode>_di_p9): Likewise.
11327 (vsx_extract_<mode>_store_p9): Likewise.
11328 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
11329 for TARGET_P9_VECTOR was used. Delete code that is now dead with
11330 the elimination of TARGET_VSX_SMALL_INTEGER.
11331 (vsx_extract_<mode>_p8): Likewise.
11332 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
11333 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
11334 (vsx_set_<mode>_p9): Likewise.
11335 (vsx_set_v4sf_p9): Likewise.
11336 (vsx_set_v4sf_p9_zero): Likewise.
11337 (vsx_insert_extract_v4sf_p9): Likewise.
11338 (vsx_insert_extract_v4sf_p9_2): Likewise.
11339 * config/rs6000/rs6000.md (sign extend splitter): Change
11340 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
11341 (floatsi<mode>2_lfiwax_mem): Likewise.
11342 (floatunssi<mode>2_lfiwzx_mem): Likewise.
11343 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
11344 since a test for TARGET_P9_VECTOR was used.
11345 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11346 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
11347 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11348 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11349 TARGET_P8_VECTOR test.
11350 (fix_trunc<mode>si2_stfiwx): Likewise.
11351 (fix_trunc<mode>si2_internal): Likewise.
11352 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
11353 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11355 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11356 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11357 TARGET_P8_VECTOR test.
11358 (fixuns_trunc<mode>si2_stfiwx): Likewise.
11359 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
11360 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11362 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11363 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
11364 since a test for TARGET_P9_VECTOR was used.
11365 (splitter for loading small constants): Likewise.
11367 2017-07-26 Andrew Pinski <apinski@cavium.com>
11369 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
11372 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
11375 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
11376 (fp_valid_at): Likewise.
11378 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11380 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
11381 (qdf24xx_addrcost_table): Likewise.
11382 (cortexa57_tunings): Update to use generic_branch_cost.
11383 (cortexa72_tunings): Likewise.
11384 (cortexa73_tunings): Likewise.
11385 (qdf24xx_tunings): Likewise.
11387 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11389 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
11390 (thunderx2t99_branch_cost): Likewise.
11391 (cortexa35_tunings): Update to use generic_branch_cost.
11392 (cortexa53_tunings): Likewise.
11393 (cortexa57_tunings): Likewise.
11394 (cortexa72_tunings): Likewise.
11395 (cortexa73_tunings): Likewise.
11396 (thunderx2t99_tunings): Likewise.
11398 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11400 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
11401 (sparc_option_override): Honour MASK_FSMULD.
11402 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
11403 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
11404 * config/sparc/sparc.opt (mfsmuld): New option.
11405 * doc/invoke.texi (mfsmuld): Document option.
11407 2017-07-26 Marek Polacek <polacek@redhat.com>
11409 PR middle-end/70992
11410 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
11412 2017-07-26 Richard Biener <rguenther@suse.de>
11414 * gimple-match-head.c (do_valueize): Return OP if valueize
11416 (get_def): New helper to get at the def stmt of a SSA name
11417 if valueize allows.
11418 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
11419 do_valueize to get at the def stmt.
11420 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
11422 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
11424 PR middle-end/46932
11425 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
11427 2017-07-26 Martin Liska <mliska@suse.cz>
11430 * function.c (expand_function_start): Make expansion of
11431 nonlocal_goto_save_area after parm_birth_insn.
11433 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11435 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
11436 from all CPU target flags enable members.
11438 2017-07-26 Richard Biener <rguenther@suse.de>
11440 * genmatch.c (dt_simplify::gen): Make iterator vars const.
11441 (decision_tree::gen): Make 'type' const.
11442 (write_predicate): Likewise.
11444 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11446 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
11447 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
11448 (rs6000_option_override_internal): Likewise.
11449 (rs6000_expand_vector_set): Likewise.
11450 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
11451 (TARGET_UPPER_REGS_SF): Likewise.
11452 (TARGET_UPPER_REGS_DI): Likewise.
11453 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
11454 (TARGET_DIRECT_MOVE_64BIT): Likewise.
11455 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
11456 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11457 (Splitters for DI constants in Altivec registers): Likewise.
11458 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
11459 (vsx_set_v4sf_p9): Likewise.
11460 (vsx_set_v4sf_p9_zero): Likewise.
11461 (vsx_insert_extract_v4sf_p9): Likewise.
11462 (vsx_insert_extract_v4sf_p9_2): Likewise.
11464 2017-07-25 Carl Love <cel@us.ibm.com>
11466 * doc/extend.texi: Update the built-in documentation file for the
11467 existing built-in functions
11468 vector signed char vec_cnttz (vector signed char);
11469 vector unsigned char vec_cnttz (vector unsigned char);
11470 vector signed short vec_cnttz (vector signed short);
11471 vector unsigned short vec_cnttz (vector unsigned short);
11472 vector signed int vec_cnttz (vector signed int);
11473 vector unsigned int vec_cnttz (vector unsigned int);
11474 vector signed long long vec_cnttz (vector signed long long);
11475 vector unsigned long long vec_cnttz (vector unsigned long long);
11477 2017-07-25 Andrew Pinski <apinski@cavium.com>
11479 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
11480 accesses where the use is for the first operand of a BIT_INSERT.
11482 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11485 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11486 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11488 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11490 * config/i386/gstabs.h: Delete.
11491 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
11493 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
11495 * config/i386/i386.c (ix86_decompose_address): Do not check for
11496 register RTX when looking at index_reg or base_reg.
11497 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
11499 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
11501 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
11502 to update EH info here.
11504 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11506 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
11508 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11510 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
11512 2017-07-25 Torsten Duwe <duwe@suse.de>
11514 * common.opt: Introduce -fpatchable-function-entry
11515 command line option, and its variables function_entry_patch_area_size
11516 and function_entry_patch_area_start.
11517 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
11518 including a two-value parser.
11519 * target.def (print_patchable_function_entry): New target hook.
11520 * targhooks.h (default_print_patchable_function_entry): New function.
11521 * targhooks.c (default_print_patchable_function_entry): Likewise.
11522 * toplev.c (process_options): Switch off IPA-RA if
11523 patchable function entries are being generated.
11524 * varasm.c (assemble_start_function): Look at the
11525 patchable-function-entry command line switch and current
11526 function attributes and maybe generate NOP instructions by
11527 calling the print_patchable_function_entry hook.
11528 * doc/extend.texi: Document patchable_function_entry attribute.
11529 * doc/invoke.texi: Document -fpatchable_function_entry
11530 command line option.
11531 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
11533 * doc/tm.texi: Re-generate.
11535 2017-07-25 Jakub Jelinek <jakub@redhat.com>
11538 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
11539 TARGET_AVX512DQ rather than TARGET_AVX512BW.
11541 2017-07-25 Tamar Christina <tamar.christina@arm.com>
11543 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
11545 2017-07-25 Richard Biener <rguenther@suse.de>
11547 PR tree-optimization/81455
11548 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
11549 not walk in cycles when looking for guards.
11551 2017-07-25 Richard Biener <rguenther@suse.de>
11553 PR tree-optimization/81529
11554 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
11555 when optimizing backedge uses.
11557 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
11559 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
11561 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
11562 to dl_section_ref. On AIX, append an expression to subtract
11563 the size of the section length to dl_section_ref.
11565 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
11567 * configure.ac: If any of the config.* scripts fail, exit 1.
11568 * configure: Regenerate.
11570 2017-07-25 Richard Biener <rguenther@suse.de>
11572 PR middle-end/81546
11573 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
11574 of immediate uses, be more verbose on errors.
11576 2017-07-25 Richard Biener <rguenther@suse.de>
11578 PR tree-optimization/81510
11579 * tree-vect-loop.c (vect_is_simple_reduction): When the
11580 reduction stmt is not inside the loop bail out.
11582 2017-07-25 Richard Biener <rguenther@suse.de>
11584 PR tree-optimization/81303
11585 * tree-vect-loop-manip.c (vect_loop_versioning): Build
11586 profitability check against LOOP_VINFO_NITERSM1.
11588 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11590 * domwalk.c (cmp_bb_postorder): Simplify.
11591 (sort_bbs_postorder): New function. Use it...
11592 (dom_walker::walk): ...here to optimize common cases.
11594 2017-07-25 Martin Liska <mliska@suse.cz>
11597 * ipa-visibility.c (function_and_variable_visibility): Make the
11598 redirection just on target that supports aliasing.
11599 Fix GNU coding style.
11601 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
11604 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
11606 * config/arm/rtems.h: Likewise.
11607 * config/bfin/rtems.h: Likewise.
11608 * config/i386/rtemself.h: Likewise.
11609 * config/lm32/rtems.h: Likewise.
11610 * config/m32c/rtems.h: Likewise.
11611 * config/m68k/rtemself.h: Likewise.
11612 * config/microblaze/rtems.h: Likewise.
11613 * config/mips/rtems.h: Likewise.
11614 * config/moxie/rtems.h: Likewise.
11615 * config/nios2/rtems.h: Likewise.
11616 * config/powerpcspe/rtems.h: Likewise.
11617 * config/rs6000/rtems.h: Likewise.
11618 * config/rtems.h: Likewise.
11619 * config/sh/rtems.h: Likewise.
11620 * config/sh/rtemself.h: Likewise.
11621 * config/sparc/rtemself.h: Likewise.
11623 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11626 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
11627 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
11628 * tree-ssa-structalias.c (alias_get_name): Same.
11630 2017-07-25 Bin Cheng <bin.cheng@arm.com>
11633 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
11634 instructions if no du chain is found.
11636 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11638 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
11640 2017-07-25 Richard Biener <rguenther@suse.de>
11642 PR middle-end/81505
11643 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
11646 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11648 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
11649 upper-regs options.
11650 (ISA_2_7_MASKS_SERVER): Likewise.
11651 (ISA_3_0_MASKS_IEEE): Likewise.
11652 (OTHER_P8_VECTOR_MASKS): Likewise.
11653 (OTHER_VSX_VECTOR_MASKS): Likewise.
11654 (POWERPC_MASKS): Likewise.
11655 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
11656 duplicate list of options.
11657 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
11658 explicit -mupper-regs options.
11659 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
11660 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
11661 alias for -mupper-regs-df.
11662 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11663 (rs6000_init_hard_regno_mode_ok): Likewise.
11664 (rs6000_option_override_internal): Likewise.
11665 (rs6000_opt_masks): Likewise.
11666 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
11667 options in terms of whether -mvsx or -mpower8-vector was used.
11668 (TARGET_UPPER_REGS_DI): Likewise.
11669 (TARGET_UPPER_REGS_SF): Likewise.
11670 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
11671 -mupper-regs-* options.
11673 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11675 * passes.c (emergency_dump_function): Print some empty lines and a
11676 header before the RTL dump.
11678 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11680 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
11682 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
11685 * config/aarch64/aarch64.c (aarch64_classify_symbol):
11686 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
11688 2017-07-24 Carl Love <cel@us.ibm.com>
11690 * config/rs6000/rs6000-c.c: Add support for built-in functions
11691 vector float vec_extract_fp32_from_shorth (vector unsigned short);
11692 vector float vec_extract_fp32_from_shortl (vector unsigned short);
11693 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
11694 vec_extract_fp_from_shortl): Add defines for the two builtins.
11695 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
11696 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
11698 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
11699 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
11700 * doc/extend.texi: Update the built-in documentation file for the
11701 new built-in function.
11703 2017-07-24 Jakub Jelinek <jakub@redhat.com>
11706 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
11708 * doc/generic.texi: Likewise.
11709 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11710 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11712 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
11714 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
11715 (aarch64_mls_elt_merge<mode>): Likewise.
11717 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
11719 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11720 having __cxa_atexit.
11722 2017-07-23 Michael Collison <michael.collison@arm.com>
11724 * config/arm/arm.c (arm_option_override): Deprecate
11725 use of -mstructure-size-boundary.
11726 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
11727 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
11729 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11732 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
11733 Reduce cost estimate for direct moves.
11735 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
11738 * config/i386/i386.c (ix86_option_override_internal): Disable
11739 BMI, BMI2 and TBM instructions for -m16.
11741 2017-07-21 Carl Love <cel@us.ibm.com>
11743 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11744 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11745 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11746 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11747 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11748 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11749 VMULOSW): New enum "unspec" values.
11750 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11751 altivec_vmulosw): New patterns.
11752 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11753 VMULOSW): Add definitions.
11755 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
11757 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
11758 (qdf24xx): Likewise.
11759 * config/aarch64/aarch64-options-extensions.def (rdma); New.
11760 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
11761 (AARCH64_FL_V8_1): Renumber.
11762 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
11763 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
11764 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
11765 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
11766 rdma to feature modifiers list.
11768 2017-07-21 Yury Gribov <tetra2005@gmail.com>
11770 PR middle-end/56727
11771 * ipa-visibility (function_and_variable_visibility): Convert
11772 recursive PLT call to direct call if appropriate.
11774 2017-07-21 Andrew Pinski <apinski@cavium.com>
11776 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
11777 operand 1 to see if the types precision matches.
11778 * fold-const.c (operand_equal_p): Likewise.
11780 2017-07-21 Richard Biener <rguenther@suse.de>
11782 PR tree-optimization/81303
11783 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
11784 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
11785 (vect_peeling_hash_get_lowest_cost): Adjust.
11786 (vect_enhance_data_refs_alignment): Likewise. Use
11787 vect_get_peeling_costs_all_drs to compute the penalty for no
11788 peeling to match up costs.
11790 2017-07-21 Richard Biener <rguenther@suse.de>
11792 PR tree-optimization/81500
11793 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
11794 we didn't identify a reduction path.
11796 2017-07-21 Tom de Vries <tom@codesourcery.com>
11797 Cesar Philippidis <cesar@codesourcery.com>
11799 PR gcov-profile/81442
11800 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
11803 2017-07-21 Tom de Vries <tom@codesourcery.com>
11806 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
11808 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
11809 nvptx_override_options_after_change.
11811 2017-07-21 Ulrich Drepper <drepper@redhat.com>
11813 * dwarf2out.c (output_file_names): Avoid double testing for
11814 dwarf_version >= 5.
11816 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
11818 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
11820 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11822 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
11824 (try_crossjump_to_edge): Do not punt on partitioned functions.
11826 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11828 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
11829 Put all BBs reachable only via paths crossing cold region to cold
11831 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
11833 2016-07-21 Richard Biener <rguenther@suse.de>
11835 PR tree-optimization/81303
11836 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
11837 into account prologue and epilogue iterations when raising
11838 min_profitable_iters to sth at least covering one vector iteration.
11840 2017-07-21 Tamar Christina <tamar.christina@arm.com>
11842 * config/arm/arm.c (arm_test_cpu_arch_dat):
11845 2017-07-20 Nathan Sidwell <nathan@acm.org>
11847 Remove TYPE_METHODS.
11848 * tree.h (TYPE_METHODS): Delete.
11849 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
11850 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
11851 (dbxout_type_methods): Scan TYPE_FIELDS.
11852 (dbxout_type): Don't check TYPE_METHODS here.
11853 * function.c (use_register_for_decl): Always ignore register for
11854 class types when not optimizing.
11855 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
11856 * tree.c (free_lang_data_in_type): Stitch out member functions and
11857 templates from TYPE_FIELDS.
11858 (build_distinct_type_copy, verify_type_variant,
11859 verify_type): Member fns are on TYPE_FIELDS.
11860 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
11861 * tree-pretty-print.c (dump_generic_node): Likewise.
11863 2017-07-20 Jakub Jelinek <jakub@redhat.com>
11866 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
11867 V2TImode and V4TImode.
11868 (ix86_expand_vector_extract): Likewise.
11869 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
11870 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
11871 (ssescalarmode): Handle V4TImode and V2TImode.
11872 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
11873 (*vec_extractv2ti, *vec_extractv4ti): New insns.
11874 (VEXTRACTI128_MODE): New mode iterator.
11875 (splitter for *vec_extractv?ti first element): New.
11876 (VEC_INIT_MODE): New mode iterator.
11877 (vec_init<mode>): Consolidate 3 expanders into one using
11878 VEC_INIT_MODE mode iterator.
11880 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11882 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
11883 non_spilled_static_chain_regno_p.
11885 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11887 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
11889 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
11891 * bb-reorder.c (connect_traces): Allow copying of blocks within
11894 2017-07-20 Richard Biener <rguenther@suse.de>
11896 * gimple.h (gimple_phi_result): Add gphi * overload.
11897 (gimple_phi_result_ptr): Likewise.
11898 (gimple_phi_arg): Likewise. Adjust index assert to only
11899 allow actual argument accesses rather than all slots available
11901 (gimple_phi_arg_def): Add gphi * overload.
11902 * tree-phinodes.c (make_phi_node): Initialize only actual
11904 (resize_phi_node): Clear memory not covered by old node,
11905 do not initialize excess argument slots.
11906 (reserve_phi_args_for_new_edge): Initialize new argument slot
11909 2017-07-20 Bin Cheng <bin.cheng@arm.com>
11911 PR tree-optimization/81388
11913 2016-07-21 Bin Cheng <bin.cheng@arm.com>
11915 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
11916 by removing computation of may_be_zero.
11918 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11919 Tom de Vries <tom@codesourcery.com>
11921 PR middle-end/81030
11922 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
11923 when gimple level profile disagrees with what RTL expander did.
11925 2017-07-20 Richard Biener <rguenther@suse.de>
11927 PR tree-optimization/61171
11928 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
11929 (vect_analyze_stmt): Add slp instance parameter.
11930 (vectorizable_reduction): Likewise.
11931 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
11932 (vect_is_simple_reduction): Deal with chains not detected
11933 as SLP reduction chain, specifically not properly associated
11934 chains containing a mix of plus/minus.
11935 (get_reduction_op): Remove.
11936 (get_initial_defs_for_reduction): Simplify, pass in whether
11937 this is a reduction chain, pass in the SLP node for the PHIs.
11938 (vect_create_epilog_for_reduction): Get the SLP instance as
11940 (vectorizable_reduction): Get the SLP instance as arg.
11941 During analysis remember the SLP node with the PHIs in the
11942 instance. Simplify getting at the vectorized reduction PHIs.
11943 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
11944 through SLP instance.
11945 (vect_slp_analyze_operations): Likewise.
11946 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
11947 (vect_transform_stmt): Likewise.
11949 2017-07-20 Tom de Vries <tom@codesourcery.com>
11951 PR tree-optimization/81489
11952 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
11953 read of phi arg location to before loop that modifies phi.
11955 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11957 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
11960 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11962 PR middle-end/81331
11963 * except.c (execute): Fix ordering issue.
11965 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11967 PR rtl-optimization/81423
11968 * combine.c (make_compound_operation_int): Don't try to optimize
11969 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
11971 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11973 PR rtl-optimization/81423
11974 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
11975 with a constant that is -1 in the truncated to mode.
11977 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11979 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
11980 (determine_unlikely_bbs): ... here.
11981 * predict.h (propagate_unlikely_bbs_forward): Declare.
11982 * cfgexpand.c (pass_expand::execute): Use it.
11983 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
11985 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
11986 propagate_unlikely_bbs_forward.
11988 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11990 PR middle-end/81331
11991 * except.c (maybe_add_nop_after_section_switch): New function.
11994 2017-07-19 Tom de Vries <tom@codesourcery.com>
11996 * gimple.h (gimple_phi_set_arg): Make assert more strict.
11998 2017-07-19 Tom de Vries <tom@codesourcery.com>
12000 * gimple.h (gimple_phi_arg): Make assert more strict.
12002 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
12004 * config.gcc (powerpc*-*-*): Add mmintrin.h.
12005 * config/rs6000/mmintrin.h: New file.
12006 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
12008 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12010 PR tree-optimization/81346
12011 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
12013 2017-07-19 Tom de Vries <tom@codesourcery.com>
12015 * config/nvptx/nvptx.md (VECIM): Add V2DI.
12017 2017-07-19 Tom de Vries <tom@codesourcery.com>
12019 * config/nvptx/nvptx-modes.def: Add V2DImode.
12020 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
12021 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
12022 (nvptx_output_mov_insn): Handle lack of mov.b128.
12023 (nvptx_print_operand): Handle 'H' and 'L' codes.
12024 (nvptx_vector_mode_supported): Allow V2DImode.
12025 (nvptx_preferred_simd_mode): New function.
12026 (nvptx_data_alignment): New function.
12027 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
12028 nvptx_preferred_simd_mode.
12029 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
12031 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
12033 2017-07-19 Tom de Vries <tom@codesourcery.com>
12035 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
12036 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
12037 (nvptx_vector_mode_supported): New function. Allow V2SImode.
12038 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
12039 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
12040 (mov<VECIM>_insn): New define_insn.
12041 (define_expand "mov<VECIM>): New define_expand.
12043 2017-07-19 Tom de Vries <tom@codesourcery.com>
12045 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
12047 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12049 PR tree-optimization/81346
12050 * fold-const.h (fold_div_compare, range_check_type): Declare.
12051 * fold-const.c (range_check_type): New function.
12052 (build_range_check): Use range_check_type.
12053 (fold_div_compare): No longer static, rewritten into
12054 a match.pd helper function.
12055 (fold_comparison): Don't call fold_div_compare here.
12056 * match.pd (X / C1 op C2): New optimization using fold_div_compare
12057 as helper function.
12059 2017-07-19 Nathan Sidwell <nathan@acm.org>
12061 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
12062 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
12063 * tree.c (find_decls_types_r, verify_type): Use
12064 TYPE_{MIN,MAX}_VALUE_RAW.
12065 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
12066 (hash_tree): Likewise.
12067 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12069 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12072 2017-07-18 Tom de Vries <tom@codesourcery.com>
12074 PR middle-end/81464
12075 * omp-expand.c (expand_omp_for_static_chunk): Handle
12076 equal-argument loop exit phi.
12078 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
12081 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
12082 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
12083 operand 2 predicate.
12084 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
12085 operand 2 predicate.
12086 (ror,rol -> rorx splitters): Use const_int_operand as
12087 operand 2 predicate.
12089 2017-06-18 Richard Biener <rguenther@suse.de>
12091 PR tree-optimization/81410
12092 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
12093 the gap in the ! slp_perm SLP case after each group.
12095 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12097 PR middle-end/81463
12098 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
12101 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12103 PR middle-end/81462
12104 * predict.c (set_even_probabilities): Cleanup; do not affect
12105 probabilities that are already known.
12106 (combine_predictions_for_bb): Call even when count is set.
12108 2017-07-18 Nathan Sidwell <nathan@acm.org>
12110 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
12113 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12116 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
12117 optimization for loop niter analysis.
12119 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
12122 * config/avr/avr.c (avr_optimize_casesi): Don't use
12123 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
12125 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12127 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
12128 body_cost_vec from _vect_peel_extended_info.
12129 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
12130 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
12133 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12135 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
12137 2017-07-18 Richard Biener <rguenther@suse.de>
12139 PR tree-optimization/80620
12140 PR tree-optimization/81403
12141 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
12142 info when re-using a VN table entry.
12144 2017-07-18 Richard Biener <rguenther@suse.de>
12146 PR tree-optimization/81418
12147 * tree-vect-loop.c (vectorizable_reduction): Properly compute
12148 vectype_in. Verify that with lane-reducing reduction operations
12149 we have a single def-use cycle.
12151 2017-07-17 Carl Love <cel@us.ibm.com>
12153 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
12155 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12156 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12157 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12158 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12159 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12160 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12161 VMULOSW): New enum "unspec" values.
12162 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12163 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12164 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12165 altivec_vmulosw): New patterns.
12166 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12167 VMULOSW): Add definitions.
12168 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
12170 * config/alpha/alpha.c: Include predict.h.
12172 2017-07-17 Yury Gribov <tetra2005@gmail.com>
12174 * tree-vrp.c (compare_assert_loc): Fix comparison function
12175 to return predictable results.
12177 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12179 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
12181 (subdi3): Likewise.
12182 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
12183 * doc/invoke.texi (mexpand-adddi): Update text.
12185 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12187 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
12188 that also clobbers the CC register. The old expand code is moved
12190 (*arc_clzsi2): ... here.
12191 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
12192 the CC register. The old expand code is moved to ...
12193 (arc_ctzsi2): ... here.
12195 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12197 * config/arc/arc.opt (mindexed-loads): Use initial value
12198 TARGET_INDEXED_LOADS_DEFAULT.
12199 (mauto-modify-reg): Use initial value
12200 TARGET_AUTO_MODIFY_REG_DEFAULT.
12201 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12202 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12203 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12204 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12206 2017-07-17 Martin Liska <mliska@suse.cz>
12209 * opts.c (finish_options): Do not allow -fgnu-tm
12210 w/ -fsanitize={kernel-,}address. Say sorry.
12212 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12215 * tree-loop-distribution.c (classify_partition): Only assert on
12216 numer of iterations.
12217 (merge_dep_scc_partitions): Delete prameter. Update function call.
12218 (distribute_loop): Remove code handling loop with unknown niters.
12219 (pass_loop_distribution::execute): Skip loop with unknown niters.
12221 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12224 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
12225 function sort_partitions_by_post_order.
12227 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12229 PR tree-optimization/81374
12230 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
12231 the max index of basic blocks, rather than number of basic blocks.
12233 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12235 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
12237 (arc_legitimate_pic_operand_p): Likewise.
12238 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
12240 (arc_needs_pcl_p): Likewise.
12241 (arc_legitimate_pc_offset_p): Likewise.
12242 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
12243 function is also used in constrains.md.
12244 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
12245 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
12246 PLUS. Only return true/false in known cases, otherwise assert.
12247 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
12248 is already called in arc_legitimate_constant_p.
12249 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
12251 (LEGITIMATE_PIC_OPERAND_P): Use
12252 arc_raw_symbolic_reference_mentioned_p function.
12253 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
12258 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12259 Andrew Burgess <andrew.burgess@embecosm.com>
12261 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
12262 (arc_return_address_register): New function.
12263 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
12264 (arc_handle_fndecl_attribute): Add naked attribute.
12265 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12266 (TARGET_WARN_FUNC_RETURN): Likewise.
12267 (arc_allocate_stack_slots_for_args): New function.
12268 (arc_warn_func_return): Likewise.
12269 (machine_function): Change type fn_type.
12270 (arc_compute_function_type): Consider new naked function type,
12271 change function return type.
12272 (arc_must_save_register): Adapt to handle new
12273 arc_compute_function_type's return type.
12274 (arc_expand_prologue): Likewise.
12275 (arc_expand_epilogue): Likewise.
12276 (arc_return_address_regs): Delete.
12277 (arc_return_address_register): New function.
12278 (arc_epilogue_uses): Use above function.
12279 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
12280 (arc_function_type): Change encoding, add naked type.
12281 (ARC_INTERRUPT_P): Change to handle the new encoding.
12282 (ARC_FAST_INTERRUPT_P): Likewise.
12283 (ARC_NORMAL_P): Define.
12284 (ARC_NAKED_P): Likewise.
12285 (arc_compute_function_type): Delete prototype.
12286 * config/arc/arc.md (in_ret_delay_slot): Use
12287 arc_return_address_register function.
12288 (simple_return): Likewise.
12289 (p_return_i): Likewise.
12291 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12293 PR tree-optimization/81428
12294 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
12295 can't be built for those types.
12297 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12299 Remove stuff dead since r239246.
12301 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
12302 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
12303 (avr_inform_devices): Remove dead stuff.
12305 2017-07-17 Tamar Christina <tamar.christina@arm.com>
12307 * config/arm/arm_neon.h: Fix softp typo.
12309 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12311 PR tree-optimization/81365
12312 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
12313 aggregate moves onto bb predecessor edges, make sure there are no
12314 loads that could alias the lhs in between the start of bb and the
12317 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12320 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
12321 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
12322 [LSHIFTRT, outer_code = TRUNCATE]: Same.
12324 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12326 PR tree-optimization/81396
12327 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
12328 (init_symbolic_number): Initialize it to 1.
12329 (perform_symbolic_merge): Add n_ops from both operands into the new
12331 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
12332 without base_addr as useless if they need more than one operation.
12333 (bswap_replace): Handle !bswap case for NULL base_addr.
12335 2017-07-17 Tom de Vries <tom@codesourcery.com>
12338 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
12341 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
12343 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
12344 conditional builtin define __FIX_LEON3FT_B2BST.
12346 2017-07-17 Daniel Cederman <cederman@gaisler.com>
12348 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
12349 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
12352 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12354 PR rtl-optimization/81424
12355 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
12356 to remove potential trapping from operands if -fnon-call-exceptions.
12358 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12360 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
12361 profile_proability for scalling.
12362 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
12364 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12366 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
12368 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12370 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
12371 fixpoint arithmetics.
12373 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12375 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
12376 fixpoint arithmetics.
12378 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12380 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
12381 fixpoint arithmetics.
12383 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12385 * profile-count.h (profile_probability::from_reg_br_prob_note,
12386 profile_probability::to_reg_br_prob_note): New functions.
12387 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
12388 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
12389 * predict.c (probability_reliable_p): Update.
12390 (edge_probability_reliable_p): Update.
12391 (br_prob_note_reliable_p): Update.
12392 (invert_br_probabilities): Update.
12393 (add_reg_br_prob_note): New function.
12394 (combine_predictions_for_insn): Update.
12395 * asan.c (asan_clear_shadow): Update.
12396 * cfgbuild.c (compute_outgoing_frequencies): Update.
12397 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12398 (update_br_prob_note): Update.
12399 (rtl_verify_edges): Update.
12400 (purge_dead_edges): Update.
12401 (fixup_reorder_chain): Update.
12402 * emit-rtl.c (try_split): Update.
12403 * ifcvt.c (cond_exec_process_insns): Update.
12404 (cond_exec_process_if_block): Update.
12405 (dead_or_predicable): Update.
12406 * internal-fn.c (expand_addsub_overflow): Update.
12407 (expand_neg_overflow): Update.
12408 (expand_mul_overflow): Update.
12409 * loop-doloop.c (doloop_modify): Update.
12410 * loop-unroll.c (compare_and_jump_seq): Update.
12411 * optabs.c (emit_cmp_and_jump_insn_1): Update.
12412 * predict.h: Update.
12413 * reorg.c (mostly_true_jump): Update.
12415 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
12416 * config/alpha/alpha.c (emit_unlikely_jump): Update.
12417 * config/arc/arc.c: (emit_unlikely_jump): Update.
12418 * config/arm/arm.c: (emit_unlikely_jump): Update.
12419 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
12420 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
12421 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
12422 (ix86_print_operand): Update.
12423 (ix86_split_fp_branch): Update.
12424 (predict_jump): Update.
12425 * config/ia64/ia64.c (ia64_print_operand): Update.
12426 * config/mmix/mmix.c (mmix_print_operand): Update.
12427 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
12428 (rs6000_expand_split_stack_prologue): Update.
12429 * config/rs6000/rs6000.c: Update.
12430 * config/s390/s390.c (s390_expand_vec_strlen): Update.
12431 (s390_expand_vec_movstr): Update.
12432 (s390_expand_cs_tdsi): Update.
12433 (s390_expand_split_stack_prologue): Update.
12434 * config/sh/sh.c (sh_print_operand): Update.
12435 (expand_cbranchsi4): Update.
12436 (expand_cbranchdi4): Update.
12437 * config/sparc/sparc.c (output_v9branch): Update.
12438 * config/spu/spu.c (get_branch_target): Update.
12439 (ea_load_store_inline): Update.
12440 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
12441 * config/tilepro/tilepro.c: Update.
12443 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12445 * gimplify.c (mostly_copy_tree_r): Revert latest change.
12446 (gimplify_save_expr): Likewise.
12448 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12450 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
12452 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12454 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
12456 * timevar.def (TV_IPA_FNSUMMARY): Define.
12458 2017-07-16 Daniel Cederman <cederman@gaisler.com>
12460 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
12461 to back store errata sensitive sequence from being generated.
12462 (sqrtdf2_fix): Likewise.
12464 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12466 * tree-ssa-threadupdate.c (compute_path_counts,
12467 update_joiner_offpath_counts): Use profile_probability.
12469 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
12472 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12474 * config/arm/arm-c.c (arm_cpu_builtins): Define
12475 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12477 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
12479 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12480 array entries to represent __ieee128 versions of the
12481 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
12482 scalar_extract_sig, and scalar_insert_exp built-in functions.
12483 (altivec_resolve_overloaded_builtin): Add special case handling
12484 for the __builtin_scalar_insert_exp function, as represented by
12485 the P9V_BUILTIN_VEC_VSIEDP constant.
12486 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
12487 exponent support for __ieee128 argument.
12488 (VSESQP): Add scalar extract signature support for __ieee128
12490 (VSTDCNQP): Add scalar test negative support for __ieee128
12492 (VSIEQP): Add scalar insert exponent support for __int128 argument
12493 with __ieee128 result.
12494 (VSIEQPF): Add scalar insert exponent support for __ieee128
12495 argument with __ieee128 result.
12496 (VSTDCQP): Add scalar test data class support for __ieee128
12498 (VSTDCNQP): Add overload support for scalar test negative with
12499 __ieee128 argument.
12500 (VSTDCQP): Add overload support for scalar test data class
12501 __ieee128 argument.
12502 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
12503 UNSPEC_VSX_SXSIGDP.
12504 (UNSPEC_VSX_SIEXPQP): New constant.
12505 (xsxexpqp): New insn for VSX scalar extract exponent quad
12507 (xsxsigqp): New insn for VSX scalar extract significand quad
12509 (xsiexpqpf): New insn for VSX scalar insert exponent quad
12510 precision with floating point argument.
12511 (xststdcqp): New expand for VSX scalar test data class quad
12513 (xststdcnegqp): New expand for VSX scalar test negative quad
12515 (xststdcqp): New insn to match expansions for VSX scalar test data
12516 class quad precision and VSX scalar test negative quad precision.
12517 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
12518 special case operand checking to enforce that second operand of
12519 VSX scalar test data class with quad precision argument is a 7-bit
12521 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12522 prototypes and descriptions of __ieee128 versions of
12523 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
12524 scalar_test_data_class, and scalar_test_neg built-in functions.
12526 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12528 PR tree-optimization/81162
12529 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
12530 replace a negate with an add.
12532 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
12534 * doc/invoke.texi (arm/-mcpu): Document +crypto.
12536 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12538 * config/arm/arm-c.c (arm_cpu_builtins): Define
12539 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12541 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12543 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
12544 (armv8-r): Set ARM Cortex-R52 as default CPU.
12545 * config/arm/arm-tables.opt: Regenerate.
12546 * config/arm/arm-tune.md: Regenerate.
12547 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12549 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
12550 extension for -mcpu=cortex-r52.
12552 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12554 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
12555 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
12556 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
12557 (fp-armv8): Define it as FP_ARMv8 only.
12558 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
12559 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
12561 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
12562 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
12563 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
12564 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
12565 than TARGET_FPU_ARMV8.
12566 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
12567 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
12568 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
12570 * config/arm/neon.md (neon_vrint): Likewise.
12571 (neon_vcvt): Likewise.
12572 (neon_<fmaxmin_op><mode>): Likewise.
12573 (<fmaxmin><mode>3): Likewise.
12574 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
12575 * config/arm/predicates.md (arm_cond_move_operator): Check against
12576 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
12578 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
12580 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
12581 to top of function.
12583 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12585 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
12586 loop in comment with memset.
12588 2017-07-14 Martin Liska <mliska@suse.cz>
12590 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
12591 * dwarf2out.c (is_java): Remove the function.
12592 (output_pubname): Remove usage of the function.
12593 (lower_bound_default): Remove usage of DW_LANG_Java.
12594 (gen_compile_unit_die): Likewise.
12595 * gcc.c: Remove compiler defaults for .java and .zip files.
12596 * gimple-expr.c (remove_suffix): Change as there's no longer
12597 extension than 4-letter one.
12598 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
12599 (gimplify_save_expr): Likewise.
12600 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
12601 as it's possible even for other languages than Java.
12602 * langhooks.h (struct lang_hooks): Remove Java from a comment.
12603 * lto-opts.c (lto_write_options): Remove reference to Java.
12604 * opts.c (strip_off_ending): Update file extension handling.
12605 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
12606 * tree-eh.c (lower_resx): Likewise.
12607 * tree.c (free_lang_data_in_type): Remove dead code.
12608 (find_decls_types_r): Likewise.
12609 (build_common_builtin_nodes): Remove Java from a comment.
12610 (verify_type): Remove dead code.
12611 * varasm.c (assemble_external): Remove Java from a comment.
12613 2017-07-14 Martin Liska <mliska@suse.cz>
12615 * opts.c (finish_options): Add quotes.
12616 (common_handle_option): Likewise.
12618 2017-07-14 Martin Liska <mliska@suse.cz>
12620 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
12621 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
12622 Remove N_SO_PASCAL.
12623 * dwarf2out.c (lower_bound_default): Do not handle
12625 (gen_compile_unit_die): Likewise.
12626 * gcc.c: Remove default extension binding for GNU Pascal.
12627 * stmt.c: Remove Pascal language from a comment.
12628 * xcoffout.c: Likewise.
12630 2017-07-13 David Malcolm <dmalcolm@redhat.com>
12633 * diagnostic-show-locus.c (fixit_cmp): New function.
12634 (layout::layout): Sort m_fixit_hints.
12635 (column_range::column_range): Assert that the values are valid.
12636 (struct char_span): New struct.
12637 (correction::overwrite): New method.
12638 (struct source_line): New struct.
12639 (line_corrections::add_hint): Add assertions. Reimplement memcpy
12640 calls in terms of classes source_line and char_span, and
12641 correction::overwrite.
12642 (selftest::test_overlapped_fixit_printing_2): New function.
12643 (selftest::diagnostic_show_locus_c_tests): Call it.
12645 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
12647 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
12648 early if there is no lhs.
12650 2017-07-13 Martin Liska <mliska@suse.cz>
12652 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
12653 (gen_reference_type_die): Likewise.
12654 * stor-layout.c: Remove Pascal-related comment.
12656 2017-07-13 Martin Liska <mliska@suse.cz>
12658 * opts.c (finish_options): Add quotes to error messages.
12659 (parse_sanitizer_options): Likewise.
12661 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
12663 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
12665 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
12667 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
12669 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
12671 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
12673 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
12675 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12678 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
12679 provides the hardware capability bits, define the macro
12680 __BUILTIN_CPU_SUPPORTS__.
12681 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
12682 if GLIBC does not provide the hardware capability bits. Add a
12683 gcc_unreachable call if the built-in cpu function is neither
12684 __builtin_cpu_is nor __builtin_cpu_supports.
12685 (rs6000_get_function_versions_dispatcher): Change the warning
12686 that an old GLIBC is used which does not export the capability
12687 bits to be an error.
12688 * doc/extend.texi (target_clones attribute): Document the
12689 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
12690 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
12691 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
12692 the macros defined by GCC if the newer GLIBC is available.
12694 2017-07-12 Jeff Law <law@redhat.com>
12696 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
12697 remaining includes slightly.
12698 * config/riscv/riscv-builtins.c: Include profile-count.h.
12700 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12703 * config/avr/avr.c (avr_set_current_function): In diagnostic
12704 messages: Quote keywords and (parts of) identifiers.
12705 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
12708 2017-07-12 Carl Love <cel@us.ibm.com>
12710 * config/rs6000/rs6000-c.c: Add support for built-in functions
12711 vector bool char vec_revb (vector bool char);
12712 vector bool short vec_revb (vector short char);
12713 vector bool int vec_revb (vector bool int);
12714 vector bool long long vec_revb (vector bool long long);
12715 * doc/extend.texi: Update the built-in documentation file for the
12716 new built-in functions.
12718 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12720 * config/s390/s390.md: Remove movcc splitter.
12722 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12724 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
12725 load/store on condition.
12727 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12730 * config/avr/avr.c (avr_encode_section_info)
12731 [progmem && !TREE_READONLY]: Error if progmem object needs
12734 2017-07-11 Michael Collison <michael.collison@arm.com>
12736 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
12739 2017-07-11 Carl Love <cel@us.ibm.com>
12741 * config/rs6000/rs6000-c.c: Add support for builtins
12742 vector unsigned int vec_parity_lsbb (vector signed int);
12743 vector unsigned int vec_parity_lsbb (vector unsigned int);
12744 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
12745 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
12746 vector unsigned long long vec_parity_lsbb (vector signed long long);
12747 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
12748 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
12749 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
12750 * doc/extend.texi: Update the built-in documentation file for the
12751 new built-in functions.
12753 2017-07-11 David Malcolm <dmalcolm@redhat.com>
12755 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
12756 (layout::m_primary_loc): New field.
12757 (layout::layout): Initialize new field. Move location filtering
12758 logic from here to...
12759 (layout::maybe_add_location_range): ...this new method. Add
12760 support for filtering to just the lines already specified by other
12762 (layout::will_show_line_p): New method.
12763 (gcc_rich_location::add_location_if_nearby): New method.
12764 (selftest::test_add_location_if_nearby): New test function.
12765 (selftest::diagnostic_show_locus_c_tests): Call it.
12766 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
12769 2017-07-11 Tom de Vries <tom@codesourcery.com>
12771 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
12772 (bb_first_real_insn): New function.
12773 (nvptx_single): Add extra initialization of broadcasted condition
12776 2017-07-11 Nathan Sidwell <nathan@acm.org>
12778 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
12780 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
12782 * doc/extend.texi (AVR Function Attributes): Remove weblink to
12783 Binutils doc as TEXI will mess them up.
12784 * doc/invoke.texi (AVR Options): Same here.
12786 2017-07-11 Daniel Cederman <cederman@gaisler.com>
12788 * config/sparc/sparc.opt (mfix-ut700): New option.
12789 (mfix-gr712rc): Likewise.
12790 (sparc_fix_b2bst): New variable.
12791 * doc/invoke.texi (SPARC options): Document them.
12792 (ARM options): Fix warnings.
12793 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
12794 instructions to prevent sequences that can trigger the store-store
12795 errata for certain LEON3FT processors.
12796 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
12797 (sparc_option_override): Set sparc_fix_b2bst appropriately.
12798 * config/sparc/sparc.md (fix_b2bst): New attribute.
12799 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
12801 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
12804 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
12806 (*rsqrtsf2_sse): Ditto.
12808 (div<mode>3): Macroize insn from divdf3 and divsf3
12809 using MODEF mode iterator.
12811 2017-07-10 Martin Sebor <msebor@redhat.com>
12813 PR tree-optimization/80397
12814 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
12815 instead of testing for equality to INTEGER_TYPE.
12817 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
12819 * config.gcc: Remove uclibc from arc target spec.
12821 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
12823 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
12825 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12828 * lto-wrapper.c (remove_option): New function.
12829 (merge_and_complain): Merge PIC/PIE options more realistically.
12831 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12833 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
12837 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
12838 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
12839 * config.in: Regenerate.
12840 * configure: Regenerate.
12841 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
12842 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
12843 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
12844 (TARGET_GASISR_PROLOGUES): ...target mask.
12845 * common/config/avr/avr-common.c
12846 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
12847 Set -mgas-isr-prologues.
12848 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
12849 INSERT_PASS_BEFORE for it.
12850 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
12851 * config/avr/avr.c (avr_option_override)
12852 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
12853 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
12854 (avr_attribute_table) <no_gccisr>: Add new function attribute.
12855 (avr_set_current_function) <is_no_gccisr>: Init machine field.
12856 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
12858 (make_avr_pass_pre_proep): New function.
12859 (emit_push_sfr) <treg>: Add argument to function and use it
12860 instead of TMP_REG.
12861 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
12862 and set machine->gasisr.yes.
12863 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
12864 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
12865 __gcc_isr.n_pushed to .L__stack_usage.
12866 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
12867 (avr_asm_final_postscan_insn): ...this new static function.
12868 * config/avr/avr.h (machine_function)
12869 <is_no_gccisr, use_L__stack_usage>: New fields.
12870 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
12871 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
12872 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
12873 (gasisr, *gasisr): New expander and insn.
12874 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12875 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
12876 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
12878 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
12880 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
12883 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12885 Move jump-tables out of .text again.
12888 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
12889 (ASM_OUTPUT_ADDR_VEC): New function.
12890 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
12891 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
12892 INSN_ADDRESSes as asm comment.
12893 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
12894 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
12895 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
12896 * config/avr/avr.md (*tablejump): Adjust comment.
12897 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
12898 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
12900 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
12901 (avr_output_addr_vec): New proto.
12902 (avr_log_t) <insn_addresses>: New field.
12904 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
12907 * config/i386/i386.c (ix86_function_arg_advance): Set
12908 outgoing_args_on_stack to true if there are outgoing arguments
12910 (ix86_function_arg): Likewise.
12911 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
12912 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
12913 * config/i386/i386.h (machine_function): Add
12914 outgoing_args_on_stack.
12916 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
12918 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12919 supporting pthreds.
12920 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
12922 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
12924 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
12925 (REAL_H): Remove $(MACHMODE_H).
12926 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
12928 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
12929 $(MACHMODE_H) and double-int.h.
12930 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
12932 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
12935 2017-07-07 Andrew Pinski <apinski@cavium.com>
12937 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
12938 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
12940 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12942 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
12943 Add warning if GCC was not configured to link against a GLIBC that
12944 exports the hardware capability bits.
12945 (make_resolver_func): Make resolver function private and not a
12946 COMDAT function. Create the name with clone_function_name instead
12947 of make_unique_name.
12950 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
12951 correct operand in doing the split.
12953 2017-07-07 Carl Love <cel@us.ibm.com>
12955 * config/rs6000/rs6000-c: Add support for built-in function
12956 vector unsigned short vec_pack_to_short_fp32 (vector float,
12958 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
12959 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
12960 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
12961 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
12962 (convert_4f32_8i16): Add define_expand.
12963 * doc/extend.texi: Update the built-in documentation file for the
12964 new built-in function.
12966 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12968 * config/sparc/m8.md: New file.
12969 * config/sparc/sparc.md: Include m8.md.
12971 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12973 * config/sparc/sparc.opt: New option -mvis4b.
12974 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
12975 (sparc_option_override): Handle VIS4B.
12976 (enum sparc_builtins): Define
12977 SPARC_BUILTIN_DICTUNPACK{8,16,32},
12978 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
12979 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
12980 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
12981 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
12982 (check_constant_argument): New function.
12983 (sparc_vis_init_builtins): Define builtins
12984 __builtin_vis_dictunpack{8,16,32},
12985 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
12986 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
12987 __builtin_vis_fpcmpde{8,16,32}shl and
12988 __builtin_vis_fpcmpur{8,16,32}shl.
12989 (sparc_expand_builtin): Check that the constant operands to
12990 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
12991 constant and in range.
12992 * config/sparc/sparc-c.c (sparc_target_macros): Handle
12994 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
12995 (SPARC_IMM5_P): Likewise.
12996 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
12997 (enabled): Handle vis4b.
12998 (UNSPEC_DICTUNPACK): New unspec.
12999 (UNSPEC_FPCMPSHL): Likewise.
13000 (UNSPEC_FPUCMPSHL): Likewise.
13001 (UNSPEC_FPCMPDESHL): Likewise.
13002 (UNSPEC_FPCMPURSHL): Likewise.
13003 (cpu_feature): New CPU feature `vis4b'.
13004 (dictunpack{8,16,32}): New insns.
13005 (FPCSMODE): New mode iterator.
13006 (fpcscond): New code iterator.
13007 (fpcsucond): Likewise.
13008 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
13009 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
13010 (fpcmpde{8,16,32}{si,di}shl): Likewise.
13011 (fpcmpur{8,16,32}{si,di}shl): Likewise.
13012 * config/sparc/constraints.md: Define constraints `q' for unsigned
13013 2-bit integer constants and `t' for unsigned 5-bit integer
13015 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
13017 (imm5_operand_dictunpack16): Likewise.
13018 (imm5_operand_dictunpack32): Likewise.
13019 (imm2_operand): Likewise.
13020 * doc/invoke.texi (SPARC Options): Document -mvis4b.
13021 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
13022 ditunpack* and fpcmp*shl builtins.
13024 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13026 * config.gcc: Handle m8 in --with-{cpu,tune} options.
13027 * config.in: Add HAVE_AS_SPARC6 define.
13028 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
13030 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
13032 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
13033 (CPP_CPU_SPEC): Handle m8.
13034 (ASM_CPU_SPEC): Likewise.
13035 * config/sparc/sparc-opts.h (enum processor_type): Add
13037 * config/sparc/sparc.c (m8_costs): New struct.
13038 (sparc_option_override): Handle TARGET_CPU_m8.
13039 (sparc32_initialize_trampoline): Likewise.
13040 (sparc64_initialize_trampoline): Likewise.
13041 (sparc_issue_rate): Likewise.
13042 (sparc_register_move_cost): Likewise.
13043 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
13044 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
13045 (ASM_CPU64_DEFAULT_SPEC): Likewise.
13046 (CPP_CPU_SPEC): Handle M8.
13047 (ASM_CPU_SPEC): Likewise.
13048 (AS_M8_FLAG): Define.
13049 * config/sparc/sparc.md: Add m8 to the cpu attribute.
13050 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
13051 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
13053 * configure: Regenerate.
13054 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
13057 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13059 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
13061 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
13062 ("*movdi_insn_sp32"): Do not set v3pipe.
13063 ("*movsi_insn"): Likewise.
13064 ("*movdi_insn_sp64"): Likewise.
13065 ("*movsf_insn"): Likewise.
13066 ("*movdf_insn_sp32"): Likewise.
13067 ("*movdf_insn_sp64"): Likewise.
13068 ("*zero_extendsidi2_insn_sp64"): Likewise.
13069 ("*sign_extendsidi2_insn"): Likewise.
13070 ("*mov<VM32:mode>_insn"): Likewise.
13071 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13072 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13073 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13074 ("<vlop:code><VL:mode>3"): Likewise.
13075 ("*not_<vlop:code><VL:mode>3"): Likewise.
13076 ("*nand<VL:mode>_vis"): Likewise.
13077 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
13078 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
13079 ("one_cmpl<VL:mode>2"): Likewise.
13080 ("faligndata<VM64:mode>_vis"): Likewise.
13081 ("alignaddrsi_vis"): Likewise.
13082 ("alignaddrdi_vis"): Likweise.
13083 ("alignaddrlsi_vis"): Likewise.
13084 ("alignaddrldi_vis"): Likewise.
13085 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13086 ("bmaskdi_vis"): Likewise.
13087 ("bmasksi_vis"): Likewise.
13088 ("bshuffle<VM64:mode>_vis"): Likewise.
13089 ("cmask8<P:mode>_vis"): Likewise.
13090 ("cmask16<P:mode>_vis"): Likewise.
13091 ("cmask32<P:mode>_vis"): Likewise.
13092 ("pdistn<P:mode>_vis"): Likewise.
13093 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13095 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13097 * config/sparc/sparc.md ("subtype"): New insn attribute.
13098 ("*wrgsr_sp64"): Set insn subtype.
13099 ("*rdgsr_sp64"): Likewise.
13100 ("alignaddrsi_vis"): Likewise.
13101 ("alignaddrdi_vis"): Likewise.
13102 ("alignaddrlsi_vis"): Likewise.
13103 ("alignaddrldi_vis"): Likewise.
13104 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13105 ("fexpand_vis"): Likewise.
13106 ("fpmerge_vis"): Likewise.
13107 ("faligndata<VM64:mode>_vis"): Likewise.
13108 ("bshuffle<VM64:mode>_vis"): Likewise.
13109 ("cmask8<P:mode>_vis"): Likewise.
13110 ("cmask16<P:mode>_vis"): Likewise.
13111 ("cmask32<P:mode>_vis"): Likewise.
13112 ("fchksm16_vis"): Likewise.
13113 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
13114 ("fmean16_vis"): Likewise.
13115 ("fp<plusminus_insn>64_vis"): Likewise.
13116 ("<plusminus_insn>v8qi3"): Likewise.
13117 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13118 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
13119 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
13120 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
13121 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
13122 ("*movqi_insn"): Likewise.
13123 ("*movhi_insn"): Likewise.
13124 ("*movsi_insn"): Likewise.
13125 ("movsi_pic_gotdata_op"): Likewise.
13126 ("*movdi_insn_sp32"): Likewise.
13127 ("*movdi_insn_sp64"): Likewise.
13128 ("movdi_pic_gotdata_op"): Likewise.
13129 ("*movsf_insn"): Likewise.
13130 ("*movdf_insn_sp32"): Likewise.
13131 ("*movdf_insn_sp64"): Likewise.
13132 ("*zero_extendhisi2_insn"): Likewise.
13133 ("*zero_extendqihi2_insn"): Likewise.
13134 ("*zero_extendqisi2_insn"): Likewise.
13135 ("*zero_extendqidi2_insn"): Likewise.
13136 ("*zero_extendhidi2_insn"): Likewise.
13137 ("*zero_extendsidi2_insn_sp64"): Likewise.
13138 ("ldfsr"): Likewise.
13139 ("prefetch_64"): Likewise.
13140 ("prefetch_32"): Likewise.
13141 ("tie_ld32"): Likewise.
13142 ("tie_ld64"): Likewise.
13143 ("*tldo_ldub_sp32"): Likewise.
13144 ("*tldo_ldub1_sp32"): Likewise.
13145 ("*tldo_ldub2_sp32"): Likewise.
13146 ("*tldo_ldub_sp64"): Likewise.
13147 ("*tldo_ldub1_sp64"): Likewise.
13148 ("*tldo_ldub2_sp64"): Likewise.
13149 ("*tldo_ldub3_sp64"): Likewise.
13150 ("*tldo_lduh_sp32"): Likewise.
13151 ("*tldo_lduh1_sp32"): Likewise.
13152 ("*tldo_lduh_sp64"): Likewise.
13153 ("*tldo_lduh1_sp64"): Likewise.
13154 ("*tldo_lduh2_sp64"): Likewise.
13155 ("*tldo_lduw_sp32"): Likewise.
13156 ("*tldo_lduw_sp64"): Likewise.
13157 ("*tldo_lduw1_sp64"): Likewise.
13158 ("*tldo_ldx_sp64"): Likewise.
13159 ("*mov<VM32:mode>_insn"): Likewise.
13160 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13161 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13163 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13165 * config/sparc/sparc.md ("type"): New insn type viscmp.
13166 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
13168 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
13169 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
13170 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13171 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
13173 ("n7_vis_logical_11cycle"): Likewise.
13174 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
13175 * config/sparc/niagara2.md ("niag3_vis": Likewise.
13176 * config/sparc/niagara.md ("niag_vis"): Likewise.
13177 * config/sparc/ultra3.md ("us3_fga"): Likewise.
13178 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
13180 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13182 * config/sparc/sparc.md: New instruction type `bmask'.
13183 (bmaskdi_vis): Use the `bmask' type.
13184 (bmasksi_vis): Likewise.
13185 * config/sparc/ultra3.md (us3_array): Likewise.
13186 * config/sparc/niagara7.md (n7_array): Likewise.
13187 * config/sparc/niagara4.md (n4_array): Likewise.
13188 * config/sparc/niagara2.md (niag2_vis): Likewise.
13189 (niag3_vis): Likewise.
13190 * config/sparc/niagara.md (niag_vis): Likewise.
13192 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13194 * ipa-comdats.c: Remove optimize check from gate.
13195 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
13196 for functions not optimized.
13197 (ipa_fn_summary_read): Skip optimize check.
13198 (ipa_fn_summary_write): Likewise.
13199 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
13201 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
13203 (can_inline_edge_p): Check flag_pcc_struct_return for match.
13204 (check_callers): Give up on caller which is not optimized.
13205 (inline_small_functions): Likewise.
13206 (ipa_inline): Do not give up when not optimizing.
13207 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
13208 away unoptimizes cdtors.
13209 (whole_program_function_and_variable_visibility): Do
13210 ipa_discover_readonly_nonaddressable_vars in LTO mode.
13211 * ipa.c (process_references): Do not check optimize.
13212 (symbol_table::remove_unreachable_nodes): Update optimize check.
13213 (set_writeonly_bit): Update optimize check.
13214 (pass_ipa_cdtor_merge::gate): Do not check optimize.
13215 (pass_ipa_single_use::gate): Remove.
13217 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13219 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
13220 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
13221 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
13222 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
13223 permute_load, permute_store, adjust_extract, adjust_splat,
13224 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
13225 replace_swap_with_copy, dump_swap_insn_table,
13226 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
13227 recombine_lvx_pattern, recombine_stvx_pattern,
13228 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
13229 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
13230 to file rs6000-p8swap.c.
13231 * config/rs6000/rs6000-p8swap.c: New file.
13232 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
13233 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
13234 and rs6000*-*-* targets.
13236 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13238 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
13240 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13242 * lto-wrapper.c (merge_and_complain): Do not merge
13243 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
13244 fsigned_zeros, ftrapping_math, fwrapv.
13245 (append_compiler_options): Do not track these options.
13246 (append_linker_options): Likewie
13248 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13250 * cgraphunit.c (cgraph_node::finalize_function): When
13251 !flag_toplevel_reorde set no_reorder flag.
13252 (varpool_node::finalize_decl): Likewise.
13253 (symbol_table::compile): Drop no toplevel reorder path.
13255 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13257 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
13258 edges; zero probability is not better than uninitialized.
13260 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13262 * asan.h (asan_sanitize_allocas_p): Declare.
13263 * asan.c (asan_sanitize_allocas_p): New function.
13264 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
13265 (handle_builtin_alloca): Likewise.
13266 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
13267 if !asan_sanitize_allocas_p.
13268 * params.def (asan-instrument-allocas): Add new option.
13269 * params.h (ASAN_PROTECT_ALLOCAS): Define.
13270 * opts.c (common_handle_option): Disable allocas sanitization for
13273 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13275 * asan.c: Include gimple-fold.h.
13276 (get_last_alloca_addr): New function.
13277 (handle_builtin_stackrestore): Likewise.
13278 (handle_builtin_alloca): Likewise.
13279 (asan_emit_allocas_unpoison): Likewise.
13280 (get_mem_refs_of_builtin_call): Add new parameter, remove const
13281 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
13282 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
13283 (instrument_builtin_call): Pass gimple iterator to
13284 get_mem_refs_of_builtin_call.
13285 (last_alloca_addr): New global.
13286 * asan.h (asan_emit_allocas_unpoison): Declare.
13287 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
13288 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
13289 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
13290 if function calls alloca.
13291 * gimple-fold.c (replace_call_with_value): Remove static keyword.
13292 * gimple-fold.h (replace_call_with_value): Declare.
13293 * internal-fn.c: Include asan.h.
13294 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
13295 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
13297 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13299 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
13300 (C_SELFTEST_FLAGS): New.
13301 (CPP_SELFTEST_FLAGS): New.
13302 (SELFTEST_DEPS): New, from deps of s-selftest.
13303 (C_SELFTEST_DEPS): New, from deps of s-selftest.
13304 (CPP_SELFTEST_DEPS): New.
13305 (selftest): Add dependency on s-selftest-c++.
13306 (s-selftest): Rename to...
13307 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
13308 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
13309 than SELFTEST_FLAGS.
13310 (selftest-gdb): Rename to...
13311 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
13313 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
13314 (selftest-valgrind): Rename to...
13315 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
13317 (selftest-valgrind): Reintroduce as an alias for
13318 selftest-c-valgrind.
13319 (s-selftest-c++): New.
13320 (selftest-c++-gdb): New.
13321 (selftest-c++-valgrind): New.
13323 2017-07-06 Olivier Hainque <hainque@adacore.com>
13325 * gcc.c (process_command): When deciding if undefined variables
13326 should be ignored when processing specs, accept "gcc -v" as well.
13328 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13330 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
13331 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
13333 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13335 * config/arm/arm-cpus.in (armv8-r): Add new entry.
13336 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
13337 * config/arm/arm-tables.opt: Regenerate.
13338 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
13340 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
13342 2017-07-06 Carl Love <cel@us.ibm.com>
13344 * ChangeLog: Clean up from mid air collision
13346 2017-07-06 Carl Love <cel@us.ibm.com>
13348 * config/rs6000/rs6000-c.c: Add support for built-in functions
13349 vector signed int vec_subc (vector signed int, vector signed int);
13350 vector signed __int128 vec_subc (vector signed __int128,
13351 vector signed __int128);
13352 vector unsigned __int128 vec_subc (vector unsigned __int128,
13353 vector unsigned __int128);
13354 vector signed int vec_sube (vector signed int, vector signed int,
13355 vector signed int);
13356 vector unsigned int vec_sube (vector unsigned int,
13357 vector unsigned int,
13358 vector unsigned int);
13359 vector signed __int128 vec_sube (vector signed __int128,
13360 vector signed __int128,
13361 vector signed__int128);
13362 vector unsigned __int128 vec_sube (vector unsigned __int128,
13363 vector unsigned __int128,
13364 vector unsigned __int128);
13365 vector signed int vec_subec (vector signed int, vector signed int,
13366 vector signed int);
13367 vector unsigned int vec_subec (vector unsigned int,
13368 vector unsigned int,
13369 vector unsigned int);
13370 vector signed __int128 vec_subec (vector signed __int128,
13371 vector signed __int128,
13372 vector signed__int128);
13373 vector unsigned __int128 vec_subec (vector unsigned __int128,
13374 vector unsigned __int128,
13375 vector unsigned __int128);
13376 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
13377 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
13378 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
13379 BU_ALTIVEC_OVERLOAD_X definitions.
13380 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
13381 * doc/extend.texi: Update the built-in documentation file for the new
13382 built-in functions.
13384 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13387 * diagnostic-show-locus.c (layout::layout): Use start and finish
13388 spelling location for the start and finish of each range.
13389 * genmatch.c (linemap_client_expand_location_to_spelling_point):
13390 Add unused aspect param.
13391 * input.c (expand_location_1): Add "aspect" param, and use it
13392 to access the correct part of the location.
13393 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
13395 (expand_location_to_spelling_point): Likewise.
13396 (linemap_client_expand_location_to_spelling_point): Add "aspect"
13397 param, and pass it to expand_location_1.
13399 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
13401 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
13402 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
13403 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
13404 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
13405 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
13406 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
13407 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
13408 _mm_maskz_getmant_ss): New intrinsics.
13409 (__builtin_ia32_getexpss128_mask): Changed to ...
13410 __builtin_ia32_getexpss128_round ... this.
13411 (__builtin_ia32_getexpsd128_mask): Changed to ...
13412 __builtin_ia32_getexpsd128_round ... this.
13413 * config/i386/i386-builtin-types.def
13414 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
13415 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
13416 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
13417 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
13418 __builtin_ia32_getmantss_mask_round): New builtins.
13419 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
13420 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
13421 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
13422 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
13423 * config/i386/sse.md
13424 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
13425 avx512f_sgetexp<mode><mask_scalar_name>
13426 <round_saeonly_scalar_name> ... this.
13427 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
13428 %0, %1, %2<round_saeonly_op3>}): Changed to ...
13429 vgetexp<ssescalarmodesuffix>
13430 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13431 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
13432 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
13433 avx512f_vgetmant<mode><mask_scalar_name>
13434 <round_saeonly_scalar_name> ... this.
13435 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
13436 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
13437 vgetmant<ssescalarmodesuffix>
13438 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
13439 %0<mask_scalar_operand4>, %1, %2
13440 <round_saeonly_scalar_mask_op4>, %3} ... this.
13441 * config/i386/subst.md (mask_scalar_operand4,
13442 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
13443 round_saeonly_scalar_nimm_predicate): New subst attributes.
13445 2017-07-06 Julia Koval <julia.koval@intel.com>
13447 * config/i386/i386.c (ix86_erase_embedded_rounding):
13448 Remove code for old rounding pattern.
13450 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
13452 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
13454 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
13456 * doc/sourcebuild.texi (Test Directives, Variants of
13457 dg-require-support): Add documentation for dg-require-stack-check.
13459 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
13461 * config/i386/subst.md (mask_scalar, round_scalar,
13462 round_saeonly_scalar): New meta-templates.
13463 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
13464 round_scalar_mask_operand3, round_scalar_mask_op3,
13465 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
13466 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
13467 round_saeonly_scalar_constraint,
13468 round_saeonly_scalar_prefix): New subst attribute.
13469 * config/i386/sse.md
13470 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
13471 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
13472 <round_scalar_name> ... this.
13473 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
13474 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
13475 <round_scalar_name> ... this.
13476 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
13477 <sse>_vm<code><mode>3<mask_scalar_name>
13478 <round_saeonly_scalar_name> ... this.
13479 (v<plusminus_mnemonic><ssescalarmodesuffix>
13480 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13481 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13482 v<plusminus_mnemonic><ssescalarmodesuffix>
13483 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13484 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13485 (v<multdiv_mnemonic><ssescalarmodesuffix>
13486 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13487 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13488 v<multdiv_mnemonic><ssescalarmodesuffix>
13489 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13490 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13491 (v<maxmin_float><ssescalarmodesuffix>
13492 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
13493 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
13494 v<maxmin_float><ssescalarmodesuffix>
13495 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13496 %0<mask_scalar_operand3>, %1, %<iptr>2
13497 <round_saeonly_scalar_mask_op3>} ... this.
13499 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
13501 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
13502 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
13504 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
13505 Alan Hayward <alan.hayward@arm.com>
13506 David Sherwood <david.sherwood@arm.com>
13508 * combine.c (simplify_if_then_else): Remove "enum" before
13510 * compare-elim.c (can_eliminate_compare): Likewise.
13511 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13513 (aarch64_lookup_simd_builtin_type): Likewise.
13514 (aarch64_simd_builtin_type): Likewise.
13515 (aarch64_init_simd_builtin_types): Likewise.
13516 (aarch64_simd_expand_args): Likewise.
13517 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
13519 (aarch64_reverse_mask): Likewise.
13520 (aarch64_simd_emit_reg_reg_move): Likewise.
13521 (aarch64_gen_adjusted_ldpstp): Likewise.
13522 (aarch64_ccmp_mode_to_code): Likewise.
13523 (aarch64_operands_ok_for_ldpstp): Likewise.
13524 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13525 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
13527 (aarch64_min_divisions_for_recip_mul): Likewise.
13528 (aarch64_reassociation_width): Likewise.
13529 (aarch64_get_condition_code_1): Likewise.
13530 (aarch64_simd_emit_reg_reg_move): Likewise.
13531 (aarch64_simd_attr_length_rglist): Likewise.
13532 (aarch64_reverse_mask): Likewise.
13533 (aarch64_operands_ok_for_ldpstp): Likewise.
13534 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13535 (aarch64_gen_adjusted_ldpstp): Likewise.
13536 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
13538 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
13539 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13540 (arm_lookup_simd_builtin_type): Likewise.
13541 (arm_simd_builtin_type): Likewise.
13542 (arm_init_simd_builtin_types): Likewise.
13543 (arm_expand_builtin_args): Likewise.
13544 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
13545 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
13546 (ft32_setup_incoming_varargs): Likewise.
13547 (ft32_function_arg): Likewise.
13548 (ft32_function_arg_advance): Likewise.
13549 (ft32_pass_by_reference): Likewise.
13550 (ft32_arg_partial_bytes): Likewise.
13551 (ft32_valid_pointer_mode): Likewise.
13552 (ft32_addr_space_pointer_mode): Likewise.
13553 (ft32_addr_space_legitimate_address_p): Likewise.
13554 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
13556 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
13557 (ix86_emit_outlined_ms2sysv_restore): Likewise.
13558 (iamcu_alignment): Likewise.
13559 (canonicalize_vector_int_perm): Likewise.
13560 (ix86_noce_conversion_profitable_p): Likewise.
13561 (ix86_mpx_bound_mode): Likewise.
13562 (ix86_operands_ok_for_move_multiple): Likewise.
13563 * config/microblaze/microblaze-protos.h
13564 (microblaze_expand_conditional_branch_reg): Likewise.
13565 * config/microblaze/microblaze.c
13566 (microblaze_expand_conditional_branch_reg): Likewise.
13567 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13569 (rs6000_reassociation_width): Likewise.
13570 (rs6000_invalid_binary_op): Likewise.
13571 (fusion_p9_p): Likewise.
13572 (emit_fusion_p9_load): Likewise.
13573 (emit_fusion_p9_store): Likewise.
13574 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
13576 (riscv_hard_regno_mode_ok_p): Likewise.
13577 (riscv_address_insns): Likewise.
13578 (riscv_split_symbol): Likewise.
13579 (riscv_legitimize_move): Likewise.
13580 (riscv_function_value): Likewise.
13581 (riscv_hard_regno_nregs): Likewise.
13582 (riscv_expand_builtin): Likewise.
13583 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
13584 (riscv_build_integer): Likewise.
13585 (riscv_split_integer): Likewise.
13586 (riscv_legitimate_constant_p): Likewise.
13587 (riscv_cannot_force_const_mem): Likewise.
13588 (riscv_regno_mode_ok_for_base_p): Likewise.
13589 (riscv_valid_base_register_p): Likewise.
13590 (riscv_valid_offset_p): Likewise.
13591 (riscv_valid_lo_sum_p): Likewise.
13592 (riscv_classify_address): Likewise.
13593 (riscv_legitimate_address_p): Likewise.
13594 (riscv_address_insns): Likewise.
13595 (riscv_load_store_insns): Likewise.
13596 (riscv_force_binary): Likewise.
13597 (riscv_split_symbol): Likewise.
13598 (riscv_force_address): Likewise.
13599 (riscv_legitimize_address): Likewise.
13600 (riscv_move_integer): Likewise.
13601 (riscv_legitimize_const_move): Likewise.
13602 (riscv_legitimize_move): Likewise.
13603 (riscv_address_cost): Likewise.
13604 (riscv_subword): Likewise.
13605 (riscv_output_move): Likewise.
13606 (riscv_canonicalize_int_order_test): Likewise.
13607 (riscv_emit_int_order_test): Likewise.
13608 (riscv_function_arg_boundary): Likewise.
13609 (riscv_pass_mode_in_fpr_p): Likewise.
13610 (riscv_pass_fpr_single): Likewise.
13611 (riscv_pass_fpr_pair): Likewise.
13612 (riscv_get_arg_info): Likewise.
13613 (riscv_function_arg): Likewise.
13614 (riscv_function_arg_advance): Likewise.
13615 (riscv_arg_partial_bytes): Likewise.
13616 (riscv_function_value): Likewise.
13617 (riscv_pass_by_reference): Likewise.
13618 (riscv_setup_incoming_varargs): Likewise.
13619 (riscv_print_operand): Likewise.
13620 (riscv_elf_select_rtx_section): Likewise.
13621 (riscv_save_restore_reg): Likewise.
13622 (riscv_for_each_saved_reg): Likewise.
13623 (riscv_register_move_cost): Likewise.
13624 (riscv_hard_regno_mode_ok_p): Likewise.
13625 (riscv_hard_regno_nregs): Likewise.
13626 (riscv_class_max_nregs): Likewise.
13627 (riscv_memory_move_cost): Likewise.
13628 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
13629 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
13630 (rl78_addr_space_address_mode): Likewise.
13631 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13633 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13634 (rs6000_reassociation_width): Likewise.
13635 (rs6000_invalid_binary_op): Likewise.
13636 (fusion_p9_p): Likewise.
13637 (emit_fusion_p9_load): Likewise.
13638 (emit_fusion_p9_store): Likewise.
13639 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
13640 (ok_for_simple_move_operands): Likewise.
13641 (ok_for_simple_move_strict_operands): Likewise.
13642 (ok_for_simple_arith_logic_operands): Likewise.
13643 (visium_legitimize_reload_address): Likewise.
13644 (visium_select_cc_mode): Likewise.
13645 (output_cbranch): Likewise.
13646 (visium_split_double_move): Likewise.
13647 (visium_expand_copysign): Likewise.
13648 (visium_expand_int_cstore): Likewise.
13649 (visium_expand_fp_cstore): Likewise.
13650 * config/visium/visium.c (visium_pass_by_reference): Likewise.
13651 (visium_function_arg): Likewise.
13652 (visium_function_arg_advance): Likewise.
13653 (visium_libcall_value): Likewise.
13654 (visium_setup_incoming_varargs): Likewise.
13655 (visium_legitimate_constant_p): Likewise.
13656 (visium_legitimate_address_p): Likewise.
13657 (visium_legitimize_address): Likewise.
13658 (visium_secondary_reload): Likewise.
13659 (visium_register_move_cost): Likewise.
13660 (visium_memory_move_cost): Likewise.
13661 (prepare_move_operands): Likewise.
13662 (ok_for_simple_move_operands): Likewise.
13663 (ok_for_simple_move_strict_operands): Likewise.
13664 (ok_for_simple_arith_logic_operands): Likewise.
13665 (visium_function_value_1): Likewise.
13666 (rtx_ok_for_offset_p): Likewise.
13667 (visium_legitimize_reload_address): Likewise.
13668 (visium_split_double_move): Likewise.
13669 (visium_expand_copysign): Likewise.
13670 (visium_expand_int_cstore): Likewise.
13671 (visium_expand_fp_cstore): Likewise.
13672 (visium_split_cstore): Likewise.
13673 (visium_select_cc_mode): Likewise.
13674 (visium_split_cbranch): Likewise.
13675 (output_cbranch): Likewise.
13676 (visium_print_operand_address): Likewise.
13677 * expmed.c (flip_storage_order): Likewise.
13678 * expmed.h (emit_cstore): Likewise.
13679 (flip_storage_order): Likewise.
13680 * genrecog.c (validate_pattern): Likewise.
13681 * hsa-gen.c (gen_hsa_addr): Likewise.
13682 * internal-fn.c (expand_arith_overflow): Likewise.
13683 * ira-color.c (allocno_copy_cost_saving): Likewise.
13684 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13685 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
13686 (process_invariant_for_inheritance): Likewise.
13687 * lra-eliminations.c (move_plus_up): Likewise.
13688 * omp-low.c (lower_oacc_reductions): Likewise.
13689 * simplify-rtx.c (simplify_subreg): Likewise.
13690 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
13691 (TARGET_CHKP_BOUND_MODE): Likewise..
13692 * targhooks.c (default_chkp_bound_mode): Likewise.
13693 (default_setup_incoming_vararg_bounds): Likewise.
13694 * targhooks.h (default_chkp_bound_mode): Likewise.
13695 (default_setup_incoming_vararg_bounds): Likewise.
13696 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
13697 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
13698 (have_whole_vector_shift): Likewise.
13699 * tree-vect-stmts.c (vectorizable_load): Likewise.
13700 * doc/tm.texi: Regenerate.
13702 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13704 Graceful degrade if Binutils PR21472 is not available.
13707 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
13708 .rodata in flash test fails.
13709 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
13710 * confgure: Regenerate.
13711 * config.in: Regenerate.
13712 * config/avr/avr.c (avr_asm_named_section)
13713 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
13714 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
13715 (avr_asm_init_sections): Same.
13717 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13719 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
13720 (fma<VH:mode>4_intrinsic): Likewise.
13721 (*fmsub<VCVTF:mode>4): Likewise.
13722 (*fmsub<VH:mode>4_intrinsic): Likewise.
13724 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13727 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
13728 Don't depend on "optimize > 0".
13729 (out_movhi_r_mr, out_movqi_mr_r): Same.
13730 (out_movhi_mr_r, out_movqi_r_mr): Same.
13731 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
13732 io_address_operand on "optimize > 0".
13734 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13736 * tree-loop-distribution.c: Add general explanantion on the pass.
13737 (generate_loops_for_partition): Mark distributed loop.
13738 (pg_add_dependence_edges): New parameter. Handle alias data
13739 dependence specially and record it in the parameter if asked.
13740 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
13741 (init_partition_graph_vertices, add_partition_graph_edge): New.
13742 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
13743 (free_partition_graph_vdata, build_partition_graph): New.
13744 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
13745 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
13746 (data_ref_segment_size, latch_dominated_by_data_ref): New.
13747 (compute_alias_check_pairs, version_loop_by_alias_check): New.
13748 (version_for_distribution_p, finalize_partitions): New.
13749 (distribute_loop): Handle alias data dependence specially. Factor
13750 out loop fusion code as functions and call these functions.
13752 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13754 * tree-loop-distribution.c (classify_partition): New parameter and
13755 better handle reduction statement.
13756 (rdg_build_partitions): Revise comment.
13757 (distribute_loop): Compute statements in all partitions and pass it
13758 to classify_partition.
13760 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13762 * tree-loop-distribution.c (enum partition_type): New.
13763 (struct partition): New field type.
13764 (partition_merge_into): Add parameter. Update partition type.
13765 (data_dep_in_cycle_p, update_type_for_merge): New functions.
13766 (build_rdg_partition_for_vertex): Compute partition type.
13767 (rdg_build_partitions): Dump partition type.
13768 (distribute_loop): Update calls to partition_merge_into.
13770 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13772 * tree-loop-distribution.c (struct ddr_hasher): New.
13773 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
13775 (classify_partition): Call get_data_dependence.
13776 (pg_add_dependence_edges): Ditto.
13777 (distribute_loop): Release data dependence hash table.
13779 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13781 * tree-loop-distribution.c (ref_base_address): Delete.
13782 (similar_memory_accesses): Rename ...
13783 (share_memory_accesses): ... to this. Check if partitions access
13784 the same memory reference.
13785 (distribute_loop): Call share_memory_accesses.
13787 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13789 * tree-loop-distribution.c (struct partition): New field recording
13790 its data reference.
13791 (partition_alloc, partition_free): Init and release data refs.
13792 (partition_merge_into): Merge data refs.
13793 (build_rdg_partition_for_vertex): Collect data refs for partition.
13794 (pg_add_dependence_edges): Change parameters from vector to bitmap.
13796 (distribute_loop): Remve data refs from vertice data of partition
13799 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13801 * tree-loop-distribution.c (params.h): Include header file.
13802 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
13803 (datarefs_vec): New global var.
13804 (create_rdg_vertices): Use datarefs_vec directly.
13805 (free_rdg): Don't free data references.
13806 (build_rdg): Update use. Don't free data references.
13807 (distribute_loop): Compute global variable for data references.
13808 Bail out if there are too many data references.
13810 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13812 * tree-loop-distribution.c (loop_nest): New global var.
13813 (build_rdg): Use loop directly, rather than loop nest.
13814 (pg_add_dependence_edges): Remove loop nest parameter. Use global
13816 (distribute_loop): Compute global variable loop nest. Update use.
13818 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13820 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
13821 (partition_merge_into): New parameter. Dump reason for fusion.
13822 (distribute_loop): Update use of partition_merge_into.
13824 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13826 * tree-loop-distribution.c (bb_top_order_index): New.
13827 (bb_top_order_index_size, bb_top_order_cmp): New.
13828 (stmts_from_loop): Use topological order.
13829 (pass_loop_distribution::execute): Compute and release topological
13830 order for basic blocks.
13832 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13834 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
13837 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13839 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
13840 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
13841 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
13842 * internal-fn.def (LOOP_DIST_ALIAS): New.
13843 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
13844 (fold_loop_internal_call): ... this.
13845 (vect_loop_dist_alias_call): New function.
13846 (set_uid_loop_bbs): Call fold_loop_internal_call.
13847 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
13850 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13853 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
13854 Require dead FLAGS_REG at the beginning of a peephole.
13856 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13859 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
13860 arguments in the call to __builtin_ia32_sbb_u32.
13861 (_subborrow_u64): Swap _X and _Y arguments in the call to
13862 __builtin_ia32_sbb_u64.
13864 2017-07-04 Jakub Jelinek <jakub@redhat.com>
13867 * tree-vrp.c (compare_assert_loc): Turn into a function template
13868 with stable template parameter. Only test if a->e is NULL,
13869 !a->e == !b->e has been verified already. Use e == NULL or
13870 e != NULL instead of e or ! e tests. If stable is true, don't use
13871 iterative_hash_expr, on the other side allow a or b or both NULL
13872 and sort the NULLs last.
13873 (process_assert_insertions): Sort using compare_assert_loc<false>
13874 instead of compare_assert_loc, later sort using
13875 compare_assert_loc<true> before calling process_assert_insertions_for
13876 in a loop. Use break instead of continue once seen NULL pointer.
13878 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13880 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13881 Cortex-R7 and Cortex-R8 processors.
13883 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13885 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
13886 uninitialized while src is not.
13888 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
13890 * common/config/arm/arm-common.c: Adjust include path for
13892 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
13893 (arm-cpu.h): Create in build directory. Adjust dependency rules.
13894 (arm-cpu-data.h): Likewise.
13895 (arm-cpu-cdata.h): Likewise.
13896 * config/arm/arm-cpu.h: Delete.
13897 * config/arm/arm-cpu-cdata.h: Delete.
13898 * config/arm/arm-cpu-data.h: Delete.
13900 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
13902 * config/arm/arm-cpus.in (cortex-a55): New.
13903 (cortex-a75): Likewise.
13904 (cortex-a75.cortex-a55): Likewise.
13905 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
13907 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
13908 * config/arm/arm-cpu-cdata.h: Regenerate.
13909 * config/arm/arm-cpu-data.h: Regenerate.
13910 * config/arm/arm-cpu.h: Regenerate.
13911 * config/arm/arm-tables.opt: Regenerate.
13912 * config/arm/arm-tune.md: Regenerate.
13914 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13916 * haifa-sched.c (sched_create_recovery_edges): Update profile.
13918 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13920 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
13923 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
13925 PR tree-optimization/81292
13926 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
13927 full_string_p, also call adjust_related_strinfos if the adjustment
13928 is simple, otherwise invalidate related strinfos.
13930 2017-07-04 Martin Liska <mliska@suse.cz>
13933 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
13934 newly created variable as DECL_IGNORED_P.
13936 2017-07-04 Martin Liska <mliska@suse.cz>
13939 * ipa-inline.c (inline_small_functions):
13940 Use xstrdup_for_dump.
13942 2017-07-04 Tom de Vries <tom@codesourcery.com>
13944 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
13946 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13949 * config/darwin.c (darwin_function_switched_text_sections):
13952 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
13954 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
13956 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13958 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
13960 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13962 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
13963 min_profitable_iters, and th as inclusive lower bounds.
13964 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
13965 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
13966 for min_profitable_iters and min_profitable_estimate.
13967 (vect_transform_loop): Treat th as an inclusive lower bound.
13968 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
13970 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13973 * config/darwin.c (darwin_function_switched_text_sections):
13974 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
13975 in two pieces, and suppress the use of buf.
13977 2017-07-03 Nathan Sidwell <nathan@acm.org>
13979 * hash-table.h (hash_table_mod1): Fix indentation.
13981 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13983 PR middle-end/81290
13984 * predict.c (force_edge_cold): Be more careful about propagation
13986 * profile-count.h (profile_probability::guessed,
13987 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
13989 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
13991 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
13993 * doc/invoke.texi (rcpc architecture extension): Document it.
13995 2017-07-03 Richard Biener <rguenther@suse.de>
13997 PR tree-optimization/60510
13998 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
13999 the scalar reduction PHI and use it.
14000 (vectorizable_reduction): Properly guard the single_defuse_cycle
14001 path for non-SLP reduction chains where we cannot use it.
14002 Rework reduc_def/index and vector type deduction. Rework
14003 vector operand gathering during reduction op code-gen.
14004 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
14005 chains dissolve the chain and leave it to non-SLP reduction
14008 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14010 * tree-data-ref.h (dr_alignment): Declare.
14011 * tree-data-ref.c (dr_alignment): New function.
14012 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
14013 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
14015 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
14017 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14019 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
14020 and base_misalignment fields.
14021 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
14022 * tree-data-ref.c: Include builtins.h.
14023 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
14024 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
14025 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
14026 * tree-vect-data-refs.c: Include tree-cfg.h.
14027 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
14028 fields instead of calculating an alignment here.
14029 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
14030 innermost_loop_behavior fields.
14032 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14034 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
14036 (DR_STEP_ALIGNMENT): New macro.
14037 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
14038 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
14039 (create_data_ref): Print it.
14040 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
14041 to tell whether the step preserves vector (mis)alignment.
14042 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14043 Move the check for an integer step and generalise to all INTEGER_CST.
14044 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
14045 Print the outer step alignment.
14047 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14049 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
14050 with offset_alignment.
14051 (DR_ALIGNED_TO): Delete.
14052 (DR_OFFSET_ALIGNMENT): New macro.
14053 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
14054 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
14055 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
14056 (create_data_ref): Likewise.
14057 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14058 (vect_analyze_data_refs): Likewise.
14059 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
14060 creating dummy innermost behavior.
14062 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14064 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
14065 with a "innermost_loop_behavior *" and refeence tree.
14066 * tree-data-ref.c (dr_analyze_innermost): Likewise.
14067 (create_data_ref): Update call accordingly.
14068 * tree-predcom.c (find_looparound_phi): Likewise.
14070 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14072 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
14073 fields with dr_wrt_vec_loop.
14074 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
14075 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
14076 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
14077 (vect_dr_behavior): New function.
14078 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14079 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
14080 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
14081 track whether the step preserves the misalignment.
14082 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14083 Use vect_dr_behavior.
14084 (vect_setup_realignment): Update call accordingly.
14085 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
14086 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
14087 call to vect_create_addr_base_for_vector_ref.
14088 (vect_create_cond_for_align_checks): Likewise.
14089 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
14090 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
14091 (vect_recog_mask_conversion_pattern): Likewise.
14092 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
14093 (new_stmt_vec_info): Remove redundant zeroing.
14095 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14097 * common/config/arm/arm-common.c (arm_be8_option): New function.
14098 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
14099 (ISA_ARMv6): Add isa_bit_be8.
14100 * config/arm/arm.h (arm_be8_option): Add prototype.
14101 (BE8_SPEC_FUNCTION): New define.
14102 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
14103 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
14104 (mlittle-endian): Similarly.
14105 (mbe8, mbe32): New options.
14106 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
14107 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
14109 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14111 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
14113 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14115 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
14116 (cleanup_tree_cfg_bb): Use it.
14117 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
14119 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
14121 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14124 * loop-doloop.c (add_test): Update profile.
14126 2017-07-03 Martin Liska <mliska@suse.cz>
14129 * sanopt.c (rewrite_usage_of_param): New function.
14130 (sanitize_rewrite_addressable_params): Likewise.
14131 (pass_sanopt::execute): Call rewrite_usage_of_param.
14133 2017-07-03 Richard Biener <rguenther@suse.de>
14135 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
14136 back to using VIEW_CONVERT_EXPR.
14138 2017-07-03 Martin Liska <mliska@suse.cz>
14141 * doc/extend.texi: Document when a resolver function is
14142 generated for target_clones.
14144 2017-07-03 Martin Liska <mliska@suse.cz>
14146 * asan.c (asan_emit_stack_protection): Unpoison just red zones
14147 and shadow memory of auto variables which are subject of
14148 use-after-scope sanitization.
14149 (asan_expand_mark_ifn): Add do set only when is_poison.
14151 2016-07-03 Richard Biener <rguenther@suse.de>
14153 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
14155 (vect_force_simple_reduction): Record reduction def -> phi mapping.
14156 (vectorizable_reduction): Perform reduction PHI creation when
14157 visiting a reduction PHI and adjust and simplify code generation
14158 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
14159 (vect_transform_loop): Visit reduction PHIs.
14160 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
14161 defs into the SLP tree.
14162 (vect_build_slp_tree): Reduction defs terminate the recursion.
14163 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
14165 (vect_get_vec_defs_for_stmt_copy): Export.
14166 (vect_get_vec_defs): Likewise.
14167 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
14169 (vect_get_vec_defs_for_stmt_copy): Declare.
14170 (vect_get_vec_defs): Likewise.
14172 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14174 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
14175 parameter with a "loop" parameter and use it instead of the
14176 loop containing DR_STMT. Don't check simple_iv when doing
14177 BB analysis. Describe the two analysis modes in the comment.
14179 2017-07-03 Tom de Vries <tom@codesourcery.com>
14181 PR tree-optimization/69468
14182 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
14183 (find_same_succ_bb): Handle ignore_edge_flags.
14185 2017-07-03 Tom de Vries <tom@codesourcery.com>
14187 PR tree-optimization/81192
14188 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
14190 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
14192 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
14194 2017-07-03 Tom de Vries <tom@codesourcery.com>
14196 PR tree-optimization/81192
14197 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
14198 BB_SAME_SUCC (bb) == NULL.
14200 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14202 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
14205 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14207 * dumpfile.c: Include profile-count.h
14208 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
14210 (insert_cond_bb): Update profile.
14211 * tree-cfg.h (insert_cond_bb): Update prototype.
14212 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
14213 * tree-dump.c: Do not include tree-cfg.
14215 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14217 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
14219 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14221 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
14224 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14226 * tree-complex.c (expand_complex_div_wide): update profile.
14228 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14229 Alan Hayward <alan.hayward@arm.com>
14230 David Sherwood <david.sherwood@arm.com>
14232 * Makefile.in (MACHMODE_H): Remove insn-modes.h
14233 (CORETYPES_H): New define.
14234 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
14235 (insn-modes-inline.h, s-modes-inline-h): New rules.
14236 (generated_files): Add insn-modes-inline.h.
14237 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
14238 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
14239 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
14240 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
14241 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
14242 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
14243 (build/gencodes.o, build/genconditions.o): Likewise.
14244 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
14245 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
14246 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
14247 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
14248 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
14249 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
14250 * coretypes.h: Include everything up to real.h for generators.
14251 Include insn-modes.h first. Include wide-int-print.h after
14252 wide-int.h. Include insn-modes-inline.h and then machmode.h.
14253 * machmode.h: Don't include insn-modes.h here.
14254 * function-tests.c: Remove includes of signop.h, machmode.h,
14255 double-int.h and wide-int.h.
14257 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
14259 * optc-save-gen.awk: Likewise.
14260 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
14261 * godump.c: Remove include of wide-int-print.h.
14262 * pretty-print.h: Likewise.
14263 * wide-int-print.cc: Likewise.
14264 * wide-int.cc: Likewise.
14265 * hash-map-tests.c: Remove include of signop.h.
14266 * hash-set-tests.c: Likewise.
14267 * rtl-tests.c: Likewise.
14268 * mkconfig.sh: Remove include of machmode.h.
14269 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
14271 (emit_insn_modes_inline_h): ...this new function. Emit the code
14272 into an insn-modes-inline.h header file, adding appropriate
14273 include guards and end comments.
14274 (emit_insn_modes_c_header): Remove include of machmode.h.
14275 (emit_min_insn_modes_c_header): Include coretypes.h rather than
14277 (main): Handle -i flag and call emit_insn_modes_inline_h when
14280 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14282 * tree-ssa-strlen.c (strinfo): Rename the length field to
14283 nonzero_chars. Add a full_string_p field.
14284 (compare_nonzero_chars, zero_length_string_p): New functions.
14285 (get_addr_stridx): Add an offset_out parameter.
14286 Use compare_nonzero_chars.
14287 (get_stridx): Update accordingly. Use compare_nonzero_chars.
14288 (new_strinfo): Update after above changes to strinfo.
14289 (set_endptr_and_length): Set full_string_p.
14290 (get_string_length): Update after above changes to strinfo.
14291 (unshare_strinfo): Update call to new_strinfo.
14292 (maybe_invalidate): Likewise.
14293 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
14294 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
14295 as a uhwi instead of an shwi. Update after above changes to
14296 strinfo and new_strinfo.
14297 (zero_length_string): Assert that chainsi contains full strings.
14298 Use zero_length_string_p. Update call to new_strinfo.
14299 (adjust_related_strinfos): Update after above changes to strinfo.
14300 Copy full_string_p from origsi.
14301 (adjust_last_stmt): Use zero_length_string_p.
14302 (handle_builtin_strlen): Update after above changes to strinfo and
14303 new_strinfo. Install the lhs as the string length if the previous
14304 entry didn't describe a full string.
14305 (handle_builtin_strchr): Update after above changes to strinfo
14307 (handle_builtin_strcpy): Likewise.
14308 (handle_builtin_strcat): Likewise.
14309 (handle_builtin_malloc): Likewise.
14310 (handle_pointer_plus): Likewise.
14311 (handle_builtin_memcpy): Likewise. Track nonzero characters
14312 that aren't necessarily followed by a nul terminator.
14313 (handle_char_store): Likewise.
14315 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14317 PR tree-optimization/80769
14318 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
14319 for malloc and calloc. Document the new invariant that all related
14320 strinfos have delayed lengths or none do.
14321 (verify_related_strinfos): Move earlier in file.
14322 (set_endptr_and_length): New function, split out from...
14323 (get_string_length): ...here. Also set the lengths of related
14325 (zero_length_string): Assert that chainsi has known (rather than
14327 (adjust_related_strinfos): Likewise.
14329 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14331 PR tree-optimization/81136
14332 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
14333 assert that two references with the same misalignment have the same
14334 compile-time misalignment if those compile-time misalignments
14337 2017-07-01 Andi Kleen <ak@linux.intel.com>
14339 * print-tree.c (print_node): Print all attributes.
14341 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14343 * cfg.c (scale_bbs_frequencies): New function.
14344 * cfg.h (scale_bbs_frequencies): Declare it.
14345 * cfgloopanal.c (single_likely_exit): Cleanup.
14346 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
14348 (scale_loop_profile): Likewise.
14349 (loop_version): Likewise.
14350 (create_empty_loop_on_edge): Update.
14351 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
14352 scale_loop_frequencies, scale_loop_profile, loopify,
14353 loop_version): Update prototypes.
14354 * modulo-sched.c (sms_schedule): Update.
14355 * predict.c (unlikely_executed_edge_p): Also check probability.
14356 (probably_never_executed_edge_p): Fix typo.
14357 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14358 * tree-parloops.c (gen_parallel_loop): Update.
14359 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
14360 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14361 * tree-ssa-loop-split.c (split_loop): Update.
14362 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14363 * tree-vect-loop-manip.c (vect_do_peeling): Update.
14364 (vect_loop_versioning): Update.
14365 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14367 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14369 * trans-mem.c (split_bb_make_tm_edge): Update profile.
14371 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14373 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
14374 to keep profile consistent.
14376 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14378 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
14379 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
14380 * profile-count.h (max_safe_multiplier): Make unsigned.
14381 (profile_count::guessed_zero): New.
14383 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14385 * bb-reorder.c (fix_up_crossing_landing_pad,
14386 fix_crossing_conditional_branches): Use make_single_succ_edge
14387 to keep profile consistent.
14389 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14391 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
14394 2017-07-01 Jakub Jelinek <jakub@redhat.com>
14397 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
14398 the right scopes, make sure cond_jump isn't preserved between multiple
14399 iterations. Search for fallthru edge whenever there are 3+ edges and
14400 use find_fallthru_edge for it.
14402 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14404 Patch by Alexander Monakov <amonakov@ispras.ru>
14405 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
14406 probabilities consistently.
14408 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14410 * pa.c (pa_expand_compare_and_swap_loop): Update call of
14411 emit_cmp_and_jump_insns.
14413 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14416 * tree-inline.c (expand_call_inline): Combine profile statuses.
14418 2017-06-30 Andrew Pinski <apinski@cavium.com>
14420 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
14421 fold_stmt returned true.
14423 2017-06-30 Nathan Sidwell <nathan@acm.org>
14425 * ggc.h (empty_string): Delete.
14426 * cfgexpand.c (expand_asm_stmt): Use plain "".
14427 * optabs.c (expand_asm_memory_barrier): Likewise.
14428 * stringpool.c (empty_string): Delete.
14429 (digit_vector, digit_string): Delete.
14430 (ggc_alloc_string): Use plain "", don't optimize single digit
14431 strings. Use ggc_alloc_atomic.
14433 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
14435 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
14436 comparison set and one other set, use the cost of the non-comparison
14439 2017-06-30 Nathan Sidwell <nathan@acm.org>
14441 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
14444 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
14446 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
14447 loops. Remove now unneeded calls to gimple_switch_set_label() that
14448 just set removed labels to NULL_TREE.
14450 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
14452 * tree-ssanames.c (set_range_info_raw): Abstract from ...
14453 (set_range_info): ...here. Only call set_range_info_raw if domain
14455 (set_nonzero_bits): Call set_range_info_raw.
14456 * tree-ssanames.h (set_range_info_raw): New.
14458 2017-06-30 Jakub Jelinek <jakub@redhat.com>
14461 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
14462 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
14463 of nonimmediate_operand and <store_mask_constraint> instead of m
14464 for the input operand. For V8FI iterator, always split if input
14465 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
14466 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
14467 <store_mask_predicate> instead of register_operand and
14468 <store_mask_constraint> instead of v for the input operand. Make
14469 sure both operands aren't MEMs for if not <mask_applied>.
14471 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
14473 * lto-wrapper.c (copy_file) Close both file descriptors before
14476 2017-06-30 Martin Liska <mliska@suse.cz>
14479 * multiple_target.c (create_dispatcher_calls): Make ifunc
14480 also for function that don't have calls or are not referenced.
14482 2017-06-30 Richard Biener <rguenther@suse.de>
14484 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
14485 analyze the first scalar stmt. Move vector type computation
14486 for the BB case here from ...
14487 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
14488 live operation processing in the SLP case properly.
14490 2017-06-30 Richard Biener <rguenther@suse.de>
14492 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
14494 2017-06-30 Martin Liska <mliska@suse.cz>
14497 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
14498 before BUILT_IN_UNWIND_RESUME when ASAN is used.
14500 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
14502 * doc/invoke.texi (AArch64): Add missing options and remove redundant
14505 2017-06-30 Richard Biener <rguenther@suse.de>
14507 PR tree-optimization/81249
14508 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
14509 condition reduction result to original scalar type.
14511 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14513 * profile-count.h (enum profile_quality): Fix typos and whitespace
14516 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14518 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
14519 type for branch probabilities.
14521 2017-06-29 Julian Brown <julian@codesourcery.com>
14522 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14524 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
14525 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
14526 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
14527 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
14529 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14531 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
14532 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
14533 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
14534 CC usage from generic code to here.
14535 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
14536 CC usage into the target macros.
14538 2017-06-29 Maya Rashish <coypu@sdf.org>
14540 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
14543 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14545 * arm/arm-builtins.c: Include profile-count.h
14546 * except.c (sjlj_emit_function_enter): Use
14547 profile_probability::unlikely.
14549 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14551 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
14552 and tocrel_offset be pointer args rather than implicitly using
14554 (legitimate_constant_pool_address_p, rs6000_emit_move,
14555 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
14556 tocrel_offset and use in toc_relative_expr_p call.
14557 (print_operand, print_operand_address): Use static tocrel_base_oac
14558 and tocrel_offset_oac.
14559 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
14562 2017-06-29 Maya Rashish <coypu@sdf.org>
14564 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
14566 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
14568 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
14569 objects, take into account only the alignment of 'op0' and 'mode1' if
14572 2017-06-29 Steve Ellcey <sellcey@cavium.com>
14574 * ccmp.c (ccmp_tree_comparison_p): New function.
14575 (ccmp_candidate_p): Update to use above function.
14576 (get_compare_parts): New function.
14577 (expand_ccmp_next): Update to use new functions.
14578 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
14580 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
14581 take mode as argument.
14582 * ccmp.h (expand_ccmp_expr): Add mode as argument.
14583 * expr.c (expand_expr_real_1): Pass mode as argument.
14585 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
14587 * combine.c (combine_instructions): Print insns to dump_file, together
14590 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14592 * asan.c (asan_emit_stack_protection): Update.
14593 (create_cond_insert_point): Update.
14594 * auto-profile.c (afdo_propagate_circuit): Update.
14595 * basic-block.h (struct edge_def): Turn probability to
14596 profile_probability.
14597 (EDGE_FREQUENCY): Update.
14598 * bb-reorder.c (find_traces_1_round): Update.
14599 (better_edge_p): Update.
14600 (sanitize_hot_paths): Update.
14601 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
14602 (make_single_succ_edge): Update.
14603 (check_bb_profile): Update.
14604 (dump_edge_info): Update.
14605 (update_bb_profile_for_threading): Update.
14606 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
14607 probabilitycount to 0.
14608 * cfgbuild.c (compute_outgoing_frequencies): Update.
14609 * cfgcleanup.c (try_forward_edges): Update.
14610 (outgoing_edges_match): Update.
14611 (try_crossjump_to_edge): Update.
14612 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
14613 (expand_gimple_tailcall): Update.
14614 (construct_init_block): Use make_single_succ_edge.
14615 (construct_exit_block): Use make_single_succ_edge.
14616 * cfghooks.c (verify_flow_info): Update.
14617 (redirect_edge_succ_nodup): Update.
14618 (split_edge): Update.
14619 (account_profile_record): Update.
14620 * cfgloopanal.c (single_likely_exit): Update.
14621 * cfgloopmanip.c (scale_loop_profile): Update.
14622 (set_zero_probability): Remove.
14623 (duplicate_loop_to_header_edge): Update.
14624 * cfgloopmanip.h (loop_version): Update prototype.
14625 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
14626 (force_nonfallthru_and_redirect): Update.
14627 (update_br_prob_note): Update.
14628 (rtl_verify_edges): Update.
14629 (purge_dead_edges): Update.
14630 (rtl_lv_add_condition_to_bb): Update.
14631 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
14632 * cgraphunit.c (init_lowered_empty_function): Update.
14633 (cgraph_node::expand_thunk): Update.
14634 * cilk-common.c: Include profile-count.h
14635 * dojump.c (inv): Remove.
14636 (jumpifnot): Update.
14637 (jumpifnot_1): Update.
14638 (do_jump_1): Update.
14640 (do_jump_by_parts_greater_rtx): Update.
14641 (do_compare_rtx_and_jump): Update.
14642 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
14643 do_jump_1. do_compare_rtx_and_jump): Update prototype.
14644 * dwarf2cfi.c: Include profile-count.h
14645 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
14646 (sjlj_emit_dispatch_table): Likewise.
14647 * explow.c: Include profile-count.h
14648 * expmed.c (emit_store_flag_force): Update.
14649 (do_cmp_and_jump): Update.
14650 * expr.c (compare_by_pieces_d::generate): Update.
14651 (compare_by_pieces_d::finish_mode): Update.
14652 (emit_block_move_via_loop): Update.
14653 (store_expr_with_bounds): Update.
14654 (store_constructor): Update.
14655 (expand_expr_real_2): Update.
14656 (expand_expr_real_1): Update.
14657 * expr.h (try_casesi, try_tablejump): Update prototypes.
14658 * gimple-pretty-print.c (dump_probability): Update.
14659 (dump_profile): New.
14660 (dump_gimple_label): Update.
14661 (dump_gimple_bb_header): Update.
14662 * graph.c (draw_cfg_node_succ_edges): Update.
14663 * hsa-gen.c (convert_switch_statements): Update.
14664 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
14665 (find_if_case_1): Update.
14666 (find_if_case_2): Update.
14667 * internal-fn.c (expand_arith_overflow_result_store): Update.
14668 (expand_addsub_overflow): Update.
14669 (expand_neg_overflow): Update.
14670 (expand_mul_overflow): Update.
14671 (expand_vector_ubsan_overflow): Update.
14672 * ipa-cp.c (good_cloning_opportunity_p): Update.
14673 * ipa-split.c (split_function): Use make_single_succ_edge.
14674 * ipa-utils.c (ipa_merge_profiles): Update.
14675 * loop-doloop.c (add_test): Update.
14676 (doloop_modify): Update.
14677 * loop-unroll.c (compare_and_jump_seq): Update.
14678 (unroll_loop_runtime_iterations): Update.
14679 * lra-constraints.c (lra_inheritance): Update.
14680 * lto-streamer-in.c (input_cfg): Update.
14681 * lto-streamer-out.c (output_cfg): Update.
14682 * mcf.c (adjust_cfg_counts): Update.
14683 * modulo-sched.c (sms_schedule): Update.
14684 * omp-expand.c (expand_omp_for_init_counts): Update.
14685 (extract_omp_for_update_vars): Update.
14686 (expand_omp_ordered_sink): Update.
14687 (expand_omp_for_ordered_loops): Update.
14688 (expand_omp_for_generic): Update.
14689 (expand_omp_for_static_nochunk): Update.
14690 (expand_omp_for_static_chunk): Update.
14691 (expand_cilk_for): Update.
14692 (expand_omp_simd): Update.
14693 (expand_omp_taskloop_for_outer): Update.
14694 (expand_omp_taskloop_for_inner): Update.
14695 * omp-simd-clone.c (simd_clone_adjust): Update.
14696 * optabs.c (expand_doubleword_shift): Update.
14697 (expand_abs): Update.
14698 (emit_cmp_and_jump_insn_1): Update.
14699 (expand_compare_and_swap_loop): Update.
14700 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
14701 * predict.c (predictable_edge_p): Update.
14702 (edge_probability_reliable_p): Update.
14703 (set_even_probabilities): Update.
14704 (combine_predictions_for_insn): Update.
14705 (combine_predictions_for_bb): Update.
14706 (propagate_freq): Update.
14707 (estimate_bb_frequencies): Update.
14708 (force_edge_cold): Update.
14709 * profile-count.c (profile_count::dump): Add missing space into dump.
14710 (profile_count::debug): Add newline.
14711 (profile_count::differs_from_p): Explicitly convert to unsigned.
14712 (profile_count::stream_in): Update.
14713 (profile_probability::dump): New member function.
14714 (profile_probability::debug): New member function.
14715 (profile_probability::differs_from_p): New member function.
14716 (profile_probability::differs_lot_from_p): New member function.
14717 (profile_probability::stream_in): New member function.
14718 (profile_probability::stream_out): New member function.
14719 * profile-count.h (profile_count_quality): Rename to ...
14720 (profile_quality): ... this one.
14721 (profile_probability): New.
14722 (profile_count): Update.
14723 * profile.c (compute_branch_probabilities): Update.
14724 * recog.c (peep2_attempt): Update.
14725 * sched-ebb.c (schedule_ebbs): Update.
14726 * sched-rgn.c (find_single_block_region): Update.
14727 (compute_dom_prob_ps): Update.
14728 (schedule_region): Update.
14729 * sel-sched-ir.c (compute_succs_info): Update.
14730 * stmt.c (struct case_node): Update.
14731 (do_jump_if_equal): Update.
14732 (get_outgoing_edge_probs): Update.
14733 (conditional_probability): Update.
14734 (emit_case_dispatch_table): Update.
14735 (expand_case): Update.
14736 (expand_sjlj_dispatch_table): Update.
14737 (emit_case_nodes): Update.
14738 * targhooks.c: Update.
14739 * tracer.c (better_p): Update.
14740 (find_best_successor): Update.
14741 * trans-mem.c (expand_transaction): Update.
14742 * tree-call-cdce.c: Update.
14743 * tree-cfg.c (gimple_split_edge): Upate.
14744 (move_sese_region_to_fn): Upate.
14745 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
14746 * tree-eh.c (lower_resx): Upate.
14747 (cleanup_empty_eh_move_lp): Upate.
14748 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14749 * tree-inline.c (copy_edges_for_bb): Update.
14750 (copy_cfg_body): Update.
14751 * tree-parloops.c (gen_parallel_loop): Update.
14752 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
14753 (gimple_gen_time_profiler): Update.
14754 * tree-ssa-dce.c (remove_dead_stmt): Update.
14755 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14756 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
14757 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
14758 (unloop_loops): Update.
14759 (try_peel_loop): Update.
14760 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14761 * tree-ssa-loop-split.c (connect_loops): Update.
14762 (split_loop): Update.
14763 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14764 (hoist_guard): Update.
14765 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
14766 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
14767 (value_replacement): Update.
14768 * tree-ssa-reassoc.c (branch_fixup): Update.
14769 * tree-ssa-tail-merge.c (replace_block_by): Update.
14770 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
14771 (create_edge_and_update_destination_phis): Update.
14772 (compute_path_counts): Update.
14773 (recompute_probabilities): Update.
14774 (update_joiner_offpath_counts): Update.
14775 (freqs_to_counts_path): Update.
14776 (duplicate_thread_path): Update.
14777 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
14778 (struct switch_conv_info): Update.
14779 (gen_inbound_check): Update.
14780 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14781 (vect_do_peeling): Update.
14782 (vect_loop_versioning): Update.
14783 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14784 (optimize_mask_stores): Update.
14785 * ubsan.c (ubsan_expand_null_ifn): Update.
14786 * value-prof.c (gimple_divmod_fixed_value): Update.
14787 (gimple_divmod_fixed_value_transform): Update.
14788 (gimple_mod_pow2): Update.
14789 (gimple_mod_pow2_value_transform): Update.
14790 (gimple_mod_subtract): Update.
14791 (gimple_mod_subtract_transform): Update.
14792 (gimple_ic): Update.
14793 (gimple_stringop_fixed_value): Update.
14794 (gimple_stringops_transform): Update.
14795 * value-prof.h: Update.
14797 2017-06-29 Carl Love <cel@us.ibm.com>
14799 * config/rs6000/rs6000-c.c: Add support for built-in functions
14800 vector signed int vec_signed (vector float);
14801 vector signed long long vec_signed (vector double);
14802 vector signed int vec_signed2 (vector double, vector double);
14803 vector signed int vec_signede (vector double);
14804 vector signed int vec_signedo (vector double);
14805 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
14806 instruction generator.
14807 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14808 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
14809 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
14811 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
14812 vunsignede_v2df): Add define_expands.
14813 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
14814 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
14815 VEC_UNSIGNEDO): Add definitions.
14816 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14817 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
14818 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
14819 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
14820 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
14821 * config/rs6000/altivec.h (vec_signed, vec_signed2,
14822 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
14823 vec_unsignede, vec_unsignedo): Add builtin defines.
14824 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
14826 * doc/extend.texi: Update the built-in documentation file for the
14827 new built-in functions.
14829 2017-06-29 Richard Biener <rguenther@suse.de>
14831 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
14832 reduction chains to LOOP_VINFO_REDUCTIONS.
14833 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
14834 SLP reductions after processing reduction chains.
14836 2017-06-29 Nathan Sidwell <nathan@acm.org>
14838 * builtins.c (fold_builtin_FUNCTION): Use
14839 lang_hooks.decl_printable_name.
14841 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
14843 PR middle-end/81194
14844 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
14845 with only one label.
14846 * stmt.c (expand_case): Assert NCASES is greater than one.
14848 2017-06-29 Richard Biener <rguenther@suse.de>
14850 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
14852 (group_case_labels): Likewise.
14853 (find_taken_edge): Push sanity checking on val to workers...
14854 (find_taken_edge_cond_expr): ... here
14855 (find_taken_edge_switch_expr): ... and here, handle cases
14856 with just a default label.
14857 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
14858 (group_case_labels): Likewise.
14859 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
14860 group_case_labels does anything cleanup the CFG again.
14862 2017-06-29 Bin Cheng <bin.cheng@arm.com>
14864 PR tree-optimization/81196
14865 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
14866 exit condition comparing two IVs.
14868 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
14870 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
14871 profile to the dummy entry at the end of the list of architectures.
14872 * config/arm/arm-cpu-cdata.h: Regenerated.
14874 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14875 Michael Collison <michael.collison@arm.com>
14878 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
14880 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
14881 (*aarch64_reg_<mode>3_minus_mask): New pattern.
14882 (*aarch64_<optab>_reg_di3_mask2): New pattern.
14883 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
14884 of shift when the shift amount is masked with constant equal to
14885 the size of the mode.
14886 * config/aarch64/predicates.md (subreg_lowpart_operator): New
14889 2017-06-29 Martin Liska <mliska@suse.cz>
14891 * config/i386/i386.opt: Change range from [1,5] to [0,5].
14893 2017-06-29 Yury Gribov <tetra2005@gmail.com>
14896 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
14899 (edge_growth_cache_entry::edge_growth_cache_entry): New
14901 (reset_edge_growth_cache): Update to use constructor.
14903 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14905 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
14906 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14907 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14909 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
14911 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
14912 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
14914 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
14916 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
14917 (*-linux-uclibc*): Add t-uclibc tmake_file.
14918 * config/t-musl: New.
14919 * config/t-uclibc: New.
14921 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
14923 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
14925 (gen_comm_data): Emit architectural setting of arch_prof.
14926 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
14928 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
14929 (armv8-m.base, armv8-m.main): Likewise.
14930 * arm-protos.h (arm_build_target): Add profile field.
14931 (arch_option): Likewise.
14932 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
14934 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
14935 arm_active_target.profile.
14937 2017-06-28 Richard Biener <rguenther@suse.de>
14939 PR middle-end/81227
14940 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
14941 TYPE_OVERFLOW_WRAPS.
14942 * match.pd (negate_expr_p): Likewise.
14943 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
14944 fold_build2, not fold_binary.
14946 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14948 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
14949 Convert memory address to Pmode.
14950 (aarch64_print_operand): Assert MEM operands are always Pmode.
14952 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14955 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
14956 Remove redundant if.
14957 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
14958 * config/arm/aarch-common-protos.h
14959 (aarch_forward_to_shift_is_not_shifted_re): Remove.
14960 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
14962 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
14965 * multiple_target.c (create_dispatcher_calls): Set the default
14966 clone to be static, not public.
14968 2017-06-28 Richard Biener <rguenther@suse.de>
14970 * tree-vect-loop.c (vectorizable_reduction): Move special
14971 cond reduction IV var creation ...
14972 (vect_create_epilog_for_reduction): ... here. Remove induction_index
14973 parameter. Use STMT_VINFO_VECTYPE.
14974 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
14977 2017-06-28 Martin Liska <mliska@suse.cz>
14980 * ipa-visibility.c (non_local_p): Handle visibility.
14982 2017-06-28 Martin Liska <mliska@suse.cz>
14985 * common.opt: Add IntegerRange to various options.
14986 * opt-functions.awk (integer_range_info): New function.
14987 * optc-gen.awk: Add integer_range_info to cl_options struct.
14988 * opts-common.c (decode_cmdline_option): Handle
14989 CL_ERR_INT_RANGE_ARG.
14990 (cmdline_handle_error): Likewise.
14991 * opts.c (print_filtered_help): Show valid interval in
14992 when --help is provided.
14993 * opts.h (struct cl_option): Add range_min and range_max fields.
14994 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
14996 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
14998 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
14999 (x * C EQ/NE y * C): New transformation.
15001 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
15003 * genmultilib (combination_space): Accept '+' in option names.
15005 2017-06-28 Martin Liska <mliska@suse.cz>
15008 * asan.c (instrument_derefs): Bail out inner references
15009 that are hard register variables.
15011 2017-06-28 Jakub Jelinek <jakub@redhat.com>
15014 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
15015 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
15017 2017-06-28 Richard Biener <rguenther@suse.de>
15019 * tree-vectorizer.h (vect_get_vec_defs): Remove.
15020 (vect_get_slp_defs): Adjust.
15021 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
15023 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
15025 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
15026 get_initial_defs_for_reduction instead of vect_get_vec_defs.
15027 (vectorizable_reduction): Adjust.
15028 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
15030 (vect_get_slp_defs): Likewise.
15031 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
15032 (vectorizable_bswap): Adjust.
15033 (vectorizable_call): Likewise.
15034 (vectorizable_conversion): Likewise.
15035 (vectorizable_assignment): Likewise.
15036 (vectorizable_shift): Likewise.
15037 (vectorizable_operation): Likewise.
15038 (vectorizable_store): Likewise.
15039 (vectorizable_condition): Likewise.
15040 (vectorizable_comparison): Likewise.
15042 2017-06-28 Michael Collison <michael.collison@arm.com>
15045 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
15047 (arm_gen_movmemqi): Removed unused variable 'i'.
15048 Convert 'for' loop into 'while' loop.
15049 (arm_expand_prologue): Remove last unnecessary set of insn.
15050 (thumb_pop): Remove unused variable 'pushed_words'.
15051 (thumb_exit): Remove last unnecessary set of regs_to_pop.
15053 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15055 * config/s390/predicates.md: Use s390_rel_address_ok_p.
15056 * config/s390/s390-protos.h: Add prototype of
15057 s390_rel_address_ok_p.
15058 * config/s390/s390.c (s390_got_symbol): New function.
15059 (s390_rel_address_ok_p): New function.
15060 (legitimize_pic_address): Use s390_rel_address_ok_p.
15061 (s390_load_got): Use s390_got_symbol.
15062 (s390_option_override): Issue error if
15063 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
15064 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
15066 * config/s390/s390.opt: New option mpic-data-is-text-relative.
15068 2017-06-27 Andrew Pinski <apinski@cavium.com>
15070 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
15071 (X * copysign (1.0, X)): New pattern.
15072 (X * copysign (1.0, -X)): New pattern.
15073 (copysign (-1.0, CST)): New pattern.
15075 2017-06-27 Joseph Myers <joseph@codesourcery.com>
15077 * genmultilib (combination_space): Remove variable.
15078 Validate reuse rules against regular expression for any sequence
15079 of multilib options in any order.
15081 2017-06-27 Michael Collison <michael.collison@arm.com>
15083 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
15084 call aarch64_split_simd_combine.
15085 * (aarch64_combine_internal<mode>): Delete pattern.
15086 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
15087 Allow register and subreg operands.
15089 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15091 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
15092 specific need, just fallback on defaults.
15093 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
15095 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15096 Olivier Hainque <hainque@adacore.com>
15098 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
15100 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
15101 targets. Pick a default if no particular attempt applied.
15102 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
15105 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15107 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
15108 (x86_64-wrs-vxworks7): Likewise.
15110 2017-06-27 Marek Polacek <polacek@redhat.com>
15113 * ubsan.c (instrument_null): Check get_base_address's result for null.
15115 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15117 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
15119 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15121 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
15122 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
15123 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
15124 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
15125 New function types.
15126 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
15127 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
15128 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
15129 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
15130 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
15131 BUILT_IN_FEUPDATEENV): New builtins.
15132 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
15133 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
15134 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
15135 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
15137 (builtin_structptr_types): Adjust size.
15138 * tree.c (builtin_structptr_types): Add four entries.
15140 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15141 Olivier Hainque <hainque@adacore.com>
15143 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
15144 (TLS_SYM): New local macro, forcing reference to __tls__ on
15145 link command lines for VxWorks 7 RTPs, triggering initialization
15147 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
15148 OS features TLS support, true for RTPs on VxWorks 7.
15149 * config/vxworks.c (vxworks_override_options): Setup emutls
15152 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15154 * predict.c (test_prediction_value_range): Use -1U instead of -1
15155 to avoid narrowing conversion warning.
15156 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
15157 to avoid narrowing conversion warning.
15158 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
15160 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
15162 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15164 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
15165 64bit configurations.
15166 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
15167 (SIZE_TYPE): Likewise.
15168 * config/vxworks.c (vxworks_emutls_var_fields): Use
15169 long_unsigned_type_node instead of unsigned_type_node as the offset
15170 field type, which is "pointer" mode in emutls.c.
15172 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15175 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
15177 PR middle-end/81207
15178 * gimple-fold.c (replace_call_with_call_and_fold): Handle
15179 gimple_vuse copying separately from gimple_vdef copying.
15181 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15183 * value-prof.c (free_hist): Remove call to memset and the enclosing if
15186 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15187 Olivier Hainque <hainque@adacore.com>
15189 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
15190 for all vxworks7 targets.
15191 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
15192 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
15193 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
15194 variations for VX6/VX7 and 32/64bits later on in ...
15195 (VXWORKS_LIB_SPEC): Leverage new macros.
15196 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
15197 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
15199 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15201 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
15202 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
15204 2017-06-26 Carl Love <cel@us.ibm.com>
15206 * config/rs6000/rs6000-c.c: Add support for built-in functions
15207 vector bool char vec_reve (vector bool char);
15208 vector signed char vec_reve (vector signed char);
15209 vector unsigned char vec_reve (vector unsigned char);
15210 vector bool int vec_reve (vector bool int);
15211 vector signed int vec_reve (vector signed int);
15212 vector unsigned int vec_reve (vector unsigned int);
15213 vector bool long long vec_reve (vector bool long long);
15214 vector signed long long vec_reve (vector signed long long);
15215 vector unsigned long long vec_reve (vector unsigned long long);
15216 vector bool short vec_reve (vector bool short);
15217 vector signed short vec_reve (vector signed short);
15218 vector double vec_reve (vector double);
15219 vector float vec_reve (vector float);
15220 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
15221 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
15222 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
15223 (altivec_vreve): New pattern.
15224 * config/rs6000/altivec.h (vec_reve): New define.
15225 * doc/extend.texi (vec_rev): Update the built-in documentation file
15226 for the new built-in functions.
15228 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15230 PR tree-optimization/71815
15231 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
15233 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
15235 (slsr_process_phi): Likewise.
15236 (replace_uncond_cands_and_profitable_phis): Don't replace a
15237 multiply candidate with a stride of 1 (copy or cast).
15238 (phi_incr_cost): Call uses_consumed_by_stmt rather than
15240 (lowest_cost_path): Likewise.
15241 (total_savings): Likewise.
15243 2017-06-26 Richard Biener <rguenther@suse.de>
15246 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
15247 Use def_builtin_pure for all gather builtins.
15249 2017-06-26 Richard Biener <rguenther@suse.de>
15251 PR tree-optimization/81203
15252 * tree-tailcall.c (find_tail_calls): Do not move stmts into
15253 non-dominating BBs.
15255 2017-06-26 Marek Polacek <polacek@redhat.com>
15258 * doc/invoke.texi: Document -Wmultistatement-macros.
15260 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
15262 * doc/sourcebuild.texi (ARM-specific attributes): Document new
15263 arm_neon_ok_no_float_abi effective target.
15265 2017-06-26 Richard Biener <rguenther@suse.de>
15267 PR tree-optimization/80928
15268 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
15269 (copy_bbs): Set BB_DUPLICATED flag early.
15270 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
15272 (execute_on_shrinking_pred): Likewise.
15273 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
15274 BB_DUPLICATED blocks.
15275 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
15276 iterate over all PHIs considering removal of *gsi.
15278 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
15280 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
15283 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15285 * config/rs6000/rs6000-string.c: (expand_block_clear,
15286 do_load_for_compare, select_block_compare_mode,
15287 compute_current_alignment, expand_block_compare,
15288 expand_strncmp_align_check, expand_strn_compare,
15289 expand_block_move, rs6000_output_load_multiple)
15290 Move functions related to string/block move/compare
15291 to a separate file.
15292 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
15293 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
15294 for this function which is now used in two files.
15295 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
15296 * config.gcc: Add rs6000-string.o to extra_objs for
15297 targets powerpc*-*-* and rs6000*-*-*.
15299 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
15302 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
15303 32-bit, since indexed is not valid for DImode.
15304 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
15305 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
15306 (define_peephole2 for Altivec d-form load): Add 32-bit support.
15307 (define_peephole2 for Altivec d-form store): Likewise.
15310 * multiple_target.c (create_dispatcher_calls): Only create the
15311 dispatcher call if the function is the default clone of a
15312 versioned function.
15314 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
15316 PR middle-end/80902
15317 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
15318 a call, force the call to not be a tail call.
15320 2017-06-23 Jeff Law <law@redhat.com>
15322 * doc/contrib.texi: Add entry for Steven Pemberton's work on
15325 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
15327 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
15328 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
15329 handling for early expansion of vector shifts (sl,sr,sra,rl).
15330 (builtin_function_type): Add vector shift right instructions
15331 to the unsigned argument list.
15333 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
15335 rtl-optimizatoin/79286
15336 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
15337 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
15338 trap. PIC register plus a const unspec without offset can never trap.
15340 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
15342 * tree.h (builtin_structptr_type): New type.
15343 (builtin_structptr_types): Declare new array.
15344 * tree.c (builtin_structptr_types): New array.
15345 (free_lang_data, build_common_tree_nodes): Use it.
15347 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
15350 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
15353 2017-06-22 Matt Turner <mattst88@gmail.com>
15355 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
15356 Lake models to skylake case. Assume skylake for unknown
15357 models with clflushopt.
15359 2017-06-22 Jeff Law <law@redhat.com>
15361 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
15362 frame sizes that do not satisfy aarch64_uimm12_shift.
15364 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
15366 * profile-count.h (apply_probability,
15367 apply_scale, probability_in): Fix checks for zero.
15369 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15371 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
15372 * doc/cppdiropts.texi (-I @var{dir}): Document it.
15374 2016-06-22 Richard Biener <rguenther@suse.de>
15376 * tree-vect-loop.c (vect_model_reduction_cost): Handle
15377 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
15378 REDUC_MAX_EXPR support.
15379 (vectorizable_reduction): Likewise.
15380 (vect_create_epilog_for_reduction): Likewise.
15382 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
15384 * match.pd (A / (1 << B) -> A >> B): New.
15385 * generic-match-head.c: Include optabs-tree.h.
15386 * gimple-match-head.c: Likewise.
15387 * optabs-tree.h (target_supports_op_p): New.
15388 * optabs-tree.c (target_supports_op_p): New.
15390 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15392 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
15393 $gcc_cv_ld --help output.
15394 (gcc_cv_ld_demangle): Likewise.
15395 (gcc_cv_ld_eh_frame_hdr): Likewise.
15396 (gcc_cv_ld_pie): Likewise.
15397 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
15398 (gcc_cv_ld_buildid): Likewise.
15399 (gcc_cv_ld_sysroot): Likewise.
15400 (ld_bndplt_support): Likewise.
15401 (ld_pushpopstate_support): Likewise.
15402 * configure: Regenerate.
15403 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
15405 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15408 * config/i386/sse.md (round<mode>2): Renumber match_dup and
15409 operands indexes to avoid gap between operands and match_dups.
15411 2017-06-21 Andrew Pinski <apinski@cavium.com>
15413 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
15414 Increment Arith_shift and Arith_shift_reg by 1.
15415 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
15417 * config/aarch64/aarch64.c (thunderx_tunings): Enable
15418 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
15419 (aarch64_strip_extend): Add new argument and test for it.
15420 (aarch64_cheap_mult_shift_p): New function.
15421 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
15422 add a cost if it is true.
15423 Update calls to aarch64_strip_extend.
15424 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
15426 2017-06-21 Andrew Pinski <apinski@cavium.com>
15428 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
15430 (thunderxt88): Likewise.
15431 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
15432 (thunderx_prefetch_tune): New variable.
15433 (thunderx2t99_prefetch_tune): Update for the correct values.
15434 (thunderxt88_tunings): New variable.
15435 (thunderx_tunings): Use thunderx_prefetch_tune instead of
15436 generic_prefetch_tune.
15437 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
15439 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15441 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
15442 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
15443 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
15444 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
15445 (aarch64_atomic_cas<mode>, GPI): Likewise.
15447 2017-06-21 Martin Liska <mliska@suse.cz>
15449 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
15450 statements on cold and hot labels.
15451 * predict.c (tree_estimate_probability_bb): Remove the
15452 prediction from this place.
15454 2017-06-21 Martin Liska <mliska@suse.cz>
15456 PR tree-optimization/79489
15457 * gimplify.c (maybe_add_early_return_predict_stmt): New
15459 (gimplify_return_expr): Call the function.
15460 * predict.c (tree_estimate_probability_bb): Remove handling
15462 * predict.def: Update comment about early return predictor.
15463 * gimple-predict.h (is_gimple_predict): New function.
15464 * predict.def: Change default value of early return to 66.
15465 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
15467 * passes.def: Put pass_strip_predict_hints to the beginning of
15470 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
15472 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
15473 FUNCTION_DECL declarations.
15474 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
15476 (dwaf2out_decl): Likewise.
15477 * godump.c (go_early_global_decl): Skip call to the real debug hook
15478 for FUNCTION_DECL declarations.
15479 * passes.c (rest_of_decl_compilation): Skip call to the
15480 early_global_decl debug hook for FUNCTION_DECL declarations, unless
15481 -fdump-go-spec is passed.
15483 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15485 * config/i386/i386.c (struct builtin_isa): New field pure_p.
15486 Reorder for compactness.
15487 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
15488 (def_builtin_pure, def_builtin_pure2): New functions.
15489 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
15491 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15493 * match.pd (nop_convert): New predicate.
15494 ((A +- CST1) +- CST2): Allow some NOP conversions.
15496 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15499 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
15500 with ctors/dtors if GOVD_SHARED is set.
15502 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15504 * config/aarch64/aarch64.md (movti_aarch64):
15505 Emit mov rather than orr.
15506 (movtf_aarch64): Likewise.
15507 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
15508 Emit mov rather than orr.
15510 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15512 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
15513 Swap alternatives, make integer dup more expensive.
15515 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15517 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
15518 Return true for non-tls symbols.
15520 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
15522 * config/aarch64/aarch64-cores.def (cortex-a55): New.
15523 (cortex-a75): Likewise.
15524 (cortex-a75.cortex-a55): Likewise.
15525 * config/aarch64/aarch64-tune.md: Regenerate.
15526 * doc/invoke.texi (-mtune): Document new values for -mtune.
15528 2017-06-21 Tom de Vries <tom@codesourcery.com>
15530 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
15531 stack_size feature.
15532 (Effective-Target Keywords, Other attributes): Suggest using
15533 dg-add-options stack_size feature to get stack limit in stack_size
15534 effective target documentation.
15536 2017-06-21 Julian Brown <julian@codesourcery.com>
15537 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15539 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
15540 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
15541 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
15543 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
15544 attribute type list for neon_multiply.
15545 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
15546 attribute type list for neon_multiply.
15547 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
15548 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
15549 attribute type list for neon_multiply.
15550 * config/arm/types.md (crypto_pmull): Add.
15551 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
15552 attribute type list.
15554 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
15556 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
15559 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15561 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
15562 to make sure not to dereference a NULL cost_classes_ptr pointer.
15564 2017-06-20 Carl Love <cel@us.ibm.com>
15566 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15567 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15568 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15569 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15570 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15571 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15572 VMULOSW): New enum "unspec" values.
15573 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15574 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15575 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15576 altivec_vmulosw): New patterns.
15577 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15578 VMULOSW): Add definitions.
15580 2017-06-20 Julia Koval <julia.koval@intel.com>
15582 * config/i386/i386.c: Fix rounding expand for new pattern.
15583 * config/i386/subst.md: Fix pattern (parallel -> unspec).
15585 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15587 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
15588 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
15590 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15592 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
15595 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15597 * config/aarch64/aarch64-cores.def: Rearrange to sort by
15598 architecture, then by implementer ID.
15599 * config/aarch64/aarch64-tune.md: Regenerate.
15601 2017-06-20 Richard Biener <rguenther@suse.de>
15603 PR middle-end/81097
15604 * fold-const.c (split_tree): Fold to type before negating.
15606 2017-06-20 David Malcolm <dmalcolm@redhat.com>
15608 * diagnostic-show-locus.c
15609 (selftest::test_fixit_deletion_affecting_newline): New function.
15610 (selftest::diagnostic_show_locus_c_tests): Call it.
15612 2017-06-20 Andreas Schwab <schwab@suse.de>
15615 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
15618 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15620 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
15621 __ARM_FEATURE_COPROC according to support.
15623 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15625 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15626 Rewritten to avoid overflow for > 32-bit pointers.
15629 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
15630 by removing enum keyword.
15631 (ubsan_type_descriptor): Likewise. Formatting fix.
15634 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
15635 splitter): Require TARGET_SSE2 in the condition.
15637 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
15640 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
15641 for doing vector set of SFmode on ISA 3.0.
15642 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
15643 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
15645 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
15646 SFmode value into a V4SF variable that was extracted from another
15647 V4SF variable without converting the element to double precision
15648 and back to single precision vector format.
15649 (vsx_insert_extract_v4sf_p9_2): Likewise.
15651 2017-06-19 Jakub Jelinek <jakub@redhat.com>
15653 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
15654 in UWHI to avoid undefined overflow.
15657 * ubsan.h (enum ubsan_encode_value_phase): New.
15658 (ubsan_encode_value): Change second argument to
15659 enum ubsan_encode_value_phase with default value of
15660 UBSAN_ENCODE_VALUE_GENERIC.
15661 * ubsan.c (ubsan_encode_value): Change second argument to
15662 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
15663 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
15664 create_tmp_var_raw instead of create_tmp_var and use a
15666 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
15667 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
15668 ubsan_encode_value callers.
15671 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
15672 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
15673 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
15675 2017-06-19 Richard Biener <rguenther@suse.de>
15677 PR middle-end/81118
15678 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
15679 estimates if we changed anything.
15681 2017-06-19 Richard Biener <rguenther@suse.de>
15683 PR tree-optimization/80887
15684 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
15685 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
15686 simplified lookups, then reset mprts_hook.
15687 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
15689 (try_to_simplify): Likewise.
15691 2017-06-19 Martin Liska <mliska@suse.cz>
15694 * gimplify.c (gimplify_switch_expr):
15695 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
15697 2017-06-19 Martin Liska <mliska@suse.cz>
15699 * doc/install.texi: Document that PGO runs in 4 stages.
15701 2017-06-19 Martin Liska <mliska@suse.cz>
15704 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
15705 to dispatcher function name.
15706 * multiple_target.c (replace_function_decl): New function.
15707 (create_dispatcher_calls): Redirect both edges and references.
15709 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
15711 * profile-count.c (profile_count::dump): Dump quality.
15712 (profile_count::differs_from_p): Update for unsigned val.
15713 * profile-count.h (profile_count_quality): New enum.
15714 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
15716 2017-06-19 Richard Biener <rguenther@suse.de>
15718 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
15719 struct function as arg.
15720 (estimate_numbers_of_iterations): Export overload with loop arg.
15721 (free_numbers_of_iterations_estimates_loop): Use an overload of
15722 free_numbers_of_iterations_estimates instead.
15723 * tree-cfg.c (remove_bb): Adjust.
15724 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
15725 * tree-parloops.c (gen_parallel_loop): Likewise.
15726 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
15728 (tree_unroll_loops_completely): Likewise.
15729 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
15730 Use an overload instead and export.
15731 (estimated_loop_iterations): Adjust.
15732 (max_loop_iterations): Likewise.
15733 (likely_max_loop_iterations): Likewise.
15734 (estimate_numbers_of_iterations): Take struct function as arg
15736 (loop_exits_before_overflow): Adjust.
15737 (free_numbers_of_iterations_estimates_loop): Use an overload.
15738 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
15739 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
15741 2017-06-19 Richard Biener <rguenther@suse.de>
15744 * ipa-prop.c (find_constructor_constant_at_offset): Handle
15745 RANGE_EXPR conservatively.
15747 2017-06-16 Carl Love <cel@us.ibm.com>
15749 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15750 definitions for vec_float, vec_float2, vec_floato,
15751 vec_floate built-ins.
15752 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
15753 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
15755 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
15756 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
15757 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
15758 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
15759 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
15760 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
15761 vec_floato): Add builtin defines.
15762 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
15763 Update the built-in documentation file for the new built-in
15766 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15768 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
15769 (mthumb): Mark as the negative of -marm.
15771 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15773 * doc/invoke.texi (ARM Options, -mcpu): Document supported
15775 (ARM Options, -mtune): Document that this accepts the same
15776 extension options as -mcpu.
15777 (ARM Options, -mfpu): Document addition of -mfpu=auto.
15779 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15781 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
15782 permitted extensions.
15784 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15786 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
15787 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
15788 (armv8-m.main): Add option +nodsp.
15789 * config/arm/arm-cpu-cdata.h: Regenerated.
15791 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15793 * config/arm/t-fuchsia: New file.
15794 * config.gcc (arm*-*-fuchsia*): Use it.
15796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15798 * config/arm/t-symbian: Rewrite for new option infrastructure.
15800 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15802 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
15803 (MULTILIB_REQUIRED): Likewise.
15805 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15807 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
15808 (MULTILIB_RESUE): Likewise.
15809 (MULTILIB_MATCHES): Likewise.
15810 (MULTLIB_REQUIRED): Likewise.
15812 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15814 * config/arm/t-rtems: Rewrite for new option framework.
15816 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15818 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
15819 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
15820 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
15821 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
15822 * config/arm/t-multilib: ... here.
15823 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
15824 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
15825 armv7-a and armv8*-a when A-profile libraries have not been built.
15826 * config/arm/t-rmprofile: Rewrite.
15828 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15830 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
15831 with a backslash. Remove the backslash after substituting unescaped
15833 * doc/fragments.texi (MULTILIB_REUSE): Document it.
15835 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15837 * config.gcc: (arm*-*-*): When building a-profile libraries, force
15838 the driver to pass through the default setting of -mfloat-abi.
15839 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
15841 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
15842 (all_feat_combs): New rule.
15843 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
15845 * config/arm/t-aprofile: Rewrite.
15847 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15849 * config/arm/arm.h (FPUTYPE_AUTO): Define.
15850 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
15851 fpu is not specified by the user/command-line.
15852 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
15853 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
15854 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
15855 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
15856 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
15857 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
15859 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15861 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
15862 * config/arm/t-arm-elf: Rewritten.
15864 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15866 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
15867 have some floating-point instructions.
15868 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
15869 (TARGET_MAYBE_HARD_FLOAT): New macro.
15870 * config/arm/arm-builtins.c (arm_init_builtins): Use
15871 TARGET_MAYBE_HARD_FLOAT.
15872 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
15874 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15876 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
15877 (configargs.h): Include it.
15878 (arm_print_hint_for_fpu_option): New function.
15879 (arm_parse_fpu_option): New function.
15880 (candidate_extension): New class.
15881 (arm_canon_for_multilib): New function.
15882 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
15883 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
15884 (ARCH_CANONICAL_SPECS): New macro.
15885 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
15887 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15889 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
15890 are set after handling multilib fragments. Set target_cpu_default2
15893 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15895 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
15897 (arm*-*-*): Set target_cpu_default2 to a quoted string.
15898 * config/arm/parsecpu.awk (check_cpu): Validate any extension
15900 (check_arch): Likewise.
15901 * config/arm/arm.c (arm_configure_build_target): Handle
15902 TARGET_CPU_DEFAULT being a string constant. Scan any feature
15903 options in the default.
15905 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15907 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
15908 when an option is an alias of another.
15909 * config/arm/parsecpu.awk (optalias): New parser token.
15910 (gen_comm_data): Mark non-alias options as such. Emit entries
15911 for extension aliases.
15912 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
15913 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
15914 (armv6kz, armv6zk, armv6t2): Likewise.
15915 (armv7): Make vfpv3-d16 an alias.
15916 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
15918 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
15919 Sort in canonical order.
15920 (armv8-a): Sort in canonical order.
15921 (armv8.1-a, armv8.2-a): Likewise.
15922 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
15924 (cortex-a9): Sort in canonical order.
15925 * config/arm/arm.c (selftests.h): Include it.
15926 (arm_test_cpu_arch_data): New function.
15927 (arm_run_self_tests): New function.
15928 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
15929 (targetm): Move declaration to the end of the file.
15930 * arm-cpu-cdata.h: Regenerated.
15932 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15934 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
15935 call to target_mode_check describing the type of option passed.
15936 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
15937 (arm_target_thumb_only): Use arm_parse_arch_option_name or
15938 arm_parse_cpu_option_name to match parameters against list of
15940 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
15941 arm_arch_core_flags data structure.
15942 * config/arm/arm-cpu_cdata.h: Regenerated.
15944 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15946 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
15948 (arm_print_hint_for_cpu_option): Likewise.
15949 (arm_print_hint_for_arch_option): Likewise.
15950 (arm_parse_cpu_option_name): Likewise.
15951 (arm_parse_arch_option_name): Likewise.
15952 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
15953 of entries in the all_fpus list.
15954 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
15955 (arm_parse_cpu_option_name): Declare.
15956 (arm_parse_arch_option_name): Declare.
15957 (arm_parse_option_features): Declare.
15958 (arm_intialize_isa): Declare.
15959 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
15961 (gen_comm_data): ... here. Make definitions non-static.
15962 * config/arm/arm-cpu-data.h: Regenerated.
15963 * config/arm/arm-cpu-cdata.h: Regenerated.
15965 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15967 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
15968 (cpu_arch_extension): New structure.
15969 (cpu_arch_option, arch_option, cpu_option): New structures.
15970 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
15971 architecture types.
15972 (gen_data): Generate new format data tables.
15973 * config/arm/arm.c (cpu_tune): New structure.
15974 (cpu_option, processors): Delete.
15975 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
15976 (arm_print_hint_for_cpu_option): ... this and ...
15977 (arm_print_hint_for_arch_option): ... this.
15978 (arm_parse_arch_cpu_name): Delete. Replace with ...
15979 (arm_parse_cpu_option_name): ... this and ...
15980 (arm_parse_arch_option_name): ... this.
15981 (arm_unrecognized_feature): Change type of target parameter to
15983 (arm_parse_arch_cpu_features): Delete. Replace with ...
15984 (arm_parse_option_features): ... this.
15985 (arm_configure_build_target): Rework to use new configuration data
15987 (arm_print_tune_info): Rework for new configuration data tables.
15988 * config/arm/arm-cpu-data.h: Regenerated.
15989 * config/arm/arm-cpu.h: Regenerated.
15991 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15993 * Makefile.in (OBJS): Move sbitmap.o from here ...
15994 (OBJS-libcommon): ... to here.
15996 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15998 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
15999 (ISA_ALL_CRYPTO): New macro.
16000 (ISA_ALL_SIMD): New macro
16001 (ISA_ALL_FP): New macro.
16002 * config/arm/arm.c (fpu_bitlist): Update initializer.
16003 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
16005 (arm9e): Add fpu. Add option for nofp
16006 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
16007 (arm926ej-s, arm1026ej-s): Likewise.
16008 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
16009 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
16010 neon-fp16, neon-vfpv4, nofp and nosimd.
16011 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
16012 (cortex-a8): Add fpu. Add option for nofp.
16013 (cortex-a9): Add fpu. Add options for nosimd and nofp.
16014 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
16015 (cortex-r4f): Add fpu.
16016 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
16017 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
16019 (cortex-r8): Likewise.
16020 (cortex-m4): Add fpu. Add option for nofp.
16021 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
16022 (cortex-a17.cortex-a7): Likewise.
16023 (cortex-a32): Add fpu. Add options for crypto and nofp.
16024 (cortex-a35, cortex-a53): Likewise.
16025 (cortex-a57): Add fpu. Add option for crypto.
16026 (cortex-a72, cortex-a73): Likewise.
16027 (exynos-m1): Likewise.
16028 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
16029 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
16030 (cortex-m33): Add fpu. Add option for nofp.
16031 * config/arm/arm-cpu-cdata.h: Regenerated
16032 * config/arm/arm-cpu-data.h: Regenerated.
16034 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16036 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
16037 (armv5te, armv5tej): Likewise.
16038 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
16039 (armv7): Add options fp and vfpv3-d16.
16040 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
16041 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
16043 (armv7ve): Likewise.
16044 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
16045 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
16046 (armv8-a): Add nocrypto option.
16047 (armv8.1-a, armv8.2-a): Likewise.
16048 (armv8-m.main): add options fp, fp.dp and nofp.
16050 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16052 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
16054 (armv8-a+crc): Delete.
16055 (armv8.1-a): Add options simd, crypto and nofp.
16056 (armv8.2-a): Add options fp16, simd, crypto and nofp.
16057 (armv8.2-a+fp16): Delete.
16058 (armv8-m.main): Add option dsp.
16059 (armv8-m.main+dsp): Delete.
16060 (cortex-a8): Add fpu. Add nofp option.
16061 (cortex-a9): Add fpu. Add nofp and nosimd options.
16062 * config/arm/parsecpu.awk (gen_data): Generate option tables and
16063 link to main cpu and architecture data structures.
16064 (gen_comm_data): Only put isa attributes from the main architecture
16066 (option): New statement for architecture and CPU entries.
16067 * arm.c (struct cpu_option): New structure.
16068 (struct processors): Add entry for options.
16069 (arm_unrecognized_feature): New function.
16070 (arm_parse_arch_cpu_name): Ignore any characters after the first
16072 (arm_parse_arch_cpu_feature): New function.
16073 (arm_configure_build_target): Separate out any CPU and architecture
16074 features and parse separately. Don't error out if -mfpu=auto is
16075 used with only an architecture string.
16076 (arm_print_asm_arch_directives): New function.
16077 (arm_file_start): Call it.
16078 * config/arm/arm-cpu-cdata.h: Regenerated.
16079 * config/arm/arm-cpu-data.h: Likewise.
16080 * config/arm/arm-tables.opt: Likewise.
16082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16084 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
16085 assembler when it is not -mfpu=auto.
16087 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16089 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
16090 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
16091 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
16092 (ASM_CPU_SPEC): Rewrite.
16093 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
16094 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
16095 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
16096 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
16097 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
16098 copied string is NUL-terminated. Also strip any characters prefixed
16100 (arm_rewrite_selected_arch): New function.
16101 (arm_rewrite_march): New function.
16103 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16105 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
16106 (x_arm_cpu_string, x_arm_tune_string): Likewise.
16107 (march, mcpu, mtune): Convert to string-based options.
16108 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
16109 (arm_parse_arch_cpu_name): New function.
16110 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
16111 identify selected architecture or CPU.
16112 (arm_option_save): New function.
16113 (TARGET_OPTION_SAVE): Redefine.
16114 (arm_option_restore): Restore string options.
16115 (arm_option_print): Print string options.
16117 2017-06-16 Martin Sebor <msebor@redhat.com>
16119 PR tree-optimization/80933
16120 PR tree-optimization/80934
16121 * builtins.c (fold_builtin_3): Do not handle bcmp here.
16122 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
16123 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
16124 (gimple_fold_builtin): Call them.
16126 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16128 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
16129 as unlikely; update profile.
16131 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16133 * predict.c (force_edge_cold): Handle declaring edges impossible
16136 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16138 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
16140 (try_unroll_loop_completely): Fix reporting.
16142 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16144 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
16146 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
16149 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
16150 if given a non-constant argument for an intrinsic which requires a
16153 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16155 * profile.c (compare_freqs): New function.
16156 (branch_prob): Sort edge list.
16157 (find_spanning_tree): Assume that the list is priority sorted.
16159 2017-06-16 Richard Biener <rguenther@suse.de>
16161 PR tree-optimization/81090
16162 * passes.def (pass_record_bounds): Remove.
16163 * tree-pass.h (make_pass_record_bounds): Likewise.
16164 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
16165 make_pass_record_bounds): Likewise.
16166 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
16167 not free niter estimates at the beginning but at the end.
16168 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
16170 2017-06-16 Richard Biener <rguenther@suse.de>
16172 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
16173 initializer to workaround ICE in host GCC 4.8.
16175 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16177 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
16179 (clone_inlined_nodes): Update.
16181 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16183 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
16184 prefetch settings, and enable prefetching by default at -O3.
16186 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16188 * config/aarch64/aarch64.c (aarch64_override_options_internal):
16189 Set flag_prefetch_loop_arrays according to tuning data.
16191 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16193 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
16194 New tune structure.
16195 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
16196 [Unrelated to main purpose of the patch] Place the pointer field last
16197 to enable type checking errors when tune structure are wrongly merged.
16198 * config/aarch64/aarch64.c (generic_prefetch_tune,)
16199 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
16200 (thunderx2t99_prefetch_tune): New tune constants.
16201 (tune_params *_tunings): Update all tunings (no functional change).
16202 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
16203 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
16204 from tunings structures.
16206 2017-06-16 Jakub Jelinek <jakub@redhat.com>
16209 * ubsan.c (instrument_null): Add T argument, use it instead
16210 of computing it based on IS_LHS.
16211 (instrument_object_size): Likewise.
16212 (pass_ubsan::execute): Adjust instrument_null and
16213 instrument_object_size callers to pass gimple_get_lhs or
16214 gimple_assign_rhs1 result to it. Use instrument_null instead of
16215 calling get_base_address and instrument_mem_ref. Handle
16216 aggregate call arguments for object-size sanitization.
16218 2017-06-16 Yury Gribov <tetra2005@gmail.com>
16220 PR tree-optimization/81089
16221 * tree-vrp.c (is_masked_range_test): Validate operands of
16224 2017-06-15 Martin Sebor <msebor@redhat.com>
16227 * dumpfile.c (dump_register): Avoid calling memset to initialize
16228 a class with a default ctor.
16229 * gcc.c (struct compiler): Remove const qualification.
16230 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
16231 * hash-table.h: Ditto.
16232 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
16234 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
16235 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
16237 * params.h (struct param_info): Make struct members non-const.
16238 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
16239 with default initialization.
16240 * vec.h (vec_copy_construct, vec_default_construct): New helper
16242 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
16243 with vec_copy_construct.
16244 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
16245 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
16246 * doc/invoke.texi (-Wclass-memaccess): Document.
16248 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16250 * emit-rtl.h (is_leaf): Update comment about local
16251 register allocator.
16253 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16256 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
16257 for a variable to have a section before checking if the section has a
16259 Set section to.persistent if persistent attribute is set.
16260 Warn if .persistent attribute is used on an automatic variable.
16262 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
16264 PR rtl-optimization/80474
16265 * reorg.c (update_block): Do not ignore instructions in a delay slot.
16267 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
16269 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
16272 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
16274 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
16275 (casesi): Emit bounds checking as RTL.
16276 (casesi_internal_mips16_<mode>): Remove bounds checking.
16278 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
16280 * config/xtensa/xtensa.c (xtensa_option_override): Append
16281 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
16282 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
16283 xtensa_doloop_hooks): Define unconditionally.
16284 (xtensa_reorg_loops): Only call reorg_loops in the presence of
16286 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
16287 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
16288 for it in xtensa_option_override.
16289 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
16290 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
16292 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
16294 * doc/cppopts.texi: Document '-' special value to -MF.
16296 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
16298 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
16299 (cortex_a53_fconst): Likewise.
16300 (cortex_a53_fpmul): Likewise.
16301 (cortex_a53_f_load_64): Likewise.
16302 (cortex_a53_f_load_many): Likewise.
16303 (cortex_a53_advsimd_alu): Likewise.
16304 (cortex_a53_advsimd_alu_q): Likewise.
16305 (cortex_a53_advsimd_mul): Likewise.
16306 (cortex_a53_advsimd_mul_q): Likewise.
16307 (fpmac bypass): Add new bypass for fpmac-fpmac case.
16308 Add missing fmul, r2f_cvt and fconst cases.
16310 2017-06-14 Richard Biener <rguenther@suse.de>
16312 PR middle-end/81088
16313 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
16315 (fold_binary_loc): When associating do not treat pre-existing
16316 TREE_OVERFLOW on literal constants as a reason to allow
16317 TREE_OVERFLOW on associated literal constants.
16319 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16321 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
16322 (MASK_FEATURES): New macro.
16323 * config/sparc/sparc.c (sparc_option_override): Remove the special
16324 handling of -mfpu and generalize it to all MASK_FEATURES switches.
16326 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16328 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
16329 a division of 0 if non-call exceptions are enabled.
16331 2017-06-14 Andrew Pinski <apinski@cavium.com>
16332 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16335 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
16336 Improve vector initialization code gen for only variable case.
16338 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16340 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
16342 2017-06-14 Richard Biener <rguenther@suse.de>
16344 PR tree-optimization/81083
16345 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
16348 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16350 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
16351 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
16352 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
16353 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
16354 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
16355 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
16357 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16359 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
16360 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
16362 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16364 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
16366 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16368 * config/rs6000/t-rtems: Don't handle SPE.
16370 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16372 * config/rs6000/t-linux: Don't handle SPE.
16374 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16376 * config/rs6000/eabispe.h: Delete file.
16378 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16380 * config/rs6000/t-spe: Delete file.
16382 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16384 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
16385 (rs6000_legitimate_offset_address_p): Return false for anything in
16386 V2SImode or V2SFmode.
16388 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16390 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
16391 except V2SF and V2SI. Rearrange the vector modes, and add comments.
16392 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
16394 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
16395 (rs6000_legitimate_offset_address_p): Ditto.
16396 (rs6000_emit_move): Ditto.
16397 (rs6000_init_builtins): Remove V4HI_type_node.
16399 2017-06-13 Martin Liska <mliska@suse.cz>
16402 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
16403 (gate_asan): Likewise.
16404 * asan.h (asan_no_sanitize_address_p): Remove the function.
16405 (sanitize_flags_p): New function.
16406 * builtins.def: Fix coding style.
16407 * common.opt: Use renamed enum value.
16408 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
16409 * doc/extend.texi: Document no_sanitize attribute.
16410 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
16411 to SANITIZE_UNDEFINED_NONDEFAULT.
16412 * gcc.c (sanitize_spec_function): Use the renamed enum value.
16413 * gimple-fold.c (optimize_atomic_compare_exchange_p):
16414 Use sanitize_flags_p.
16415 * gimplify.c (gimplify_function_tree): Likewise.
16416 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
16417 * opts.c (parse_no_sanitize_attribute): New function.
16418 (common_handle_option): Use renamed enum value.
16419 * opts.h (parse_no_sanitize_attribute): Declare.
16420 * tree.c (sanitize_flags_p): New function.
16421 * tree.h: Declared here.
16422 * tsan.c: Use sanitize_flags_p.
16423 * ubsan.c (ubsan_expand_null_ifn): Likewise.
16424 (instrument_mem_ref): Likewise.
16425 (instrument_bool_enum_load): Likewise.
16426 (do_ubsan_in_current_function): Remove the function.
16427 (pass_ubsan::execute): Use sanitize_flags_p.
16428 * ubsan.h: Remove do_ubsan_in_current_function
16429 * tree-cfg.c (print_no_sanitize_attr_value): New function.
16430 (dump_function_to_file): Use it here.
16432 2017-06-13 Martin Jambor <mjambor@suse.cz>
16434 PR tree-optimization/80803
16435 PR tree-optimization/81063
16436 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
16437 (propagate_subaccesses_across_link): Enqueue subtree whenever
16438 necessary instead of relying on the caller.
16440 2017-06-13 Martin Jambor <mjambor@suse.cz>
16442 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
16443 that have a first_link.
16444 (sort_and_splice_var_accesses): Do not check first_link before
16446 (subtree_mark_written_and_enqueue): Likewise.
16447 (propagate_all_subaccesses): Likewise and do not stop at first
16448 parent with a first_link.
16450 2017-06-13 Martin Jambor <mjambor@suse.cz>
16452 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
16455 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16457 * match.pd: New pattern.
16459 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16461 * tree-vrp.c (is_masked_range_test): New function.
16462 (register_edge_assert_for): Determine ranges for
16465 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16467 PR tree-optimization/67328
16468 * fold-const.c (maskable_range_p): New function.
16469 (build_range_check): Generate bittests if possible.
16471 2017-06-13 Martin Liska <mliska@suse.cz>
16473 * gimple-pretty-print.c (dump_probability): Add new argument.
16474 (dump_edge_probability): Dump both probability and count.
16475 (dump_gimple_label): Likewise.
16476 (dump_gimple_bb_header): Likewise.
16478 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
16481 * config/avr/avr-devices.c: Fix indentation.
16482 * config/avr/gen-avr-mmcu-specs.c: Dito.
16484 2017-06-13 Richard Biener <rguenther@suse.de>
16486 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
16487 instead get vector type from stmt_info.
16488 (vectorizable_reduction): Adjust. Remove dead code.
16490 2017-06-13 Richard Biener <rguenther@suse.de>
16492 PR middle-end/81065
16493 * fold-const.c (extract_muldiv_1): Remove bogus distribution
16494 case of C * (x * C2 + C3).
16495 (fold_addr_of_array_ref_difference): Properly fold index difference.
16497 2017-06-12 David S. Miller <davem@davemloft.net>
16500 * config/sparc/sparc.md (return expander): Emit frame blockage if
16501 function uses alloca.
16503 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16505 * combine.c (make_field_assignment): Check len rather than the mode
16506 precision when calling force_to_mode.
16508 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
16510 Support multilibs and devices that see flash in RAM address range.
16513 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
16514 (avr_mcu_t) <flash_pm_offset>: New field.
16515 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
16516 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
16517 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
16518 (AVR_TINY_PM_OFFSET): Remove macro.
16519 * config/avr/avr.opt (-mshort-calls): New option.
16520 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16521 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
16522 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
16523 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
16524 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
16525 instead of avr_arch->have_jmp_call.
16526 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
16527 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
16528 avr_arch->flash_pm_offset to define.
16529 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
16530 new field flash_pm_offset. Add entry for avrxmega3.
16531 (avr_texinfo): Add entry for avrxmega3.
16532 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
16533 attiny212, attiny214,
16534 attiny412, attiny414, attiny416, attiny417,
16535 attiny814, attiny816, attiny817,
16536 attiny1614, attiny1616, attiny1617,
16537 attiny3214, attiny3216, attiny3217.
16538 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
16539 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
16540 (avr_print_operand_address) [AVR_TINY]: Same.
16541 (avr_asm_init_sections) <readonly_data_section>: Only patch
16542 callback if avr_arch->flash_pm_offset = 0.
16543 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
16544 for rodata if avr_arch->flash_pm_offset != 0.
16545 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
16546 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
16547 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
16548 (m_options): Append opt_rcall.
16549 (m_dirnames): Append dir_rcall.
16550 * config/avr/t-multilib: Regenerate.
16552 * configure.ac [target=avr]: Check whether avrxmega3 default
16553 linker description file works as needed.
16554 * configure: Regenerate.
16555 * doc/avr-mmcu.texi: Regenerate.
16556 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
16557 <__AVR_ARCH__>: Document avrxmega3 and 103.
16558 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
16559 <__AVR_SHORT_CALLS__>: Document it.
16560 <__AVR_PM_BASE_ADDRESS__>: Document it.
16561 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
16562 (AVR Variable Attributes) <progmem>: Document this is
16563 not needed for avrxmega3.
16564 (AVR Named Address Spaces) <__flash>: Dito.
16566 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
16568 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
16570 2017-06-12 Doug Rupp <rupp@adacore.com>
16572 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
16573 Append vxworks-stdint.h to the tm_file list.
16574 * config/vxworks-stdint.h: New file.
16576 2017-06-12 Martin Liska <mliska@suse.cz>
16578 PR tree-optimization/81041
16579 * tree-profile.c (gimple_gen_ic_func_profiler):
16580 Create an extra BB in profile-generate
16581 (gimple_gen_time_profiler): Likewise.
16583 2017-06-12 Jakub Jelinek <jakub@redhat.com>
16585 PR tree-optimization/81003
16586 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
16587 (update_range_test): Use it instead of force_gimple_operand_gsi.
16589 2017-06-12 Richard Biener <rguenther@suse.de>
16591 PR tree-optimization/81053
16592 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
16593 with backedge value not defined in loop. Simplify def stmt
16596 2017-06-11 Tom de Vries <tom@codesourcery.com>
16599 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
16601 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
16602 nvptx_cannot_force_const_mem.
16604 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16606 * opts.c (finish_options): Move test for flag_split_stack after
16607 it has been initialized.
16609 2017-06-11 Jason Merrill <jason@redhat.com>
16611 * tree.h (id_equal): New.
16612 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
16613 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
16614 instead of strcmp of IDENTIFIER_POINTER.
16616 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16618 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
16619 (mark_all_inlined_calls_cdtor): Fix formating.
16620 (inline_transform): Rescale profile before inlining.
16622 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16624 * cgraph.h (cgraph_edge::clone): Update prototype.
16625 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
16626 (cgraph_node::create_clone): Update.
16627 (cgraph_node::create_version_clone): Update.
16628 * tree-inline.c (copy_bb): Update.
16629 (expand_call_inline): Update.
16631 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16633 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
16634 factored out from ...
16635 (rs6000_emit_prologue): ... here.
16637 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16639 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
16640 factored out from ...
16641 (rs6000_emit_prologue): ... here.
16643 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16645 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
16647 (handle_missing_profiles): Fix computation of tp_first_run.
16648 (counts_to_freqs): Do not touch freqs when count is 0.
16650 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16652 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
16655 2017-06-10 Tom de Vries <tom@codesourcery.com>
16657 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
16658 attributes): Document signal effective target.
16660 2017-06-10 Tom de Vries <tom@codesourcery.com>
16662 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
16663 Document effective target stack_size.
16665 2017-06-09 David Malcolm <dmalcolm@redhat.com>
16667 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
16668 to the edit_context if they can be auto-applied.
16670 2017-06-9 Ian Lance Taylor <iant@golang.org>
16672 * opts.c (finish_options): If -fsplit-stack, disable implicit
16673 -forder-blocks-and-partition.
16674 * doc/invoke.texi (Optimize Options): Document that when using
16675 -fsplit-stack -forder-blocks-and-partition is not implicitly
16678 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16680 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
16681 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
16682 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
16683 * builtins.def (abort, trap, unreachable): Declare cold.
16684 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
16685 * tree-core.h (ECF_COLD): New.
16686 * tree.c (set_call_expr_flags): Handle ECF_COLD.
16687 (build_common_builtin_nodes): Mark unreachable and abort as cold.
16689 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16691 * predict.c (unlikely_executed_stmt_p): Cleanup.
16693 2017-06-09 Richard Biener <rguenther@suse.de>
16695 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
16696 model if the ref is always written to.
16698 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16700 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
16702 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16704 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
16707 2017-06-09 Tom de Vries <tom@codesourcery.com>
16710 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
16711 "target cannot support label values" when encountering LABEL_REF.
16713 2017-06-09 Martin Liska <mliska@suse.cz>
16715 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
16716 (gimple_gen_ic_func_profiler): Emit direct comparison
16717 of __gcov_indirect_call_callee with NULL.
16718 (gimple_gen_time_profiler): Change probability from
16719 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
16721 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16723 * profile.c (edge_gcov_counts): Turn to pointer.
16724 (compute_branch_probabilities, compute_branch_probabilities): Update.
16725 (branch_prob): Do not clear edge_gcov_count.
16726 * profile.h (edge_gcov_counts): Turn to pointer.
16727 (edge_gcov_count): Update.
16729 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16731 * gimple.h (gimple_check_failed): Mark cold.
16733 2017-06-09 Richard Biener <rguenther@suse.de>
16735 PR tree-optimization/66623
16736 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
16737 refactor check_reduction into two parts, properly computing
16738 whether we have to check reduction validity for outer loop
16741 2017-06-09 Richard Biener <rguenther@suse.de>
16743 PR tree-optimization/79483
16744 * graphite-scop-detection.c (order): New global.
16745 (get_order): Compute bb to order mapping that satisfies code
16746 generation constraints.
16747 (cmp_pbbs): New helper.
16748 (build_scops): Start domwalk at entry block, sort generated
16751 2017-06-09 Richard Biener <rguenther@suse.de>
16753 PR middle-end/81007
16754 * ipa-polymorphic-call.c
16755 (ipa_polymorphic_call_context::restrict_to_inner_class):
16756 Skip FIELD_DECLs with error_mark_node type.
16757 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
16760 2017-06-09 Martin Liska <mliska@suse.cz>
16762 * predict.c (struct branch_predictor): New struct.
16763 (test_prediction_value_range): New test.
16764 (predict_c_tests): New function.
16765 * selftest-run-tests.c (selftest::run_tests): Run the function.
16766 * selftest.h: Declare new tests.
16768 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
16771 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
16772 gen_add3_insn did not fail.
16773 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
16774 r0, construct that number in a temporary reg and add that reg to r0.
16775 If asked to put the result in r0 as well, fail.
16777 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
16779 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16780 for early expansion of vec_eqv.
16782 2017-06-08 Jakub Jelinek <jakub@redhat.com>
16784 PR middle-end/81005
16785 * ubsan.c (instrument_null): Avoid pointless code temporary.
16786 (pass_ubsan::execute): Instrument aggregate arguments of calls.
16788 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
16792 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
16795 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
16798 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16800 * predict.c (unlikely_executed_edge_p): Move ahead.
16801 (probably_never_executed_edge_p): Use it.
16803 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16805 PR middle-end/79988
16806 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
16807 gimple_call_builtin_p call.
16809 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16811 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
16812 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
16813 rtl_check_failed_type2, rtl_check_failed_code1,
16814 rtl_check_failed_code2, rtl_check_failed_code_mode,
16815 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
16816 rtvec_check_failed_bounds, rtl_check_failed_flag,
16817 _fatal_insn_not_found, _fatal_insn): Likewise.
16818 * tree.h (tree_contains_struct_check_failed,
16819 tree_check_failed, tree_not_check_failed,
16820 tree_class_check_failed, tree_range_check_failed,
16821 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
16822 tree_vec_elt_check_failed, phi_node_elt_check_failed,
16823 tree_operand_check_failed, omp_clause_check_failed,
16824 omp_clause_operand_check_failed, omp_clause_range_check_failed):
16827 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16829 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
16830 flag_branch_probabilities.
16831 * ipa-inline.c (edge_badness): Likewise.
16832 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16833 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
16834 * predict.c (maybe_hot_frequency_p): Likewise.
16835 (probably_never_executed): Likewise.
16836 * sched-ebb.c (schedule_ebbs): Likewise.
16837 * sched-rgn.c (find_single_block_region): Likewise.
16838 * tracer.c (tail_duplicate): Likewise.
16840 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16842 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
16843 longer requires x_flag_profile_use.
16845 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16847 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
16848 instead of flag_reorder_blocks_and_partition.
16849 * dbxout.c (dbxout_function_end): Likewise.
16850 * dwarf2out.c (gen_subprogram_die): Likewise.
16851 * haifa-sched.c (sched_create_recovery_edges): Likewise.
16852 * hw-doloop.c (reorg_loops): Likewise.
16853 * varasm.c (assemble_start_function,
16854 assemble_end_function): Likewise.
16855 (decide_function_section): Do not check for
16856 flag_reorder_blocks_and_partition.
16858 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16860 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16862 (chkp_get_hard_register_fake_addr_expr): Ditto.
16863 (chkp_build_addr_expr): Add check for hard reg case.
16864 (chkp_parse_array_and_component_ref): Ditto.
16865 (chkp_find_bounds_1): Ditto.
16866 (chkp_process_stmt): Don't generate bounds store for
16869 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16871 * predict.c (maybe_hot_bb_p): Do not check profile status.
16872 (maybe_hot_edge_p): Likewise.
16873 (probably_never_executed): Check for zero counts even if profile
16875 (unlikely_executed_edge_p): New function.
16876 (unlikely_executed_stmt_p): New function.
16877 (unlikely_executed_bb_p): New function.
16878 (set_even_probabilities): Use unlikely predicates.
16879 (combine_predictions_for_bb): Likewise.
16880 (predict_paths_for_bb): Likewise.
16881 (predict_paths_leading_to_edge): Likewise.
16882 (determine_unlikely_bbs): New function.
16883 (estimate_bb_frequencies): Use it.
16884 (compute_function_frequency): Use zero counts even if profile is
16886 * profile-count.h: Fix typo.
16888 2017-08-08 Julia Koval <julia.koval@intel.com>
16890 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
16891 _mm512_mask_cvtsepi16_storeu_epi8,
16892 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
16893 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
16894 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
16895 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
16896 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
16897 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
16898 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
16899 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
16900 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
16901 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
16902 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
16903 __builtin_ia32_pmovuswb256mem_mask,
16904 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
16905 __builtin_ia32_pmovwb512mem_mask): New builtins.
16907 2017-08-08 Julia Koval <julia.koval@intel.com>
16909 PR target/73350,80862
16910 * config/i386/subst.md (round): Fix round pattern.
16911 * config/i386/i386.c (ix86_erase_embedded_rounding):
16912 Fix erasing rounding for the fixed pattern.
16914 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16916 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
16918 2017-06-08 Martin Liska <mliska@suse.cz>
16920 PR gcov-profile/80911
16921 * gcov.c (block_info::block_info): New constructor.
16923 2017-06-07 Carl Love <cel@us.ibm.com>
16925 * config/rs6000/rs6000-c: The return type of the following
16926 built-in functions was implemented as int not long long. Fix sign
16927 of return value for the unsigned version of vec_mulo and vec_mule.
16928 vector unsigned long long vec_bperm (vector unsigned long long,
16929 vector unsigned char)
16930 vector signed long long vec_mule (vector signed int,
16932 vector unsigned long long vec_mule (vector unsigned int,
16933 vector unsigned int)
16934 vector signed long long vec_mulo (vector signed int,
16936 vector unsigned long long vec_mulo (vector unsigned int,
16937 vector unsigned int)
16938 * doc/extend.texi: Fix the documentation for the built-in
16941 2017-06-07 Carl Love <cel@us.ibm.com>
16944 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
16947 2017-06-07 Carl Love <cel@us.ibm.com>
16949 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
16950 support, Generate doublehv for signed int/float for BE case only.
16952 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
16954 * doc/invoke.texi (mcx16): Rewrite.
16956 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16958 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
16959 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
16960 *mov<mode>_softfloat, and an anonymous splitter): Use
16961 nonimmediate_operand instead of rs6000_nonimmediate_operand.
16963 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16965 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
16967 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16968 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
16969 (rs6000_debug_reg_global): Adjust.
16970 (rs6000_init_hard_regno_mode_ok): Adjust.
16971 (rs6000_dbx_register_number): Adjust.
16972 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
16973 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16974 Remove SPE_ACC and SPEFSCR.
16975 (REG_ALLOC_ORDER): Ditto.
16976 (FRAME_POINTER_REGNUM): Change to 111.
16977 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
16978 (REG_CLASS_NAMES): Ditto.
16979 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
16980 (REGISTER_NAMES): Ditto.
16981 (ADDITIONAL_REG_NAMES): Ditto.
16982 (rs6000_reg_names): Ditto.
16983 * config/rs6000/rs6000.md: Renumber some register number
16986 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16988 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
16990 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16991 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
16993 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
16994 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16995 Delete the SPE high registers.
16996 (REG_ALLOC_ORDER): Ditto.
16997 (enum reg_class): Remove SPE_HIGH_REGS.
16998 (REG_CLASS_NAMES): Ditto.
16999 (REG_CLASS_CONTENTS): Delete the SPE high registers.
17000 (REGISTER_NAMES): Ditto.
17001 (rs6000_reg_names): Ditto.
17002 * doc/tm.texi.in: Remove SPE as example.
17003 * doc/tm.texi: Regenerate.
17005 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17007 * config/rs6000/8540.md (ppc8540_brinc): Delete.
17008 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
17009 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
17010 * config/rs6000/rs6000.md (type): Remove "brinc".
17012 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17014 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
17015 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
17016 * config/rs6000/linuxspe.h: Delete file.
17017 * config/rs6000/rs6000.md: Don't include spe.md.
17018 * config/rs6000/spe.h: Delete file.
17019 * config/rs6000/spe.md: Delete file.
17020 * config/rs6000/t-rs6000: Remove spe.md.
17022 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17024 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
17025 (reg_or_none500mem_operand): Delete.
17026 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
17027 instead of reg_or_none500mem_operand.
17029 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17031 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17032 handling of SPE flags.
17033 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
17035 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17037 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17039 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
17040 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
17041 paired_divv2sf3): Similar.
17042 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
17043 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
17044 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
17046 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
17047 Rename the paired_* instruction patterns.
17048 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
17050 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
17051 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
17052 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
17053 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
17054 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
17055 PAIRED_VECTOR_MODE.
17056 (struct machine_function): Delete field spe_insn_chain_scanned_p.
17057 (spe_func_has_64bit_regs_p): Delete.
17058 (spe_expand_predicate_builtin): Delete.
17059 (spe_expand_evsel_builtin): Delete.
17060 (TARGET_DWARF_REGISTER_SPAN): Do not define.
17061 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
17062 (invalid_e500_subreg): Delete.
17063 (rs6000_legitimize_address): Always force_reg op2 as well, for
17064 paired single memory accesses.
17065 (rs6000_member_type_forces_blk): Delete.
17066 (rs6000_spe_function_arg): Delete.
17067 (rs6000_expand_unop_builtin): Delete SPE handling.
17068 (rs6000_expand_binop_builtin): Ditto.
17069 (spe_expand_stv_builtin): Delete.
17070 (bdesc_2arg_spe): Delete.
17071 (spe_expand_builtin): Delete.
17072 (spe_expand_predicate_builtin): Delete.
17073 (spe_expand_evsel_builtin): Delete.
17074 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
17075 (spe_init_builtins): Delete.
17076 (spe_func_has_64bit_regs_p): Delete.
17077 (savres_routine_name): Delete "info" parameter. Adjust callers.
17078 (rs6000_emit_stack_reset): Ditto.
17079 (rs6000_dwarf_register_span): Delete.
17080 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
17081 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
17082 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
17084 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
17086 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
17087 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
17088 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
17089 mulv2sf3, divv2sf3): Delete expanders.
17091 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17093 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
17095 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17097 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
17098 * config/rs6000/rs6000.c: Ditto.
17100 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17102 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
17103 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
17104 comparison_operator.
17106 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17108 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
17109 * config/rs6000/rs6000.opt: Ditto.
17110 * config/rs6000/t-rtems: Ditto.
17112 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17114 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
17115 TARGET_E500_SINGLE by 0, simplify.
17116 * config/rs6000/rs6000.c: Ditto.
17117 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
17118 (spe_build_register_parallel): Delete.
17119 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
17120 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
17121 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
17122 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
17123 (E500_CONVERT): Delete.
17124 * config/rs6000/spe.md: Remove many patterns and all define_constants.
17126 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17128 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
17129 * config/rs6000/dfp.md: Ditto.
17130 (negdd2, *negdd2_fpr): Merge.
17131 (absdd2, *absdd2_fpr): Merge.
17132 (negtd2, *negtd2_fpr): Merge.
17133 (abstd2, *abstd2_fpr): Merge.
17134 * config/rs6000/e500.h: Delete file.
17135 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
17136 TARGET_FPRS by 1 and simplify.
17137 * config/rs6000/rs6000-c.c: Ditto.
17138 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
17139 TARGET_DF_SPE by 0.
17140 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
17142 * config/rs6000/rs6000.md: Ditto.
17143 (floatdidf2, *floatdidf2_fpr): Merge.
17144 (move_from_CR_gt_bit): Delete.
17145 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
17146 (E500_CR_IOR_COMPARE): Delete.
17147 (All patterns that require !TARGET_FPRS): Delete.
17148 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
17150 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17152 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
17154 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17156 * graphds.c (add_edge): Intitialize edge's attached data.
17157 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
17158 pointer parameter. Call pointed function on each edge during
17159 graph traversing. Skip traversing the edge when the function
17161 (graphds_dfs, graphds_scc): Ditto.
17162 (for_each_edge): New parameter. Pass the new parameter to callback
17164 * graphds.h (skip_edge_callback): New function pointer type.
17165 (graphds_dfs, graphds_scc): New function pointer parameter.
17166 (graphds_edge_callback, for_each_edge): New parameter.
17168 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17170 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
17171 out code checking if runtime alias check is possible to below ...
17172 Call the new function.
17173 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
17174 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
17176 2017-06-07 Marek Polacek <polacek@redhat.com>
17179 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
17180 TYPE_OVERFLOW_WRAPS check.
17182 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17184 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
17185 if versioning is required.
17186 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
17187 peeling with the check for versioning.
17189 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17191 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
17192 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
17193 Set true to new parameter if new ssa variable is defined.
17194 (vect_gen_vector_loop_niters): Refactor. Set range information
17195 for the new vector loop bound variable.
17196 (vect_do_peeling): Ditto.
17198 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17200 * tree-affine.c (ssa.h): Include header file.
17201 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
17202 has wrapping overflow behavior.
17204 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17206 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
17208 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17210 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
17211 (tree_to_aff_combination): ... here.
17213 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17215 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
17216 reg_pressure model function.
17217 (ivopts_global_cost_for_size): Delete.
17218 (determine_set_costs, iv_ca_recount_cost): Call new model function
17219 ivopts_estimate_reg_pressure.
17221 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17223 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
17224 expensive than udiv. Remove floating point cases from mod.
17226 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17228 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
17229 Increase idiv cost.
17231 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17233 * config/aarch64/aarch64.md
17234 (copysignsf3): Fix mask generation.
17236 2017-06-07 Jakub Jelinek <jakub@redhat.com>
17238 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
17240 (class dump_manager): Add register_dumps method.
17241 * dumpfile.c: Include langhooks.h.
17242 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
17243 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
17244 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
17245 (dump_manager::dump_register): Start with 512 entries instead of 32.
17246 (dump_manager::register_dumps): New method.
17247 * toplev.c (general_init): Instead of invoking register_dumps
17248 langhook, invoke register_dumps method on the dump manager.
17249 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
17252 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
17254 * doc/md.texi: Clarify the restrictions on a define_insn condition.
17255 Say that # requires an associated define_split to exist, and that
17256 the define_split must be suitable for use after register allocation.
17258 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17260 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
17261 (compute_outgoing_frequencies): Also initialize zero counts.
17262 (find_many_sub_basic_blocks): Do not produce uninitialized profile
17263 around loops; preserve more of profile when nothing changes.
17265 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
17267 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
17269 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
17270 * config/arm/arm-cpu-cdata.h: Regenerate.
17271 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
17272 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
17273 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
17274 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
17275 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
17277 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
17279 * config/arm/t-rmprofile: Likewise.
17280 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
17282 2017-06-06 David S. Miller <davem@davemloft.net>
17285 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
17286 blockage if function uses alloca.
17288 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17290 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
17291 New "uid" fields to hold pretty-print IDs of group and ref.
17292 Memory references are now identified as <group_id>:<ref_id>
17293 instead of using [random] addresses.
17294 (dump_mem_details): Simplify, no functional change.
17295 (dump_mem_ref): Simplify and make output more concise.
17296 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
17297 (find_or_create_group): Initialize group uid.
17298 (record_ref): Initialize ref uid. Improve debug output.
17299 (prune_group_by_reuse, should_issue_prefetch_p,)
17300 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
17301 (mark_nontemporal_store, determine_loop_nest_reuse):
17302 Improve debug output.
17304 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17306 * dbgcnt.def (prefetch): New debug counter.
17307 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
17308 (schedule_prefetches): Stop issueing prefetches if debug counter
17311 2017-06-06 Tom de Vries <tom@codesourcery.com>
17313 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
17314 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
17316 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17318 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
17319 Use aarch64_reg_or_zero predicate for operand 4.
17320 (aarch64_compare_and_swap<mode> define_insn_and_split):
17321 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
17322 (aarch64_store_exclusive<mode>): Likewise for operand 2.
17324 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17326 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
17327 (arm_compute_save_core_reg_mask): This.
17328 (thumb1_compute_save_reg_mask): Rename into ...
17329 (thumb1_compute_save_core_reg_mask): This.
17330 (arm_compute_save_reg0_reg12_mask): Adapt comment.
17331 (arm_compute_frame_layout): Likewise.
17333 2017-06-06 Richard Biener <rguenther@suse.de>
17335 PR tree-optimization/80974
17336 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
17337 keep or clear leaders SSA info.
17339 2017-06-06 Tom de Vries <tom@codesourcery.com>
17341 * config/nvptx/nvptx.c (split_mode_p): New function.
17342 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
17344 2017-06-06 Tom de Vries <tom@codesourcery.com>
17346 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
17348 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17351 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
17354 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17356 * shrink-wrap.c (handle_simple_exit): Update profile.
17357 (try_shrink_wrapping): Upate profile.
17359 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17361 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
17362 (tree_guess_outgoing_edge_probabilities): New.
17363 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
17364 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
17366 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17368 * ipa-split.c (split_function): Initialize return bb profile.
17370 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17372 * profile.c (compute_branch_probabilities): Also initialize
17373 EXIT_BLOCK profile.
17375 2017-06-06 Richard Biener <rguenther@suse.de>
17377 PR tree-optimization/80928
17378 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
17379 (vect_analyze_loop_operations): Properly guard analysis for
17381 (vect_transform_loop): Likewise.
17382 (vect_analyze_loop_2): Also reset SLP type on PHIs.
17383 (vect_model_induction_cost): Do not cost for pure SLP.
17384 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
17385 of induction in inner loop vectorization.
17386 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
17387 (vect_get_and_check_slp_defs): Handle vect_induction_def.
17388 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
17390 (vect_analyze_slp_cost_1): Cost induction.
17391 (vect_detect_hybrid_slp_stmts): Handle PHIs.
17392 (vect_get_slp_vect_defs): Likewise.
17393 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
17394 (vect_transform_stmt): Handle SLP reductions.
17395 * tree-vectorizer.h (vectorizable_induction): Adjust.
17397 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17399 * config/rs6000/rs6000.c (make_resolver_func): Update
17400 init_lowered_empty_function call.
17402 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
17404 * doc/invoke.texi: Document the -fprofile-abs-path option.
17405 * common.opt (fprofile-abs-path): New option.
17406 * gcov-io.h (gcov_write_filename): Declare.
17407 * gcov-io.c (gcov_write_filename): New function.
17408 * coverage.c (coverage_begin_function): Use gcov_write_filename.
17409 * profile.c (output_location): Likewise.
17411 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17413 * shring-wrap.c: Revert accidental commit.
17415 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
17417 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
17419 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17421 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
17423 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
17424 profile in callgraph edge.
17425 * profile-count.h (apply_probability): If THIS is 0, then result is 0
17426 (apply_scale): Likewise.
17427 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
17428 Also scale profile when inlining function with zero profile.
17429 (initialize_cfun): Update exit block profile even when it is zero.
17430 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
17431 when profile is read.
17433 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17435 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
17436 (CLONE_*): New constants to define the processors we can generate
17437 code for with the target_clone attribute.
17438 (rs6000_clone_map): New array to identify which clone processors
17439 the current program is running on.
17440 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
17441 target_clone attribute.
17442 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
17443 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
17444 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
17445 (cpu_expand_builtin): Add support for target_clone attribute.
17446 (rs6000_valid_attribute_p): Allow "default" attribute.
17447 (get_decl_name): New debug function to simplify printing the
17448 current function name in debugging statements.
17449 (rs6000_clone_priority): New functions to support the target_clone
17450 attribute, and be able to generate code to switch between ISA 2.05
17451 through ISA 3.0 (power6 through power9).
17452 (rs6000_compare_version_priority): Likewise.
17453 (rs6000_get_function_versions_dispatcher): Likewise.
17454 (make_resolver_func): Likewise.
17455 (add_condition_to_bb): Likewise.
17456 (dispatch_function_versions): Likewise.
17457 (rs6000_generate_version_dispatcher_body): Likewise.
17458 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
17459 (fusion_gpr_load_p): Fix a spacing issue.
17460 * doc/extend.texi (Common Function Attributes): Document that the
17461 PowerPC supports the target_clone attribute.
17463 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17465 * config/arm/arm.h: explain F symbol found in description of ARM
17466 register allocation in its legend.
17468 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17470 * config/mips/frame-header-opt.c: Include profile-count.h.
17471 * config/riscv/riscv.c: Include profile-count.h
17473 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17475 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
17477 (sm_set_flag_if_changed): Add bbs field.
17478 (execute_sm_if_changed_flag_set): Pass BBS.
17479 (execute_sm): Update.
17481 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17483 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
17486 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17488 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
17489 (peephole2): New peephole2 to emit the above.
17490 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
17492 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17494 * config/aarch64/aarch64.c (define_peephole2 above
17495 *sub_<shift>_<mode>): New peephole.
17497 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17499 * config/i386/i386.c (make_resolver_func): Update.
17500 * Makefile.in: Add profile-count.h and profile-count.o
17501 * auto-profile.c (afdo_indirect_call): Update to new API.
17502 (afdo_set_bb_count): Update.
17503 (afdo_propagate_edge): Update.
17504 (afdo_propagate_circuit): Update.
17505 (afdo_calculate_branch_prob): Update.
17506 (afdo_annotate_cfg): Update.
17507 * basic-block.h: Include profile-count.h
17508 (struct edge_def): Turn count to profile_count.
17509 (struct basic_block_def): Likewie.
17510 (REG_BR_PROB_BASE): Move to profile-count.h
17511 (RDIV): Move to profile-count.h
17512 * bb-reorder.c (max_entry_count): Turn to profile_count.
17513 (find_traces): Update.
17514 (rotate_loop):Update.
17515 (connect_traces):Update.
17516 (sanitize_hot_paths):Update.
17517 * bt-load.c (migrate_btr_defs): Update.
17518 * cfg.c (RDIV): Remove.
17519 (init_flow): Use alloc_block.
17520 (alloc_block): Uninitialize count.
17521 (unchecked_make_edge): Uninitialize count.
17522 (check_bb_profile): Update.
17523 (dump_edge_info): Update.
17524 (dump_bb_info): Update.
17525 (update_bb_profile_for_threading): Update.
17526 (scale_bbs_frequencies_int): Update.
17527 (scale_bbs_frequencies_gcov_type): Update.
17528 (scale_bbs_frequencies_profile_count): New.
17529 * cfg.h (update_bb_profile_for_threading): Update.
17530 (scale_bbs_frequencies_profile_count): Declare.
17531 * cfgbuild.c (compute_outgoing_frequencies): Update.
17532 (find_many_sub_basic_blocks): Update.
17533 * cfgcleanup.c (try_forward_edges): Update.
17534 (try_crossjump_to_edge): Update.
17535 * cfgexpand.c (expand_gimple_tailcall): Update.
17536 (construct_exit_block): Update.
17537 * cfghooks.c (verify_flow_info): Update.
17538 (dump_bb_for_graph): Update.
17539 (split_edge): Update.
17540 (make_forwarder_block): Update.
17541 (duplicate_block): Update.
17542 (account_profile_record): Update.
17543 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
17544 (get_estimated_loop_iterations): Update.
17545 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
17546 (single_likely_exit): Update.
17547 * cfgloopmanip.c (scale_loop_profile): Update.
17549 (set_zero_probability): Update.
17550 (lv_adjust_loop_entry_edge): Update.
17551 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17552 (purge_dead_edges): Update.
17553 (rtl_account_profile_record): Update.
17554 * cgraph.c (cgraph_node::create): Uninitialize count.
17555 (symbol_table::create_edge): Uninitialize count.
17556 (cgraph_update_edges_for_call_stmt_node): Update.
17557 (cgraph_edge::dump_edge_flags): Update.
17558 (cgraph_node::dump): Update.
17559 (cgraph_edge::maybe_hot_p): Update.
17560 * cgraph.h: Include profile-count.h
17561 (create_clone), create_edge, create_indirect_edge): Update.
17562 (cgraph_node): Turn count to profile_count.
17563 (cgraph_edge0: Likewise.
17564 (make_speculative, clone): Update.
17565 (create_edge): Update.
17566 (init_lowered_empty_function): Update.
17567 * cgraphclones.c (cgraph_edge::clone): Update.
17568 (duplicate_thunk_for_node): Update.
17569 (cgraph_node::create_clone): Update.
17570 * cgraphunit.c (cgraph_node::analyze): Update.
17571 (cgraph_node::expand_thunk): Update.
17572 * final.c (dump_basic_block_info): Update.
17573 * gimple-streamer-in.c (input_bb): Update.
17574 * gimple-streamer-out.c (output_bb): Update.
17575 * graphite.c (print_global_statistics): Update.
17576 (print_graphite_scop_statistics): Update.
17577 * hsa-brig.c: Include basic-block.h.
17578 * hsa-dump.c: Include basic-block.h.
17579 * hsa-gen.c (T sum_slice): Update.
17580 (convert_switch_statements):Update.
17581 * hsa-regalloc.c: Include basic-block.h.
17582 * ipa-chkp.c (chkp_produce_thunks): Update.
17583 * ipa-cp.c (struct caller_statistics): Update.
17584 (init_caller_stats): Update.
17585 (gather_caller_stats): Update.
17586 (ipcp_cloning_candidate_p): Update.
17587 (good_cloning_opportunity_p): Update.
17588 (get_info_about_necessary_edges): Update.
17589 (dump_profile_updates): Update.
17590 (update_profiling_info): Update.
17591 (update_specialized_profile): Update.
17592 (perhaps_add_new_callers): Update.
17593 (decide_about_value): Update.
17594 (ipa_cp_c_finalize): Update.
17595 * ipa-devirt.c (struct odr_type_warn_count): Update.
17596 (struct decl_warn_count): Update.
17597 (struct final_warning_record): Update.
17598 (possible_polymorphic_call_targets): Update.
17599 (ipa_devirt): Update.
17600 * ipa-fnsummary.c (redirect_to_unreachable): Update.
17601 * ipa-icf.c (sem_function::merge): Update.
17602 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
17603 * ipa-inline.c (compute_uninlined_call_time): Update.
17604 (compute_inlined_call_time): Update.
17605 (want_inline_small_function_p): Update.
17606 (want_inline_self_recursive_call_p): Update.
17607 (edge_badness): Update.
17608 (lookup_recursive_calls): Update.
17609 (recursive_inlining): Update.
17610 (inline_small_functions): Update.
17611 (dump_overall_stats): Update.
17612 (dump_inline_stats): Update.
17613 * ipa-profile.c (ipa_profile_generate_summary): Update.
17614 (ipa_propagate_frequency): Update.
17615 (ipa_profile): Update.
17616 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17617 * ipa-utils.c (ipa_merge_profiles): Update.
17618 * loop-doloop.c (doloop_modify): Update.
17619 * loop-unroll.c (report_unroll): Update.
17620 (unroll_loop_runtime_iterations): Update.
17621 * lto-cgraph.c (lto_output_edge): Update.
17622 (lto_output_node): Update.
17623 (input_node): Update.
17624 (input_edge): Update.
17625 (merge_profile_summaries): Update.
17626 * lto-streamer-in.c (input_cfg): Update.
17627 * lto-streamer-out.c (output_cfg): Update.
17628 * mcf.c (create_fixup_graph): Update.
17629 (adjust_cfg_counts): Update.
17630 (sum_edge_counts): Update.
17631 * modulo-sched.c (sms_schedule): Update.
17632 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
17633 * predict.c (maybe_hot_count_p): Update.
17634 (probably_never_executed): Update.
17635 (dump_prediction): Update.
17636 (combine_predictions_for_bb): Update.
17637 (propagate_freq): Update.
17638 (handle_missing_profiles): Update.
17639 (counts_to_freqs): Update.
17640 (rebuild_frequencies): Update.
17641 (force_edge_cold): Update.
17642 * predict.h: Include profile-count.h
17643 (maybe_hot_count_p, counts_to_freqs): UPdate.
17644 * print-rtl-function.c: Do not include cfg.h
17645 * print-rtl.c: Include basic-block.h
17646 * profile-count.c: New file.
17647 * profile-count.h: New file.
17648 * profile.c (is_edge_inconsistent): Update.
17649 (correct_negative_edge_counts): Update.
17650 (is_inconsistent): Update.
17651 (set_bb_counts): Update.
17652 (read_profile_edge_counts): Update.
17653 (compute_frequency_overlap): Update.
17654 (compute_branch_probabilities): Update; Initialize and deinitialize
17656 (branch_prob): Update.
17657 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
17658 (edge_gcov_count): New.
17659 (bb_gcov_count): New.
17660 * shrink-wrap.c (try_shrink_wrapping): Update.
17661 * tracer.c (better_p): Update.
17662 * trans-mem.c (expand_transaction): Update.
17663 (ipa_tm_insert_irr_call): Update.
17664 (ipa_tm_insert_gettmclone_call): Update.
17665 * tree-call-cdce.c: Update.
17666 * tree-cfg.c (gimple_duplicate_sese_region): Update.
17667 (gimple_duplicate_sese_tail): Update.
17668 (gimple_account_profile_record): Update.
17669 (execute_fixup_cfg): Update.
17670 * tree-inline.c (copy_bb): Update.
17671 (copy_edges_for_bb): Update.
17672 (initialize_cfun): Update.
17673 (freqs_to_counts): Update.
17674 (copy_cfg_body): Update.
17675 (expand_call_inline): Update.
17676 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17677 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
17678 (try_unroll_loop_completely): Update.
17679 (try_peel_loop): Update.
17680 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17681 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
17682 * tree-ssa-loop-split.c (connect_loops): Update.
17683 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
17684 * tree-ssa-reassoc.c (branch_fixup): Update.
17685 * tree-ssa-tail-merge.c (replace_block_by): Update.
17686 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
17687 (compute_path_counts): Update.
17688 (update_profile): Update.
17689 (recompute_probabilities): Update.
17690 (update_joiner_offpath_counts): Update.
17691 (estimated_freqs_path): Update.
17692 (freqs_to_counts_path): Update.
17693 (clear_counts_path): Update.
17694 (ssa_fix_duplicate_block_edges): Update.
17695 (duplicate_thread_path): Update.
17696 * tree-switch-conversion.c (case_bit_test_cmp): Update.
17697 (struct switch_conv_info): Update.
17698 * tree-tailcall.c (decrease_profile): Update.
17699 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17700 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17701 * value-prof.c (check_counter): Update.
17702 (gimple_divmod_fixed_value): Update.
17703 (gimple_mod_pow2): Update.
17704 (gimple_mod_subtract): Update.
17705 (gimple_ic_transform): Update.
17706 (gimple_stringop_fixed_value): Update.
17707 * value-prof.h (gimple_ic): Update.
17709 2017-06-02 Carl Love <cel@us.ibm.com>
17711 * config/rs6000/rs6000-c: Add support for built-in functions
17712 vector double vec_doublee (vector signed int);
17713 vector double vec_doublee (vector unsigned int);
17714 vector double vec_doublee (vector float);
17715 vector double vec_doubleh (vector signed int);
17716 vector double vec_doubleh (vector unsigned int);
17717 vector double vec_doubleh (vector float);
17718 vector double vec_doublel (vector signed int);
17719 vector double vec_doublel (vector unsigned int);
17720 vector double vec_doublel (vector float);
17721 vector double vec_doubleo (vector signed int);
17722 vector double vec_doubleo (vector unsigned int);
17723 vector double vec_doubleo (vector float);.
17724 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
17725 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
17727 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
17728 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
17729 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
17731 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
17732 vec_doublel, vec_doubleh.
17733 * doc/extend.texi: Update the built-in documentation file for the
17734 new built-in functions.
17736 2017-06-02 David Malcolm <dmalcolm@redhat.com>
17739 * ipa-inline-analysis.c (free_growth_caches): Set
17740 edge_removal_hook_holder to NULL after removing it.
17742 2017-06-02 Sudakshina Das <sudi.das@arm.com>
17744 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
17745 comparision with zero.
17747 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17748 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17749 for early expansion of vec_min and vec_max builtins.
17750 (builtin_function_type): Add min/max unsigned variants to those
17751 identified as having unsigned arguments.
17753 2017-06-02 Olivier Hainque <hainque@adacore.com>
17755 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
17757 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17759 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
17760 Use VALL_F16 iterator rather than VALL.
17762 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17764 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
17765 Emit CBNZ inside loop when doing a strong exchange and comparing
17766 against zero. Generate the CC flags after the loop.
17768 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
17770 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
17771 (dl_section_ref): New.
17772 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
17773 On AIX, append an expression to subtract the size of the
17774 section length to dl_section_ref.
17776 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17778 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17779 for early expansion of vector absolute builtins.
17781 2017-06-02 Richard Biener <rguenther@suse.de>
17783 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
17784 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
17786 2017-06-02 Richard Biener <rguenther@suse.de>
17788 PR tree-optimization/80948
17789 * tree-tailcall.c (find_tail_calls): Track stmts to move in
17790 stmt order as well.
17792 2017-06-02 Richard Biener <rguenther@suse.de>
17794 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
17796 * tree-vect-stmts.c (process_use): Do not mark backedge defs
17797 for inductions as relevant.
17799 2017-06-02 Richard Biener <rguenther@suse.de>
17801 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
17802 (vectorizable_induction): ... this. Remove dead code.
17804 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
17806 * builtins. (expand_builtin_alloca): Remove second parameter and
17807 infer its value from the first parameter instead.
17808 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
17810 2017-06-02 Jakub Jelinek <jakub@redhat.com>
17812 PR rtl-optimization/80903
17813 * loop-doloop.c (add_test): Unshare sequence.
17815 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
17817 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
17819 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
17821 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
17823 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
17824 xlogue_layout::get_instance, logue_layout::xlogue_layout,
17825 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
17826 (xlogue_layout::get_stub_rtx): Make static.
17827 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
17828 (xlogue_layout::compute_stub_managed_regs): Rename to...
17829 (xlogue_layout::count_stub_managed_regs): ...this.
17830 (xlogue_layout::is_stub_managed_reg): New function.
17831 (xlogue_layout::m_stub_names): Rename to...
17832 (xlogue_layout::s_stub_names): ...this, make static.
17833 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
17834 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
17835 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
17836 xlogue_layout::s_stub_names): Instantiate statics.
17837 (stub_managed_regs): Remove.
17838 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
17839 (disable_call_ms2sysv_xlogues): Rename to...
17840 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
17841 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
17843 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
17844 change after reload_completed.
17845 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
17847 (ix86_expand_prologue): Likewise.
17848 (ix86_expand_epilogue): Likewise.
17849 (ix86_expand_split_stack_prologue): Likewise.
17850 (ix86_compute_frame_layout): Remove frame parameter ...
17851 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
17852 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
17854 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
17855 (ix86_frame): Move from here ...
17856 * config/i386/i386.h (ix86_frame): ... to here.
17857 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
17858 complete ix86_frame data structure instead. Remove some_ld_name.
17860 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17862 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
17863 symbols that hold a DECL_VALUE_EXPR.
17865 2017-06-01 Martin Jambor <mjambor@suse.cz>
17867 PR tree-optimization/80898
17868 * tree-sra.c (process_subtree_disqualification): Removed.
17869 (disqualify_candidate): Do not acll
17870 process_subtree_disqualification.
17871 (subtree_mark_written_and_enqueue): New function.
17872 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
17873 RHS has been disqualified and re-queue LHS if necessary. Apart
17874 from that, ignore disqualified RHS.
17876 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17878 * config/s390/s390.c (s390_emit_epilogue): Disable early return
17879 address fetch for z10 or later.
17881 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17883 * config/arc/arc.md (tst_movb): Add guard when splitting.
17885 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17887 * config/arc/arc.c (arc_can_eliminate): Test against
17888 arc_frame_pointer_needed.
17890 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17892 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
17893 to prevent store reordering.
17894 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
17895 (type): Add block type.
17896 (stack_tie): Define special instruction to be used in
17899 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17901 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
17902 constraint. It is not valid for the pattern.
17903 (noncommutative_binary_comparison): Likewise.
17905 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17907 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
17910 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17912 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
17913 be used by the reg-alloc.
17915 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17917 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
17918 reg-alloc when having mul64 or mul32x16 instructions.
17919 (mulsidi3): Likewise.
17920 (umulsidi3): Likewise.
17921 (mulsi32x16): New pattern.
17922 (mulsi64): Likewise.
17923 (mulsidi64): Likewise.
17924 (umulsidi64): Likewise.
17925 (MUL32x16_REG): Define.
17926 (mul64_600): Use MUL32x16_REG.
17927 (mac64_600): Likewise.
17928 (umul64_600): Likewise.
17929 (umac64_600): Likewise.
17931 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17933 * config/arc/arc.md (mulsi3_700): Make it commutative.
17935 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
17937 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
17939 (*sign_extendsidi2_insn): Likewise for movstosw.
17941 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17943 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
17944 the type of the input discriminant value. Convert the
17945 discriminant value of signedness vary.
17947 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
17949 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
17950 Add to -Wall section.
17952 2017-06-01 Richard Biener <rguenther@suse.de>
17954 PR middle-end/66313
17955 * fold-const.c (fold_plusminus_mult_expr): If the factored
17956 factor may be zero use a wrapping type for the inner operation.
17957 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
17958 and handle moved defs.
17959 (process_assignment): Properly guard the unary op case. Return a
17960 tri-state indicating that moving the stmt before the call may allow
17961 to continue. Pass through to_move.
17962 (find_tail_calls): Handle moving unrelated defs before
17965 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
17968 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
17969 splitter result in the canonical way.
17971 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
17973 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
17974 also for 32bit target. Update insn attributes.
17975 (zero-extendsidi2 splitter): Allow all registers for operand 1.
17977 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
17979 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
17980 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
17981 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
17982 (_mm_maskz_min_ss): New intrinsics.
17984 2017-05-31 Martin Liska <mliska@suse.cz>
17986 * tree-vect-loop.c (vect_create_epilog_for_reduction):
17987 Change comment style to one we normally use.
17988 (vectorizable_reduction): Likewise.
17989 (vectorizable_induction): Likewise.
17990 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17991 (vectorizable_call): Likewise.
17992 (vectorizable_simd_clone_call): Likewise.
17993 (vectorizable_conversion): Likewise.
17994 (vectorizable_assignment): Likewise.
17995 (vectorizable_shift): Likewise.
17996 (vectorizable_operation): Likewise.
17997 (vectorizable_store): Likewise.
17998 (vectorizable_load): Likewise.
17999 * tree-vectorizer.h: Likewise.
18001 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
18003 * passes.c (emergency_dump_function): New.
18004 * tree-pass.h (emergency_dump_function): Declare.
18005 * plugin.c (plugins_internal_error_function): Remove.
18006 * plugin.h (plugins_internal_error_function): Remove declaration.
18007 * toplev.c (internal_error_function): New static function. Use it...
18008 (general_init): ...here.
18010 2017-05-31 Graham Markall <graham.markall@embecosm.com>
18012 * config/arc/arc.c (arc_print_operand): Handle constant operands.
18013 (arc_rtx_costs): Add costs for new patterns.
18014 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
18015 * config/arc/predicates.md: Add _1_2_3_operand predicate.
18017 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18019 * tree-ssa-strlen.c (get_next_strinfo): New function.
18020 (get_stridx_plus_constant): Use it.
18021 (zero_length_string): Likewise.
18022 (adjust_related_strinfos): Likewise.
18023 (adjust_last_stmt): Likewise.
18025 2017-05-31 Richard Biener <rguenther@suse.de>
18028 * config/i386/i386.c (ix86_expand_builtin): Remove assert
18029 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
18031 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18033 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
18034 loop_vinfo argument and use of dependence distance vectors.
18035 Check instead whether the two references differ only in their
18036 initial value and assume that they have the same alignment if the
18037 difference is a multiple of the vector alignment.
18038 (vect_analyze_data_refs_alignment): Update call accordingly.
18040 2017-05-31 Martin Liska <mliska@suse.cz>
18043 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
18045 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18047 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
18048 (create_intersect_range_checks): Move from ...
18049 * tree-data-ref.c (create_intersect_range_checks_index)
18050 (create_intersect_range_checks): ... to here.
18051 (create_runtime_alias_checks): New function factored from ...
18052 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
18053 here. Call above function.
18054 * tree-data-ref.h (create_runtime_alias_checks): New function.
18056 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18058 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
18059 segment length for dr_b and compute it in wide_int.
18061 2017-05-31 Richard Biener <rguenther@suse.de>
18063 PR tree-optimization/80906
18064 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
18065 and pass through iv_map.
18066 (copy_bb_and_scalar_dependences): Adjust.
18067 (translate_pending_phi_nodes): Likewise.
18068 (copy_loop_close_phi_args): Handle code-generating IVs instead
18071 2017-05-30 David Malcolm <dmalcolm@redhat.com>
18073 * diagnostic-color.c (color_dict): Add "type-diff".
18074 (parse_gcc_colors): Update comment.
18075 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
18076 -fdiagnostics-show-template-tree and -fno-elide-type.
18077 (GCC_COLORS): Add type-diff to example.
18079 (-fdiagnostics-show-template-tree): New.
18080 (-fno-elide-type): New.
18081 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
18082 the pp_format_decoder callback. Call any m_format_postprocessor's
18084 (pretty_printer::pretty_printer): Initialize
18085 m_format_postprocessor.
18086 (pretty_printer::~pretty_printer): Delete any
18087 m_format_postprocessor.
18088 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
18089 (class format_postprocessor): New class.
18090 (struct pretty_printer::format_decoder): Document the new parameters.
18091 (struct pretty_printer::m_format_postprocessor): New field.
18092 * tree-diagnostic.c (default_tree_printer): Update for new
18093 bool and const char ** params.
18094 * tree-diagnostic.h (default_tree_printer): Likewise.
18096 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
18098 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
18099 (lwa_operand): Delete rs6000_gen_cell_microcode test.
18100 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18101 rs6000_gen_cell_microcode code.
18102 (rs6000_final_prescan_insn): Delete.
18103 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
18104 "warn-cell-microcode" entries.
18105 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
18106 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
18107 throughout. Change cc_reg_not_micro_cr0_operand to
18108 cc_reg_not_cr0_operand throughout.
18109 (*extendhi<mode>2_noload): Delete.
18110 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
18111 (mwarn-cell-microcode): Delete.
18112 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
18113 -mgen-cell-microcode and -mwarn-cell-microcode.
18115 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
18118 * config/i386/constraints.md (Yd): New constraint.
18120 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
18121 and (?Yd, r) alternatives. Update insn attributes.
18122 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
18123 and (?*Yd, r) alternatives. Update insn attributes.
18124 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
18126 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
18128 * gimplify.c (gimplify_modify_expr): Don't create a
18129 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
18132 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
18134 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
18136 2017-05-30 Richard Biener <rguenther@suse.de>
18138 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
18139 and reduc_def fields.
18140 (STMT_VINFO_REDUC_TYPE): New define.
18141 (STMT_VINFO_REDUC_DEF): Likewise.
18142 (vect_force_simple_reduction): Adjust prototype.
18143 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
18144 (vect_is_simple_reduction): Remove check_reduction argument.
18145 (vect_force_simple_reduction): Adjust and set
18146 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18147 (vectorizable_reduction): Do not re-do reduction analysis
18148 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18149 * tree-parloops.c (gather_scalar_reductions): Adjust.
18151 2017-05-30 Richard Biener <rguenther@suse.de>
18153 PR middle-end/80901
18154 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
18157 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
18159 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
18160 Introduce unknown_misalignment parameter and remove vf.
18161 (vect_peeling_hash_get_lowest_cost):
18162 Pass unknown_misalignment parameter.
18163 (vect_enhance_data_refs_alignment):
18164 Fix unsupportable data ref treatment.
18166 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18168 * tree-vect-data-refs.c (vect_get_data_access_cost):
18169 Workaround for SLP handling.
18170 (vect_enhance_data_refs_alignment):
18171 Compute costs for doing no peeling at all, compare to the best
18172 peeling costs so far and avoid peeling if cheaper.
18174 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18176 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
18177 Return peeling info and set costs to zero for unlimited cost
18179 (vect_enhance_data_refs_alignment): Also inspect all datarefs
18180 with unknown misalignment. Compute and costs for unknown
18181 misalignment, compare them to the costs for known misalignment
18182 and choose the cheapest for peeling.
18184 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18186 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
18187 (vect_get_peeling_costs_all_drs): Create function.
18188 (vect_peeling_hash_get_lowest_cost):
18189 Use vect_get_peeling_costs_all_drs.
18190 (vect_peeling_supportable): Create function.
18191 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
18193 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18195 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
18196 DR_HAS_NEGATIVE_STEP.
18197 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
18198 (vect_enhance_data_refs_alignment): Use.
18199 (vect_duplicate_ssa_name_ptr_info): Use.
18200 * tree-vectorizer.h (dr_misalignment): Use.
18201 (known_alignment_for_access_p): Use.
18203 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18206 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
18208 (has_section_name): New function.
18210 2017-05-30 Martin Liska <mliska@suse.cz>
18213 * auto-profile.c (get_function_decl_from_block): Fix
18216 2017-05-30 Richard Biener <rguenther@suse.de>
18218 PR middle-end/80876
18219 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
18221 2017-05-30 Martin Liska <mliska@suse.cz>
18223 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
18224 * dumpfile.h (struct dump_file_info): Remove ctors.
18226 2017-05-30 Martin Liska <mliska@suse.cz>
18228 * predict.def: Fix GNU coding style.
18230 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18232 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18233 Mark 'to' argument with ATTRIBUTE_UNUSED.
18235 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18237 * config/xtensa/xtensa.c (xtensa_emit_call): Use
18238 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
18239 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
18242 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
18244 * doc/install.texi (Options specification): Restore entry of
18245 --enable-sjlj-exceptions.
18247 2017-05-27 Michael Eager <eager@eagercon.com>
18250 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
18252 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
18254 * config/microblaze/microblaze.h
18255 (FIXED_REGISTERS): Update in macro.
18256 (CALL_USED_REGISTERS): Update in macro.
18258 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
18260 * doc/install.texi: Add links to macOS binary distributions.
18262 2017-05-27 Jakub Jelinek <jakub@redhat.com>
18266 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18268 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18270 2017-05-26 Martin Liska <mliska@suse.cz>
18272 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
18274 2017-05-26 Martin Liska <mliska@suse.cz>
18276 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
18277 always leading ';; '.
18278 (dump_bb_info): Likewise.
18279 (brief_dump_cfg): Likewise.
18280 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
18281 * dumpfile.c: Remove usage of TDF_VERBOSE.
18282 * dumpfile.h (enum dump_kind): Likewise.
18283 (dump_gimple_bb_header): Do not use TDF_COMMENT.
18284 * print-tree.c (debug_verbose): Remove.
18285 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
18286 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
18287 * tree-diagnostic.c (default_tree_printer): Replace
18288 TDF_DIAGNOSTIC with TDF_SLIM.
18290 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18292 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
18293 in parameter loop, rather than loop_vinfo.
18294 (create_intersect_range_checks): Ditto.
18295 (vect_create_cond_for_alias_checks): Update call to above functions.
18297 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18299 PR tree-optimization/80815
18300 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
18301 for merging runtime alias checks. Handle negative DR_STEPs.
18303 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18305 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
18307 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
18308 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
18309 out code pruning runtime alias checks.
18310 * tree-data-ref.c (prune_runtime_alias_test_list): New function
18311 factored out from above.
18312 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18314 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18316 (prune_runtime_alias_test_list): New decalaration.
18318 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18320 * tree-vect-data-refs.c (compare_tree): Rename and move ...
18321 * tree-data-ref.c (data_ref_compare_tree): ... to here.
18322 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
18323 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
18324 (operator==, comp_dr_with_seg_len_pair): Ditto.
18325 (vect_prune_runtime_alias_test_list): Ditto.
18327 2017-05-26 Martin Liska <mliska@suse.cz>
18330 * params.def: Bound partial-inlining-entry-probability param.
18332 2017-05-26 Marek Polacek <polacek@redhat.com>
18335 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
18338 2017-05-26 Richard Biener <rguenther@suse.de>
18340 PR tree-optimization/80842
18341 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
18344 2017-05-26 Richard Biener <rguenther@suse.de>
18346 PR tree-optimization/80844
18347 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
18349 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
18351 * doc/md.texi (Machine Constraints): Update x86 family
18352 machine constraints section to match 'config/i386/constraints.md'.
18354 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
18356 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
18358 2017-05-25 Nathan Sidwell <nathan@acm.org>
18360 * doc/invoke.texi (--enable-languages): Update documentation.
18362 2017-05-25 Martin Liska <mliska@suse.cz>
18364 * dumpfile.c: Add TDF_FOLDING.
18365 * dumpfile.h (enum dump_kind): Likewise.
18366 * genmatch.c (dt_simplify::gen_1): Use it.
18368 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18370 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
18372 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18374 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18375 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
18377 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18379 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
18380 * match.pd (X == C): Rewrite it here.
18381 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
18382 with_certain_nonzero_bits2): New predicates.
18383 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
18385 2017-05-24 Nathan Sidwell <nathan@acm.org>
18387 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
18390 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
18393 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18395 * config/powerpcspe: New port. Files are copied from the rs6000
18396 port, with "rs6000" in filenames replaced by "powerpcspe".
18398 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
18400 PR rtl-optimization/80754
18401 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
18403 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
18405 * config/sparc/sparc.md (length): Return the correct value for -mflat
18406 sibcalls to match output_sibcall.
18408 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18412 * config/rs6000/rs6000.c (struct machine_function): Add new field
18414 (rs6000_get_separate_components): Init that field, use it.
18415 (rs6000_components_for_bb): Use the field.
18417 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
18419 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
18421 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
18423 PR middle-end/80823
18424 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
18426 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18429 * config/s390/s390.c (s390_check_qrst_address): Check incoming
18430 address against address_operand predicate.
18431 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
18433 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
18435 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
18436 parameters passed indirectly.
18438 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
18440 * config/i386/i386.md (*movdi_internal): Remove SSE4
18441 alternative 18 (?r, *v). Update insn attributes.
18442 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
18443 Update insn attributes.
18444 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
18445 Update insn attributes.
18446 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
18447 alternative 1 (r, v). Remove isa attribute.
18448 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
18449 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
18450 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18452 2017-05-23 Tom de Vries <tom@codesourcery.com>
18454 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
18457 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18459 * cgraphunit.c (symbol_table::process_new_functions): Update.
18460 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
18461 (inline_generate_summary): Rename to ...
18462 (ipa_fn_summary_generate): ... this one.
18463 (inline_read_summary): Rename to ...
18464 (ipa_fn_summary_read): ... this one.
18465 (inline_write_summary): Rename to ...
18466 (ipa_fn_summary_write): ... this one.
18467 (inline_free_summary): Rename to ...
18468 (ipa_free_fn_summary): ... this one.
18469 (pass_data_local_fn_summary, pass_local_fn_summary,
18470 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
18471 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
18472 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
18473 make_pass_ipa_fn_summary): New.
18474 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
18475 inline_write_summary, inline_free_summary): Remove.
18476 (ipa_free_fn_summary) : New.
18477 * ipa-inline.c (ipa_inline): Update.
18478 (pass_ipa_inline): Do not generate summaries.
18479 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
18481 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
18482 and add pass_ipa_fn_summary.
18483 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
18485 (make_pass_inline_parameters): Remove.
18487 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
18489 * omp-low.c (struct omp_context): Remove "default_kind" member.
18492 * omp-offload.c (execute_oacc_device_lower): Remove the
18493 parallelism dimensions function attributes for unparallelized
18494 OpenACC kernels constructs.
18496 2017-05-23 Martin Liska <mliska@suse.cz>
18498 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
18500 (cgraph_edge::make_speculative): Likewise.
18501 (cgraph_edge::resolve_speculation): Likewise.
18502 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
18503 (cgraph_node::dump): Likewise.
18504 * cgraph.h: Likewise.
18505 * cgraphunit.c (analyze_functions): Likewise.
18506 (symbol_table::compile): Likewise.
18507 * ipa-cp.c (print_all_lattices): Likewise.
18508 (determine_versionability): Likewise.
18509 (initialize_node_lattices): Likewise.
18510 (ipcp_verify_propagated_values): Likewise.
18511 (estimate_local_effects): Likewise.
18512 (update_profiling_info): Likewise.
18513 (create_specialized_node): Likewise.
18514 (perhaps_add_new_callers): Likewise.
18515 (decide_about_value): Likewise.
18516 (decide_whether_version_node): Likewise.
18517 (identify_dead_nodes): Likewise.
18518 (ipcp_store_bits_results): Likewise.
18519 * ipa-devirt.c (dump_targets): Likewise.
18520 (ipa_devirt): Likewise.
18521 * ipa-icf.c (sem_item::dump): Likewise.
18522 (sem_function::equals): Likewise.
18523 (sem_variable::equals): Likewise.
18524 (sem_item_optimizer::read_section): Likewise.
18525 (sem_item_optimizer::execute): Likewise.
18526 (congruence_class::dump): Likewise.
18527 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
18528 (dump_inline_summary): Likewise.
18529 (estimate_node_size_and_time): Likewise.
18530 (inline_analyze_function): Likewise.
18531 * ipa-inline-transform.c (inline_call): Likewise.
18532 * ipa-inline.c (report_inline_failed_reason): Likewise.
18533 (want_early_inline_function_p): Likewise.
18534 (edge_badness): Likewise.
18535 (update_edge_key): Likewise.
18536 (inline_small_functions): Likewise.
18537 * ipa-profile.c (ipa_profile): Likewise.
18538 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
18539 (ipa_make_edge_direct_to_target): Likewise.
18540 (remove_described_reference): Likewise.
18541 (ipa_impossible_devirt_target): Likewise.
18542 (propagate_controlled_uses): Likewise.
18543 (ipa_print_node_params): Likewise.
18544 (ipcp_transform_function): Likewise.
18545 * ipa-pure-const.c (pure_const_read_summary): Likewise.
18546 (propagate_pure_const): Likewise.
18547 * ipa-reference.c (generate_summary): Likewise.
18548 (read_write_all_from_decl): Likewise.
18549 (propagate): Likewise.
18550 (ipa_reference_read_optimization_summary): Likewise.
18551 * ipa-utils.c (ipa_merge_profiles): Likewise.
18552 * ipa.c (walk_polymorphic_call_targets): Likewise.
18553 (symbol_table::remove_unreachable_nodes): Likewise.
18554 (ipa_single_use): Likewise.
18555 * passes.c (execute_todo): Likewise.
18556 * predict.c (drop_profile): Likewise.
18557 * symtab.c (symtab_node::get_dump_name): New function.
18558 (symtab_node::dump_name): Likewise.
18559 (symtab_node::dump_asm_name): Likewise.
18560 (symtab_node::dump_references): Likewise.
18561 (symtab_node::dump_referring): Likewise.
18562 (symtab_node::dump_base): Likewise.
18563 (symtab_node::debug_symtab): Likewise.
18564 * tree-sra.c (convert_callers_for_node): Likewise.
18565 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
18566 * value-prof.c (init_node_map): Likewise.
18568 2017-05-23 Martin Liska <mliska@suse.cz>
18570 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
18571 and symtab_node::debug_symtab to symbol_table::debug.
18572 * cgraphunit.c (analyze_functions): Use the renamed function.
18573 (symbol_table::compile): Likewise.
18574 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
18575 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
18576 * passes.c (execute_todo): Likewise.
18577 * symtab.c (symbol_table::dump): New function.
18578 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
18580 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18582 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
18583 that nonconst implies exec.
18585 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18587 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18588 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18589 (inline_edge_summary_vec): Turn into ...
18590 (ipa_call_summaries): ... this one.
18591 (redirect_to_unreachable, edge_set_predicate,
18592 evaluate_properties_for_edge, inline_summary_alloc,
18593 reset_ipa_call_summary, reset_inline_summary,
18594 inline_summary_t::duplicate): Update.
18595 (inline_edge_duplication_hook): Turn to ...
18596 (ipa_call_summary_t::duplicate): ... this one.
18597 (inline_edge_removal_hook): Turn to ...
18598 (ipa_call_summary_t::remove): ... this one.
18599 (dump_inline_edge_summary): Turn to ...
18600 (dump_ipa_call_summary): ... this one.
18601 (estimate_function_body_sizes): Update.
18602 (inline_update_callee_summaries): Update.
18603 (remap_edge_change_prob): Update.
18604 (remap_edge_summaries): Update.
18605 (inline_merge_summary): Update.
18606 (do_estimate_edge_time): Update.
18607 (inline_generate_summary): Update.
18608 (inline_read_section): Update.
18609 (inline_read_summary): Update.
18610 (inline_free_summary): Update.
18611 * ipa-inline.c (can_inline_edge_p): Update.
18612 (compute_inlined_call_time): Update.
18613 (want_inline_small_function_p): Update.
18614 (edge_badness): Update.
18615 (early_inliner): Update.
18616 * ipa-inline.h (inline_edge_summary): Turn to ...
18617 (ipa_call_summary): ... this one.
18618 (ipa_call_summary_t): New class.
18619 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18620 (ipa_call_summaries): New.
18621 (inline_edge_summary): Remove.
18622 (estimate_edge_growth): Update.
18623 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18624 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18625 * ipa-split.c (execute_split_functions): Update.
18626 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18628 2017-05-23 Tom de Vries <tom@codesourcery.com>
18630 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18631 attributes): Document rdrand effective target.
18633 2017-05-23 Tom de Vries <tom@codesourcery.com>
18635 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18636 attributes): Sort alphabetically.
18638 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
18640 * config/avr/genmultilib.awk: Use gsub instead of gensub.
18642 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
18645 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
18646 V2DF/V2DI splat into two separate patterns, one that handles
18647 registers, and the other that only handles memory. Drop support
18648 for splatting from a GPR on ISA 2.07 and then splitting the
18649 splat into direct move and splat.
18650 (vsx_splat_<mode>_reg): Likewise.
18651 (vsx_splat_<mode>_mem): Likewise.
18653 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
18655 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
18657 2017-05-22 Jakub Jelinek <jakub@redhat.com>
18659 PR middle-end/80809
18660 * omp-low.c (finish_taskreg_remap): New function.
18661 (finish_taskreg_scan): If unit size of ctx->record_type
18662 is non-constant, unshare the size expression and replace
18663 decls in it with possible outer var refs.
18665 PR middle-end/80809
18666 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
18667 GOVD_SHARED rather than GOVD_PRIVATE with it.
18668 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
18669 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
18671 PR middle-end/80853
18672 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
18673 as last argument to build_outer_var_ref for pointer bases of array
18674 section reductions.
18676 2017-05-19 Martin Sebor <msebor@redhat.com>
18678 * print-tree.c (print_node): Print DECL_READ_P flag.
18680 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18682 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
18683 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
18684 * cgraph.c: Likewise.
18685 * cgraphunit.c: Likewise.
18686 * gengtype.c: Likewise.
18687 * ipa-cp.c: Likewise.
18688 * ipa-devirt.c: Likewise.
18689 * ipa-icf.c: Likewise.
18690 * ipa-predicate.c: Likewise.
18691 * ipa-profile.c: Likewise.
18692 * ipa-prop.c: Likewise.
18693 * ipa-split.c: Likewise.
18695 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
18696 edge_predicate_pool, dump_inline_hints,
18697 inline_summary::account_size_time, redirect_to_unreachable,
18698 edge_set_predicate, set_hint_predicate,
18699 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
18700 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
18701 inline_summary_t::remove, remap_hint_predicate_after_duplication,
18702 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
18703 ipa_call_summary_t::remove, initialize_growth_caches,
18704 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
18705 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
18706 mark_modified, unmodified_parm_1, unmodified_parm,
18707 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
18708 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
18709 compute_bb_predicates, will_be_nonconstant_expr_predicate,
18710 will_be_nonconstant_predicate, record_modified_bb_info,
18711 get_minimal_bb, record_modified, param_change_prob,
18712 phi_result_unknown_predicate, predicate_for_phi_result,
18713 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
18714 estimate_function_body_sizes, compute_inline_parameters,
18715 compute_inline_parameters_for_curren, pass_data_inline_parameters,
18716 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
18717 inline_update_callee_summaries, remap_edge_change_prob,
18718 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
18719 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
18720 inline_analyze_function, inline_summary_t::insert,
18721 inline_generate_summary, read_ipa_call_summary, inline_read_section,
18722 inline_read_summary, write_ipa_call_summary, inline_write_summary,
18723 inline_free_summary): Move to ipa-fnsummary.h
18724 (predicate_t): Remove.
18725 * ipa-fnsummary.c: New file.
18726 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
18727 (enum inline_hints_vals, inline_hints, agg_position_info,
18728 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
18729 inline_summaries, ipa_call_summary, ipa_call_summary_t,
18730 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
18731 dump_inline_summary, dump_inline_hints, inline_generate_summary,
18732 inline_read_summary, inline_write_summary, inline_free_summary,
18733 inline_analyze_function, initialize_inline_failed,
18734 inline_merge_summary, inline_update_overall_summary,
18735 compute_inline_parameters): Move to ipa-fnsummary.h
18736 * ipa-fnsummary.h: New file.
18737 * ipa-inline-transform.h: Include ipa-inline.h.
18738 * ipa-inline.c: LIkewise.
18740 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18742 * ipa-inline.c (edge_badness): Use inlined_time instead of
18743 inline_summaries->get.
18745 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18747 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
18749 2017-05-22 Nathan Sidwell <nathan@acm.org>
18751 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
18752 (fdump-lang): Document 'raw' option.
18753 * dumpfile.h (TDI_tu): Delete.
18754 * dumpfile.c (dump_files): Remove translation-unit.
18755 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
18757 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
18759 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
18760 command option from $(AWK) call.
18761 * config/avr/genmultilib.awk: Simplify and rewrite so that it
18762 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
18763 [FORMAT]: Remove handling of variable.
18764 * config/avr/t-multilib: Regenerate.
18766 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18768 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
18770 (dump_inline_summary): Do not print self_time.
18771 (estimate_function_body_sizes): Do not set self_time.
18772 (compute_inline_parameters): Likewise.
18773 (inline_read_section, inline_write_summary): Do not stream self_time.
18774 * ipa-inline.h (inline_summary): Drop self_time.
18776 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18778 * ipa-inline-analysis.c (account_size_time): Rename to ...
18779 (inline_summary::account_size_time): ... this one.
18780 (reset_ipa_call_summary): Turn to ...
18781 (ipa_call_summary::reset): ... this one.
18782 (reset_inline_summary): Turn to ...
18783 (inline_summary::reset): ... this one.
18784 (inline_summary_t::remove): Update.
18785 (inline_summary_t::duplicate): Update.
18786 (ipa_call_summary_t::remove): Update.
18787 (dump_inline_summary): Update.
18788 (estimate_function_body_sizes): Update.
18789 (compute_inline_parameters): Update.
18790 (estimate_node_size_and_time): Update.
18791 (inline_merge_summary): Update.
18792 (inline_update_overall_summary): Update.
18793 (inline_read_section): Update.
18794 (inline_write_summary): Update.
18795 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
18796 add account_size_time and reset member functions.
18797 (ipa_call_summary): Add reset function.
18798 * ipa-predicate.h (predicate::operator &): Constify.
18800 2017-05-22 Richard Biener <rguenther@suse.de>
18802 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
18804 2017-05-19 Jason Merrill <jason@redhat.com>
18806 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
18808 2017-05-19 Marek Polacek <polacek@redhat.com>
18811 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
18812 TYPE_OVERFLOW_WRAPS checks.
18814 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
18816 * tree-core.h (enum omp_clause_default_kind): Add
18817 "OMP_CLAUSE_DEFAULT_PRESENT".
18818 * tree-pretty-print.c (dump_omp_clause): Handle it.
18819 * gimplify.c (enum gimplify_omp_var_data): Add
18820 "GOVD_MAP_FORCE_PRESENT".
18821 (gimplify_adjust_omp_clauses_1): Map it to
18822 "GOMP_MAP_FORCE_PRESENT".
18823 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
18825 * gimplify.c (oacc_default_clause): Clarify.
18827 2017-05-19 Nathan Sidwell <nathan@acm.org>
18829 LANG_HOOK_REGISTER_DUMPS
18830 * toplev.c (general_init): Call register dump lang hook.
18831 * doc/invoke.texi: Document -fdump-lang option family.
18832 * dumpfile.c (dump_files): Remove class dump here.
18833 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
18834 * dumpfile.h (tree_dump_index): Remove TDI_class.
18835 * langhooks-def.h (lhd_register_dumps): Declare.
18836 (LANG_HOOKS_REGISTER_DUMPS): Define.
18837 (LANG_HOOKS_INITIALIZER): Add it.
18838 * langhooks.c (lhd_register_dumps): Define.
18839 * langhooks.h (struct lang_hooks): Add register_dumps.
18841 2017-05-19 Nathan Sidwell <nathan@acm.org>
18843 * context.h (context::set_passes): New.
18844 * context.c (context::context): Do not create pass manager.
18845 * toplev.c (general_init): Create pass manager here.
18847 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
18849 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
18850 use this splitter if two add or or instructions would also work for
18851 the constant we want to generate.
18853 2017-05-19 Richard Biener <rguenther@suse.de>
18856 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
18857 predicate evaluation.
18859 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18861 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
18863 * ipa-inline.c (want_inline_small_function_p): Do not cast to
18866 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18868 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18869 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18870 (inline_edge_summary_vec): Turn into ...
18871 (ipa_call_summaries): ... this one.
18872 (redirect_to_unreachable, edge_set_predicate,
18873 evaluate_properties_for_edge, inline_summary_alloc,
18874 reset_ipa_call_summary, reset_inline_summary,
18875 inline_summary_t::duplicate): Update.
18876 (inline_edge_duplication_hook): Turn to ...
18877 (ipa_call_summary_t::duplicate): ... this one.
18878 (inline_edge_removal_hook): Turn to ...
18879 (ipa_call_summary_t::remove): ... this one.
18880 (dump_inline_edge_summary): Turn to ...
18881 (dump_ipa_call_summary): ... this one.
18882 (estimate_function_body_sizes): Update.
18883 (inline_update_callee_summaries): Update.
18884 (remap_edge_change_prob): Update.
18885 (remap_edge_summaries): Update.
18886 (inline_merge_summary): Update.
18887 (do_estimate_edge_time): Update.
18888 (inline_generate_summary): Update.
18889 (inline_read_section): Update.
18890 (inline_read_summary): Update.
18891 (inline_free_summary): Update.
18892 * ipa-inline.c (can_inline_edge_p): Update.
18893 (compute_inlined_call_time): Update.
18894 (want_inline_small_function_p): Update.
18895 (edge_badness): Update.
18896 (early_inliner): Update.
18897 * ipa-inline.h (inline_edge_summary): Turn to ...
18898 (ipa_call_summary): ... this one.
18899 (ipa_call_summary_t): New class.
18900 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18901 (ipa_call_summaries): New.
18902 (inline_edge_summary): Remove.
18903 (estimate_edge_growth): Update.
18904 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18905 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18906 * ipa-split.c (execute_split_functions): Update.
18907 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18909 2017-05-19 Richard Biener <rguenther@suse.de>
18911 PR middle-end/80764
18912 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
18914 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
18916 * config/rs6000/rs6000.c (struct machine_function): Add field
18917 fpr_is_wrapped_separately.
18918 (rs6000_get_separate_components): Use 64 components. Handle the
18919 new FPR components.
18920 (rs6000_components_for_bb): Handle the FPR components.
18921 (rs6000_emit_prologue_components): Handle the FPR components.
18922 (rs6000_emit_epilogue_components): Handle the FPR components.
18923 (rs6000_set_handled_components): Handle the FPR components.
18924 (rs6000_emit_prologue): Don't output prologue code for those FPRs
18925 that are already separately shrink-wrapped.
18926 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
18927 that are already separately shrink-wrapped.
18929 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18932 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
18935 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
18936 (define_peephole2 for Altivec d-form load): Add peepholes to catch
18937 cases where the register allocator uses a move and an offsettable
18938 memory operation to/from a FPR register on ISA 2.06/2.07.
18939 (define_peephole2 for Altivec d-form store): Likewise.
18941 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
18944 * config/i386/mmx.md (*mov<mode>_internal): Enable
18945 alternatives 11, 12, 13 and 14 also for 32bit targets.
18946 Remove alternatives 15, 16, 17 and 18.
18947 * config/i386/sse.md (vec_concatv2di): Change
18948 alternative (!x, *y) to (x, ?!*Yn).
18950 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
18952 * dumpfile.h (enum dump_kind): Remove stray comma.
18954 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18956 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
18957 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
18958 predicate::num_conditions
18959 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
18960 (CHANGED): turn into predicate::changed.
18961 (agg_position_info): Move to ipa-predicate.h
18962 (add_condition, predicate::add_clause, predicate::operator &=,
18963 predicate::or_with, predicate::evaluate, predicate::probability,
18964 dump_condition, dump_clause, predicate::dump,
18965 predicate::remap_after_duplication, predicate::remap_after_inlining,
18966 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
18967 (evaluate_conditions_for_known_args): Update.
18968 (set_cond_stmt_execution_predicate): Update.
18969 * ipa-inline.h: Include ipa-predicate.h
18970 (condition, inline_param_summary, conditions, agg_position_info,
18971 predicate): Move to ipa-predicate.h
18972 * ipa-predicate.c: New file.
18973 * ipa-predicate.h: New file.
18975 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
18977 * final.c (leaf_function_p): Check we are not in a sequence.
18979 2017-05-18 Martin Liska <mliska@suse.cz>
18981 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
18982 * dumpfile.c (dump_register): Use new enum dump_kind.
18983 (get_dump_file_name): Likewise.
18984 (dump_enable_all): Likewise.
18985 (dump_switch_p_1): Likewise.
18986 (enable_rtl_dump_file): Remove usage of TDF_RTL.
18987 * dumpfile.h (enum dump_kind): New enum type.
18988 (struct dump_file_info): Create constructor and
18989 format fields and comments.
18990 * passes.c (pass_manager::register_one_dump_file):
18992 * statistics.c (statistics_early_init): Likewise.
18993 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
18994 TDF_TREE with TDF_SLIM.
18995 (gather_memory_references_ref): Likewise.
18997 2017-05-18 Martin Liska <mliska@suse.cz>
18999 * vec.h (struct vnull): Use it.
19001 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19003 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
19004 (true_predicate, false_predicate, true_predicate_p,
19005 false_predicate_p): Remove.
19006 (single_cond_predicate, not_inlined_predicate): Turn to member function
19008 (add_condition): Update.
19009 (add_clause): Turn to...
19010 (predicate::add_clause): ... this one; update; allow passing NULL
19012 (and_predicates): Turn to ...
19013 (predicate::operator &=): ... this one.
19014 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
19015 (or_predicates): Turn to ...
19016 (predicate::or_with): ... this one.
19017 (evaluate_predicate): Turn to ...
19018 (predicate::evaluate): ... this one.
19019 (predicate_probability): Turn to ...
19020 (predicate::probability): ... this one.
19021 (dump_condition): Update.
19022 (dump_predicate): Turn to ...
19023 (predicate::dump): ... this one.
19024 (account_size_time): Update.
19025 (edge_set_predicate): Update.
19026 (set_hint_predicate): UPdate.
19027 (evaluate_conditions_for_known_args): Update.
19028 (evaluate_properties_for_edge): Update.
19029 (remap_predicate_after_duplication): Turn to...
19030 (predicate::remap_after_duplication): ... this one.
19031 (remap_hint_predicate_after_duplication): Update.
19032 (inline_summary_t::duplicate): UPdate.
19033 (dump_inline_edge_summary): Update.
19034 (dump_inline_summary): Update.
19035 (set_cond_stmt_execution_predicate): Update.
19036 (set_switch_stmt_execution_predicate): Update.
19037 (compute_bb_predicates): Update.
19038 (will_be_nonconstant_expr_predicate): Update.
19039 (will_be_nonconstant_predicate): Update.
19040 (phi_result_unknown_predicate): Update.
19041 (predicate_for_phi_result): Update.
19042 (array_index_predicate): Update.
19043 (estimate_function_body_sizes): Update.
19044 (estimate_node_size_and_time): Update.
19045 (estimate_ipcp_clone_size_and_time): Update.
19046 (remap_predicate): Rename to ...
19047 (predicate::remap_after_inlining): ... this one.
19048 (remap_hint_predicate): Update.
19049 (inline_merge_summary): Update.
19050 (inline_update_overall_summary): Update.
19051 (estimate_size_after_inlining): Update.
19052 (read_predicate): Rename to ...
19053 (predicate::stream_in): ... this one.
19054 (read_inline_edge_summary): Update.
19055 (write_predicate): Rename to ...
19056 (predicate::stream_out): ... this one.
19057 (write_inline_edge_summary): Update.
19058 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
19059 (clause_t): Turn to uint32_t
19060 (predicate): Turn to class; implement constructor and operators
19062 (size_time_entry): Update.
19063 (inline_summary): Update.
19064 (inline_edge_summary): Update.
19066 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
19068 * fold-const.c (fold_binary_loc): Move transformation...
19069 * match.pd (C - X CMP X): ... here.
19071 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
19073 * config/sparc/sparc.c (sparc_option_override): Set function
19074 alignment for -mcpu=niagara7 to 64 to match the I$ line.
19075 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
19077 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
19079 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
19081 2017-05-18 Marek Polacek <polacek@redhat.com>
19084 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
19085 (pass_ubsan::execute): Call gimple_assign_single_p instead of
19086 gimple_assign_load_p.
19088 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
19090 PR middle-end/80692
19091 * real.c (do_compare): Give decimal_do_compare preference over
19092 comparing just the signs.
19094 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
19096 * doc/md.texi (Canonicalization of Instructions): Describe the
19097 canonical form of instructions that inherently set a condition
19100 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
19102 PR middle-end/80775
19103 * tree-cfg.c: Move deletion of unreachable case statements to after
19104 the merging of consecutive case labels.
19106 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
19108 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
19109 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
19110 restoring of callee-saved registers.
19112 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
19114 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
19115 * config/visium/visium.c (single_set_and_flags): Likewise.
19116 * config/visium/visium.md (Substitutions): Likewise.
19118 2017-05-17 Martin Liska <mliska@suse.cz>
19120 * cfg.c: Introduce dump_flags_t type and
19121 use it instead of int type.
19123 * cfghooks.c: Likewise.
19124 * cfghooks.h (struct cfg_hooks): Likewise.
19125 * cfgrtl.c: Likewise.
19126 * cfgrtl.h: Likewise.
19127 * cgraph.c (cgraph_node::get_body): Likewise.
19128 * coretypes.h: Likewise.
19129 * domwalk.c: Likewise.
19130 * domwalk.h: Likewise.
19131 * dumpfile.c (struct dump_option_value_info): Likewise.
19132 (dump_enable_all): Likewise.
19133 (dump_switch_p_1): Likewise.
19134 (opt_info_switch_p): Likewise.
19135 * dumpfile.h (enum tree_dump_index): Likewise.
19136 (struct dump_file_info): Likewise.
19137 * genemit.c: Likewise.
19138 * generic-match-head.c: Likewise.
19139 * gengtype.c (open_base_files): Likewise.
19140 * gimple-pretty-print.c: Likewise.
19141 * gimple-pretty-print.h: Likewise.
19142 * graph.c (print_graph_cfg): Likewise.
19143 * graphite-scop-detection.c (dot_all_sese): Likewise.
19144 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
19145 * loop-unroll.c (report_unroll): Likewise.
19146 * passes.c (pass_manager::register_one_dump_file): Likewise.
19147 * print-tree.c: Likewise.
19148 * statistics.c: Likewise.
19149 * tree-cfg.c: Likewise.
19150 * tree-cfg.h: Likewise.
19151 * tree-dfa.c: Likewise.
19152 * tree-dfa.h: Likewise.
19153 * tree-dump.c (dump_function): Likewise.
19154 * tree-dump.h (struct dump_info): Likewise.
19155 * tree-pretty-print.c: Likewise.
19156 * tree-pretty-print.h: Likewise.
19157 * tree-ssa-live.c: Likewise.
19158 * tree-ssa-live.h: Likewise.
19159 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
19160 * tree-vect-loop.c: Likewise.
19161 * tree-vect-slp.c: Likewise.
19163 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
19164 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19166 PR tree-optimization/80457
19167 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
19168 of all arguments to a statement as scalar_to_vec operations.
19169 (vectorizable_call): Adjust call to vect_model_simple_cost for
19171 (vectorizable_conversion): Likewise.
19172 (vectorizable_assignment): Likewise.
19173 (vectorizable_shift): Likewise.
19174 (vectorizable_operation): Likewise.
19175 (vectorizable_comparison): Likewise.
19176 (vect_is_simple_cond): Record the def types for operands.
19177 (vectorizable_condition): Likewise, call vect_model_simple_cost.
19178 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
19179 for statement argument count.
19181 2017-05-16 Carl Love <cel@us.ibm.com>
19183 * config/rs6000/rs6000-c: Add support for built-in functions
19184 vector unsigned long long vec_bperm (vector unsigned long long,
19185 vector unsigned char)
19186 vector signed long long vec_mule (vector signed int,
19188 vector unsigned long long vec_mule (vector unsigned int,
19189 vector unsigned int)
19190 vector signed long long vec_mulo (vector signed int,
19192 vector unsigned long long vec_mulo (vector unsigned int,
19193 vector unsigned int)
19194 vector signed char vec_sldw (vector signed char,
19195 vector signed char,
19197 vector unsigned char vec_sldw (vector unsigned char,
19198 vector unsigned char,
19200 vector signed short vec_sldw (vector signed short,
19201 vector signed short,
19203 vector unsigned short vec_sldw (vector unsigned short,
19204 vector unsigned short,
19206 vector signed int vec_sldw (vector signed int,
19209 vector unsigned int vec_sldw (vector unsigned int,
19210 vector unsigned int,
19212 vector signed long long vec_sldw (vector signed long long,
19213 vector signed long long,
19215 vector unsigned long long vec_sldw (vector unsigned long long,
19216 vector unsigned long long,
19218 * config/rs6000/rs6000-c: Add support for built-in functions
19219 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
19220 * config/rs6000/altivec.h: Add defintion for vec_sldw.
19221 * doc/extend.texi: Update the built-in documentation for the
19222 new built-in functions.
19224 2017-05-16 Marek Polacek <polacek@redhat.com>
19228 * tree.c (save_expr): Don't fold the expression.
19230 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
19232 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
19233 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
19234 and (?*y,m). Update insn attributes.
19236 2017-05-16 Martin Liska <mliska@suse.cz>
19238 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
19239 flags argument of print_gimple_stmt, print_gimple_expr,
19240 print_generic_stmt and print_generic_expr.
19241 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
19242 * coretypes.h: Likewise.
19243 * except.c (dump_eh_tree): Likewise.
19244 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
19245 * gimple-pretty-print.h: Likewise.
19246 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
19247 (backprop::push_to_worklist): Likewise.
19248 (backprop::pop_from_worklist): Likewise.
19249 (backprop::process_use): Likewise.
19250 (backprop::intersect_uses): Likewise.
19251 (note_replacement): Likewise.
19252 * gimple-ssa-store-merging.c
19253 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
19254 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
19255 (pass_store_merging::execute): Likewise.
19256 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
19257 (ssa_base_cand_dump_callback): Likewise.
19258 (dump_incr_vec): Likewise.
19259 (replace_refs): Likewise.
19260 (replace_mult_candidate): Likewise.
19261 (create_add_on_incoming_edge): Likewise.
19262 (create_phi_basis): Likewise.
19263 (insert_initializers): Likewise.
19264 (all_phi_incrs_profitable): Likewise.
19265 (introduce_cast_before_cand): Likewise.
19266 (replace_one_candidate): Likewise.
19267 * gimplify.c (gimplify_expr): Likewise.
19268 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
19269 (set_rename): Likewise.
19270 (rename_uses): Likewise.
19271 (copy_loop_phi_nodes): Likewise.
19272 (add_close_phis_to_merge_points): Likewise.
19273 (copy_loop_close_phi_args): Likewise.
19274 (copy_cond_phi_args): Likewise.
19275 (graphite_copy_stmts_from_block): Likewise.
19276 (translate_pending_phi_nodes): Likewise.
19277 * graphite-poly.c (print_pdr): Likewise.
19278 (dump_gbb_cases): Likewise.
19279 (dump_gbb_conditions): Likewise.
19280 (print_scop_params): Likewise.
19281 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
19282 (build_cross_bb_scalars_use): Likewise.
19283 (gather_bbs::before_dom_children): Likewise.
19284 * hsa-dump.c (dump_hsa_immed): Likewise.
19285 * ipa-cp.c (print_ipcp_constant_value): Likewise.
19286 (get_replacement_map): Likewise.
19287 * ipa-inline-analysis.c (dump_condition): Likewise.
19288 (estimate_function_body_sizes): Likewise.
19289 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
19290 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
19291 * ipa-prop.c (ipa_dump_param): Likewise.
19292 (ipa_print_node_jump_functions_for_edge): Likewise.
19293 (ipa_modify_call_arguments): Likewise.
19294 (ipa_modify_expr): Likewise.
19295 (ipa_dump_param_adjustments): Likewise.
19296 (ipa_dump_agg_replacement_values): Likewise.
19297 (ipcp_modif_dom_walker::before_dom_children): Likewise.
19298 * ipa-pure-const.c (check_stmt): Likewise.
19299 (pass_nothrow::execute): Likewise.
19300 * ipa-split.c (execute_split_functions): Likewise.
19301 * omp-offload.c (dump_oacc_loop_part): Likewise.
19302 (dump_oacc_loop): Likewise.
19303 * trans-mem.c (tm_log_emit): Likewise.
19304 (tm_memopt_accumulate_memops): Likewise.
19305 (dump_tm_memopt_set): Likewise.
19306 (dump_tm_memopt_transform): Likewise.
19307 * tree-cfg.c (gimple_verify_flow_info): Likewise.
19308 (print_loop): Likewise.
19309 * tree-chkp-opt.c (chkp_print_addr): Likewise.
19310 (chkp_gather_checks_info): Likewise.
19311 (chkp_get_check_result): Likewise.
19312 (chkp_remove_check_if_pass): Likewise.
19313 (chkp_use_outer_bounds_if_possible): Likewise.
19314 (chkp_reduce_bounds_lifetime): Likewise.
19315 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
19316 (chkp_mark_completed_bounds): Likewise.
19317 (chkp_register_incomplete_bounds): Likewise.
19318 (chkp_mark_invalid_bounds): Likewise.
19319 (chkp_maybe_copy_and_register_bounds): Likewise.
19320 (chkp_build_returned_bound): Likewise.
19321 (chkp_get_bound_for_parm): Likewise.
19322 (chkp_build_bndldx): Likewise.
19323 (chkp_get_bounds_by_definition): Likewise.
19324 (chkp_generate_extern_var_bounds): Likewise.
19325 (chkp_get_bounds_for_decl_addr): Likewise.
19326 * tree-chrec.c (chrec_apply): Likewise.
19327 * tree-data-ref.c (dump_data_reference): Likewise.
19328 (dump_subscript): Likewise.
19329 (dump_data_dependence_relation): Likewise.
19330 (analyze_overlapping_iterations): Likewise.
19331 * tree-inline.c (expand_call_inline): Likewise.
19332 (tree_function_versioning): Likewise.
19333 * tree-into-ssa.c (dump_defs_stack): Likewise.
19334 (dump_currdefs): Likewise.
19335 (dump_names_replaced_by): Likewise.
19336 (dump_update_ssa): Likewise.
19337 (update_ssa): Likewise.
19338 * tree-object-size.c (pass_object_sizes::execute): Likewise.
19339 * tree-parloops.c (build_new_reduction): Likewise.
19340 (try_create_reduction_list): Likewise.
19341 (ref_conflicts_with_region): Likewise.
19342 (oacc_entry_exit_ok_1): Likewise.
19343 (oacc_entry_exit_single_gang): Likewise.
19344 * tree-pretty-print.h: Likewise.
19345 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
19346 (get_scalar_evolution): Likewise.
19347 (add_to_evolution): Likewise.
19348 (get_loop_exit_condition): Likewise.
19349 (analyze_evolution_in_loop): Likewise.
19350 (analyze_initial_condition): Likewise.
19351 (analyze_scalar_evolution): Likewise.
19352 (instantiate_scev): Likewise.
19353 (number_of_latch_executions): Likewise.
19354 (gather_chrec_stats): Likewise.
19355 (final_value_replacement_loop): Likewise.
19356 (scev_const_prop): Likewise.
19357 * tree-sra.c (dump_access): Likewise.
19358 (disqualify_candidate): Likewise.
19359 (create_access): Likewise.
19360 (reject): Likewise.
19361 (maybe_add_sra_candidate): Likewise.
19362 (create_access_replacement): Likewise.
19363 (analyze_access_subtree): Likewise.
19364 (analyze_all_variable_accesses): Likewise.
19365 (sra_modify_assign): Likewise.
19366 (initialize_constant_pool_replacements): Likewise.
19367 (find_param_candidates): Likewise.
19368 (decide_one_param_reduction): Likewise.
19369 (replace_removed_params_ssa_names): Likewise.
19370 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
19371 * tree-ssa-copy.c (dump_copy_of): Likewise.
19372 (copy_prop_visit_cond_stmt): Likewise.
19373 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
19374 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
19375 (record_equivalences_from_stmt): Likewise.
19376 * tree-ssa-dse.c (compute_trims): Likewise.
19377 (delete_dead_call): Likewise.
19378 (delete_dead_assignment): Likewise.
19379 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
19380 (forward_propagate_into_cond): Likewise.
19381 (pass_forwprop::execute): Likewise.
19382 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
19383 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
19385 (move_computations_worker): Likewise.
19386 (execute_sm): Likewise.
19387 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
19388 (remove_exits_and_undefined_stmts): Likewise.
19389 (remove_redundant_iv_tests): Likewise.
19390 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
19391 (adjust_iv_update_pos): Likewise.
19392 * tree-ssa-math-opts.c (bswap_replace): Likewise.
19393 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
19394 (value_replacement): Likewise.
19395 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
19396 * tree-ssa-pre.c (print_pre_expr): Likewise.
19397 (get_representative_for): Likewise.
19398 (create_expression_by_pieces): Likewise.
19399 (insert_into_preds_of_block): Likewise.
19400 (eliminate_insert): Likewise.
19401 (eliminate_dom_walker::before_dom_children): Likewise.
19402 (eliminate): Likewise.
19403 (remove_dead_inserted_code): Likewise.
19404 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
19405 * tree-ssa-reassoc.c (get_rank): Likewise.
19406 (eliminate_duplicate_pair): Likewise.
19407 (eliminate_plus_minus_pair): Likewise.
19408 (eliminate_not_pairs): Likewise.
19409 (undistribute_ops_list): Likewise.
19410 (eliminate_redundant_comparison): Likewise.
19411 (update_range_test): Likewise.
19412 (optimize_range_tests_var_bound): Likewise.
19413 (optimize_vec_cond_expr): Likewise.
19414 (rewrite_expr_tree): Likewise.
19415 (rewrite_expr_tree_parallel): Likewise.
19416 (linearize_expr): Likewise.
19417 (break_up_subtract): Likewise.
19418 (linearize_expr_tree): Likewise.
19419 (attempt_builtin_powi): Likewise.
19420 (attempt_builtin_copysign): Likewise.
19421 (transform_stmt_to_copy): Likewise.
19422 (transform_stmt_to_multiply): Likewise.
19423 (dump_ops_vector): Likewise.
19424 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
19425 (print_scc): Likewise.
19426 (set_ssa_val_to): Likewise.
19427 (visit_reference_op_store): Likewise.
19428 (visit_use): Likewise.
19429 (sccvn_dom_walker::before_dom_children): Likewise.
19430 (run_scc_vn): Likewise.
19431 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
19433 (expr_hash_elt::print): Likewise.
19434 (const_and_copies::pop_to_marker): Likewise.
19435 (const_and_copies::record_const_or_copy_raw): Likewise.
19436 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
19437 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
19438 (dump_predicates): Likewise.
19439 (find_uninit_use): Likewise.
19440 (warn_uninitialized_phi): Likewise.
19441 (pass_late_warn_uninitialized::execute): Likewise.
19442 * tree-ssa.c (verify_vssa): Likewise.
19443 (verify_ssa): Likewise.
19444 (maybe_optimize_var): Likewise.
19445 * tree-vrp.c (dump_value_range): Likewise.
19446 (dump_all_value_ranges): Likewise.
19447 (dump_asserts_for): Likewise.
19448 (register_edge_assert_for_2): Likewise.
19449 (vrp_visit_cond_stmt): Likewise.
19450 (vrp_visit_switch_stmt): Likewise.
19451 (vrp_visit_stmt): Likewise.
19452 (vrp_visit_phi_node): Likewise.
19453 (simplify_cond_using_ranges_1): Likewise.
19454 (fold_predicate_in): Likewise.
19455 (evrp_dom_walker::before_dom_children): Likewise.
19456 (evrp_dom_walker::push_value_range): Likewise.
19457 (evrp_dom_walker::pop_value_range): Likewise.
19458 (execute_early_vrp): Likewise.
19460 2017-05-16 Richard Biener <rguenther@suse.de>
19462 * dwarf2out.c (loc_list_from_tree_1): Do not create
19463 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
19465 2017-05-16 Richard Biener <rguenther@suse.de>
19467 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
19469 (note_variable_value_in_expr): If we resolved the decl ref
19470 do not push to the stack.
19472 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
19474 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
19475 operations in fast-math mode.
19476 (vaddq_f16): Likewise.
19477 (vmul_f16): Likewise.
19478 (vmulq_f16): Likewise.
19479 (vsub_f16): Likewise.
19480 (vsubq_f16): Likewise.
19481 * config/arm/neon.md (add<mode>3): New.
19483 (fma:<VH:mode>3): New. Also remove outdated comment.
19486 2017-05-16 Martin Liska <mliska@suse.cz>
19490 * ipa-devirt.c (warn_types_mismatch): Fix typo.
19491 (odr_types_equivalent_p): Likewise.
19493 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
19495 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
19497 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
19500 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
19501 non-interunit SSE move alternatives with '?'.
19502 (zero-extendsidi peephole2): New peephole to skip intermediate
19503 general register in SSE zero-extend sequence.
19505 2017-05-15 Jeff Law <law@redhat.com>
19507 * reorg.c (relax_delay_slots): Create a new variable to hold
19508 the temporary target rather than clobbering TARGET_LABEL.
19510 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
19511 missing argument to extract_bit_field call.
19512 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
19514 2017-05-15 Martin Liska <mliska@suse.cz>
19517 * gcc.c (process_command): Do not allow empty argument of -o option.
19519 2017-05-15 Renlin Li <renlin.li@arm.com>
19521 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
19522 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
19523 * config/aarch64/constraints.md (Usf): Add long call check.
19524 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
19525 (call_value): Likewise.
19526 (sibcall): Likewise.
19527 (sibcall_value): Likewise.
19529 (call_value_insn): New.
19530 (sibcall_insn): Update rtx pattern.
19531 (sibcall_value_insn): Likewise.
19532 (call_internal): Remove.
19533 (call_value_internal): Likewise.
19534 (sibcall_internal): Likewise.
19535 (sibcall_value_internal): Likewise.
19536 (call_reg): Likewise.
19537 (call_symbol): Likewise.
19538 (call_value_reg): Likewise.
19539 (call_value_symbol): Likewise.
19541 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
19544 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
19546 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
19548 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
19549 compatible with CCGOCmode and with CCZmode.
19551 2017-05-14 Martin Sebor <msebor@redhat.com>
19553 PR middle-end/77671
19554 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
19555 (gimple_fold_builtin_snprintf): Same.
19556 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
19557 (gimple_fold_builtin_snprintf): Same.
19558 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
19559 of character types.
19560 (is_call_safe): New function.
19561 (try_substitute_return_value): Call it.
19562 (try_simplify_call): New function.
19563 (pass_sprintf_length::handle_gimple_call): Call it.
19565 2017-05-14 Martin Sebor <msebor@redhat.com>
19567 PR middle-end/80669
19568 * builtins.c (expand_builtin_stpncpy): Simplify.
19570 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
19572 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
19573 * config/i386/i386.h
19574 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
19575 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
19576 (struct machine_function): Add new members call_ms2sysv,
19577 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
19578 (struct machine_frame_state): New fields sp_realigned and
19579 sp_realigned_offset.
19580 * config/i386/i386.c
19581 (enum xlogue_stub): New enum.
19582 (enum xlogue_stub_sets): New enum.
19583 (class xlogue_layout): New class.
19584 (struct ix86_frame): New fields stack_realign_allocate_offset,
19585 stack_realign_offset and outlined_save_offset. Modify comments to
19586 detail stack layout when using out-of-line stubs.
19587 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
19588 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
19589 -mcall-ms2sysv-xlogues.
19590 (stub_managed_regs): New static variable.
19591 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
19592 registers managed by out-of-line stub.
19593 (disable_call_ms2sysv_xlogues): New function.
19594 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
19595 m->call_ms2sysv when appropriate and compute frame layout for
19597 (sp_valid_at, fp_valid_at): New inline functions.
19598 (choose_basereg): New function.
19599 (choose_baseaddr): Add align parameter, use choose_basereg and modify
19601 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
19602 Use align parameter of choose_baseaddr to generated aligned SSE movs
19604 (pro_epilogue_adjust_stack): Modify to track
19605 machine_frame_state::sp_realigned.
19606 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
19607 (ix86_nsaved_sseregs): Likewise.
19608 (ix86_emit_save_regs): Likewise.
19609 (ix86_emit_save_regs_using_mov): Likewise.
19610 (ix86_emit_save_sse_regs_using_mov): Likewise.
19611 (get_scratch_register_on_entry): Likewise.
19612 (gen_frame_set): New function.
19613 (gen_frame_load): Likewise.
19614 (gen_frame_store): Likewise.
19615 (emit_outlined_ms2sysv_save): Likewise.
19616 (emit_outlined_ms2sysv_restore): Likewise.
19617 (ix86_expand_prologue): Modify stack re-alignment code and call
19618 emit_outlined_ms2sysv_save when appropriate.
19619 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
19620 parameter rtx_insn *insn, which allows the function to be used to only
19621 generate the notes.
19622 (ix86_expand_epilogue): Modify validity checks of frame and stack
19623 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
19624 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
19625 * config/i386/predicates.md
19626 (save_multiple): New predicate.
19627 (restore_multiple): Likewise.
19628 * config/i386/sse.md
19629 (save_multiple<mode>): New pattern.
19630 (save_multiple_realign<mode>): Likewise.
19631 (restore_multiple<mode>): Likewise.
19632 (restore_multiple_and_return<mode>): Likewise.
19633 (restore_multiple_leave_return<mode>): Likewise.
19634 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
19636 2017-05-14 Julia Koval <julia.koval@intel.com>
19638 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
19639 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
19640 (__builtin_ia32_xsetbv): New builtins.
19641 * config/i386/i386.c (ix86_expand_special_args_builtin):
19643 (ix86_expand_builtin): Special expand for new intrinsics.
19644 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
19645 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
19646 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
19648 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19650 * cfganal.c (inverted_post_order_compute): Change argument type
19652 * cfganal.h (inverted_post_order_compute): Adjust prototype.
19653 * df-core.c (rest_of_handle_df_initialize): Adjust.
19654 (rest_of_handle_df_finish): Likewise.
19655 (df_analyze_1): Likewise.
19656 (df_analyze): Likewise.
19657 (loop_inverted_post_order_compute): Change argument to be a vec *.
19658 (df_analyze_loop): Adjust.
19659 (df_get_n_blocks): Likewise.
19660 (df_get_postorder): Likewise.
19661 * df.h (struct df_d): Change field to be a vec.
19662 * lcm.c (compute_laterin): Adjust.
19663 (compute_available): Likewise.
19664 * lra-lives.c (lra_create_live_ranges_1): Likewise.
19665 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
19666 * tree-ssa-pre.c (compute_antic): Likewise.
19668 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19670 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
19671 (depth_first_search::depth_first_search): Change structure init
19672 function to this constructor.
19673 (depth_first_search::add_bb): Rename function to this member.
19674 (depth_first_search::execute): Likewise.
19675 (flow_dfs_compute_reverse_finish): Adjust.
19677 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19679 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
19680 (longest_simple_path): Likewise.
19681 * shrink-wrap.c (spread_components): Likewise.
19682 (disqualify_problematic_components): Likewise.
19683 (emit_common_heads_for_components): Likewise.
19684 (emit_common_tails_for_components): Likewise.
19685 (insert_prologue_epilogue_for_components): Likewise.
19687 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19689 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
19692 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19694 * df-core.c (df_set_blocks): Start using auto_bitmap.
19695 (df_compact_blocks): Likewise.
19696 * df-problems.c (df_rd_confluence_n): Likewise.
19697 * df-scan.c (df_insn_rescan_all): Likewise.
19698 (df_process_deferred_rescans): Likewise.
19699 (df_update_entry_block_defs): Likewise.
19700 (df_update_exit_block_uses): Likewise.
19701 (df_entry_block_bitmap_verify): Likewise.
19702 (df_exit_block_bitmap_verify): Likewise.
19703 (df_scan_verify): Likewise.
19704 * lra-constraints.c (lra_constraints): Likewise.
19705 (undo_optional_reloads): Likewise.
19706 (lra_undo_inheritance): Likewise.
19707 * lra-remat.c (calculate_gen_cands): Likewise.
19708 (do_remat): Likewise.
19709 * lra-spills.c (assign_spill_hard_regs): Likewise.
19710 (spill_pseudos): Likewise.
19711 * tree-ssa-pre.c (bitmap_set_and): Likewise.
19712 (bitmap_set_subtract_values): Likewise.
19714 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19716 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
19717 management with auto_bitmap.
19718 (fix_inter_tick): Likewise.
19719 (fix_recovery_deps): Likewise.
19720 * ira.c (add_store_equivs): Likewise.
19721 (find_moveable_pseudos): Likewise.
19722 (split_live_ranges_for_shrink_wrap): Likewise.
19723 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
19724 (rtx_reuse_manager::seen_def_p): Likewise.
19725 (rtx_reuse_manager::set_seen_def): Likewise.
19726 * print-rtl.h (class rtx_reuse_manager): Likewise.
19728 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19730 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
19732 (migrate_btr_def): Likewise.
19733 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
19734 * df-core.c (loop_post_order_compute): Likewise.
19735 (loop_inverted_post_order_compute): Likewise.
19736 * hsa-common.h: Likewise.
19737 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
19738 * init-regs.c (initialize_uninitialized_regs): Likewise.
19739 * ipa-inline.c (resolve_noninline_speculation): Likewise.
19740 (inline_small_functions): Likewise.
19741 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
19742 * ira.c (combine_and_move_insns): Likewise.
19743 (build_insn_chain): Likewise.
19744 * loop-invariant.c (find_invariants): Likewise.
19745 * lower-subreg.c (propagate_pseudo_copies): Likewise.
19746 * predict.c (tree_predict_by_opcode): Likewise.
19747 (predict_paths_leading_to): Likewise.
19748 (predict_paths_leading_to_edge): Likewise.
19749 (estimate_loops_at_level): Likewise.
19750 (estimate_loops): Likewise.
19751 * shrink-wrap.c (try_shrink_wrapping): Likewise.
19752 (spread_components): Likewise.
19753 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
19754 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
19755 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
19756 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
19757 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
19758 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
19759 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
19760 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
19761 (mark_threaded_blocks): Likewise.
19762 (thread_through_all_blocks): Likewise.
19763 * tree-ssa.c (verify_ssa): Likewise.
19764 (execute_update_addresses_taken): Likewise.
19765 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
19767 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19769 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
19771 (post_order_compute): Likewise.
19772 (inverted_post_order_compute): Likewise.
19773 (pre_and_rev_post_order_compute_fn): Likewise.
19775 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19777 * genrecog.c (int_set::int_set): Explicitly construct our
19778 auto_vec base class.
19779 * vec.h (auto_vec::auto_vec): New constructor.
19781 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19783 * bitmap.h (class auto_bitmap): New constructor taking
19784 bitmap_obstack * argument.
19786 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19788 * bitmap.h (class auto_bitmap): Change type of m_bits to
19789 bitmap_head, and adjust ctor / dtor and member operators.
19791 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
19793 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
19794 when returned register mode doesn't match original mode.
19796 2017-05-12 Jeff Law <law@redhat.com>
19797 Jakub Jelinek <jakub@redhat.com>
19799 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
19800 we look for cc setter after the compare-elim changes.
19801 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
19802 within the vector to match what compare-elim now expects.
19803 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
19804 (xorsi3_flags, one_cmplsi2_flags): Likewise.
19806 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
19807 after the compare-elim changes.
19808 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
19809 the vector to match what compare-elim now expects.
19810 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
19811 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
19812 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
19813 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
19814 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
19816 * config/visium/visium.c (single_set_and_flags): Fix where
19817 we look for cc setter after the compare-elim changes.
19818 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
19819 with the vector to match what compare-elim now expects.
19820 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
19821 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
19822 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
19823 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
19824 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
19825 (neg<mode>2_insn_set_overflow): Likewise.
19827 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
19829 PR middle-end/79794
19830 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
19831 maybe_expand_insn call, set ops[0].target. If still set after call,
19832 set alt_rtl. Add extra arg to recursive calls.
19833 (extract_bit_field): Add alt_rtl argument. Pass to
19835 * expmed.h (extract_bit_field): Fix prototype.
19836 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
19837 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
19838 to extract_bit_field_calls.
19839 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
19840 Pass alt_rtl to extract_bit_field calls.
19841 * calls.c (store_unaligned_arguments_into_psuedos)
19842 load_register_parameters): Pass extra NULL to extract_bit_field calls.
19843 * optabs.c (maybe_legitimize_operand): Clear op->target when call
19845 * optabs.h (struct expand_operand): Add target bitfield.
19847 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19849 * compare-elim.c (try_eliminate_compare): Canonicalize
19850 operation with embedded compare to
19851 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
19852 (set (reg) (operation)].
19854 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
19856 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19859 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
19860 cost of adding a carry flag for ADC instruction.
19861 [case MINUS]: Ignore the cost of subtracting a carry flag
19862 for SBB instruction.
19864 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
19866 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
19868 * config/rs6000/bmiintrin.h: New file.
19869 * config/rs6000/bmi2intrin.h: New file.
19870 * config/rs6000/x86intrin.h: New file.
19872 2017-05-12 Jeff Law <law@redhat.com>
19874 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
19877 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
19879 PR middle-end/80707
19880 * tree-cfg.c: Remove cfg edges of unreachable case statements.
19882 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19884 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19885 early expansion of vector divide builtins.
19886 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
19887 builtins identified as having unsigned arguments.
19889 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19891 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
19892 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
19893 expansion of vector logical operations (and, andc, or, xor,
19896 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19898 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
19899 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
19901 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19903 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19904 early GIMPLE expansion of vector multiplies.
19906 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19908 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
19909 TARGET_HAVE_MOVT conditional.
19910 (movt splitter): Likewise.
19912 2017-05-12 Richard Biener <rguenther@suse.de>
19914 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
19915 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19916 Fold all stmts not inplace.
19918 2017-05-12 Richard Biener <rguenther@suse.de>
19920 PR tree-optimization/80713
19921 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
19922 inserted_exprs bit for not removed stmts.
19924 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
19926 PR middle-end/69921
19927 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
19928 parallelized" attribute for parallelized OpenACC kernels.
19929 * omp-offload.c (execute_oacc_device_lower): Use it.
19931 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
19932 Set "oacc kernels" attribute.
19933 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
19934 parameter. Adjust all users.
19935 (oacc_fn_attrib_kernels_p): Remove function.
19936 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
19937 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
19938 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
19939 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
19940 assert "oacc kernels" attribute is set.
19942 2017-05-11 Carl Love <cel@us.ibm.com>
19944 * config/rs6000/rs6000-c: Add support for built-in functions
19945 vector unsigned char vec_popcnt (vector signed char)
19946 vector unsigned char vec_popcnt (vector unsigned char)
19947 vector unsigned short vec_popcnt (vector signed short)
19948 vector unsigned short vec_popcnt (vector unsigned short)
19949 vector unsigned int vec_popcnt (vector signed int)
19950 vector unsigned int vec_popcnt (vector unsigned int)
19951 vector unsigned long long vec_popcnt (vector signed long long)
19952 vector unsigned long long vec_popcnt (vector unsigned long long)
19953 vector signed long long vec_slo (vector signed long long,
19954 vector signed char)
19955 vector signed long long vec_slo (vector signed long long,
19956 vector unsigned char)
19957 vector unsigned long long vec_slo (vector unsigned long long,
19958 vector signed char)
19959 vector unsigned long long vec_slo (vector unsigned long long,
19960 vector unsigned char)
19961 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
19962 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
19963 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
19964 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
19965 * doc/extend.texi: Update the built-in documentation file for the
19966 new built-in functions.
19968 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19970 * attribs.h (sorted_attr_string): Move machine independent
19971 functions for target clone support from the i386 port to common
19972 code. Rename ix86_function_versions to common_function_versions.
19973 Rename make_name to make_unique_name.
19974 (common_function_versions): Likewise.
19975 (make_unique_name): Likewise.
19976 (make_dispatcher_decl): Likewise.
19977 (is_function_default_version): Likewise.
19978 * attribs.c (attr_strcmp): Likewise.
19979 (sorted_attr_string): Likewise.
19980 (common_function_versions): Likewise.
19981 (make_unique_name): Likewise.
19982 (make_dispatcher_decl): Likewise.
19983 (is_function_default_version): Likewise.
19984 * config/i386/i386.c (attr_strcmp): Likewise.
19985 (sorted_attr_string): Likewise.
19986 (ix86_function_versions): Likewise.
19987 (make_name): Likewise.
19988 (make_dispatcher_decl): Likewise.
19989 (is_function_default_version): Likewise.
19990 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
19992 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19995 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19996 Account for direct move costs for vec_construct of integer
19999 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
20002 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
20003 (UNSPEC_STX_ATOMIC): Ditto.
20004 (loaddi_via_sse): New insn.
20005 (storedi_via_sse): Ditto.
20006 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
20007 Update corresponding peephole2 patterns.
20008 (atomic_storedi_fpu): Ditto.
20010 2017-05-11 Julia Koval <julia.koval@intel.com>
20012 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
20013 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
20015 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
20016 (__builtin_ia32_rsqrt14ss_mask): New builtins.
20017 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
20019 2017-05-11 Nathan Sidwell <nathan@acm.org>
20021 * graphite-poly.c: Include dumpfile.h.
20023 * dumpfle.h (dump_function): Declare here ...
20024 * tree-dump.h (dump_function): ... not here.
20025 * dumpfile.c: #include tree-cfg.h.
20026 (dump_function): Move here from ...
20027 * tree-dump.c (dump_function): ... here.
20028 * gimplify.c: #include splay-tree.h, not tree-dump.h.
20029 * graphite-poly.c: Don't include tree-dump.h.
20030 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
20031 * print-tree.c: Likewise.
20032 * stor-layout.c: Likewise.
20033 * tree-nested.c: Likewise.
20035 * dumpfile.c (dump_start): Use TDF_FLAGS.
20036 (dump_enable_all): Fix TDF_KIND check thinko.
20038 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
20040 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20041 array entries to represent two legal parameterizations of the
20042 overloaded __builtin_cmpb function, as represented by the
20043 P6_OV_BUILTIN_CMPB constant.
20044 (altivec_resolve_overloaded_builtin): Add special case handling
20045 for the __builtin_cmpb function, as represented by the
20046 P6_OV_BUILTIN_CMPB constant.
20047 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
20048 (BU_P6_64BIT_2): New macro.
20049 (BU_P6_OVERLOAD_2): New macro
20050 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
20051 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
20052 (CMPB): Add overload support to represent both 32-bit and 64-bit
20053 compare-bytes function.
20054 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
20055 support for TARGET_CMPB.
20056 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
20057 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20058 documentation of the __builtin_cmpb overloaded built-in function.
20060 2017-05-11 Richard Biener <rguenther@suse.de>
20062 PR tree-optimization/80705
20063 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
20064 bases are not vectorizable.
20066 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20068 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
20069 when counting register pressure.
20071 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20073 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
20074 (iv_ca_more_deps): Renamed to ...
20075 (iv_ca_compare_deps): ... this.
20076 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
20078 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20080 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
20082 (determine_group_iv_costs): ... here.
20083 (find_inv_vars_cb): Record inv var if it's not recorded before.
20085 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20087 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
20088 (get_shiftadd_cost): Ditto.
20090 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20092 * tree-ssa-address.c: Include header file.
20093 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
20095 (add_to_parts): Refactor.
20096 (addr_to_parts): New parameter. Update use of move_hint_to_base.
20097 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
20100 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20102 PR tree-optimization/53090
20103 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
20105 (extract_cond_operands): Detect condition with IV on both sides
20106 and return COMP_IV_EXPR_2.
20107 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
20108 (rewrite_use_compare): Simplify by removing call to function
20109 extract_cond_operands.
20111 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20113 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
20114 (extract_cond_operands): Detect condition comparing against non-
20115 invariant bound and return appropriate enum value.
20116 (find_interesting_uses_cond): Update use of extract_cond_operands.
20117 Handle its return value accordingly.
20118 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
20120 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20122 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
20123 nonlinear iv_use computation in loop invariant sensitive way.
20125 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20127 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
20128 (find_iv_candidates): Call relate_compare_use_with_all_cands.
20130 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20132 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
20133 (dump_cand): Support iv_cand.inv_exprs.
20134 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
20136 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
20139 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20141 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
20143 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
20144 as local function. Include necessary header files.
20145 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
20147 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20149 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
20151 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20153 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
20154 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
20155 RSHIFT_EXPR and BIT_NOT_EXPR.
20157 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20159 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
20160 (adjust_setup_cost): New parameter supporting round up adjustment.
20161 (struct address_cost_data): Delete.
20162 (force_expr_to_var_cost): Don't bound cost with spill_cost.
20163 (split_address_cost, ptr_difference_cost): Delete.
20164 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
20165 (struct ainc_cost_data): New struct.
20166 (get_address_cost_ainc): New function.
20167 (get_address_cost, get_computation_cost): Reimplement.
20168 (determine_group_iv_cost_address): Record inv_expr for all uses of
20170 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
20171 (iv_ca_has_deps): Reimplemented to ...
20172 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
20174 (iv_ca_extend): Call iv_ca_more_deps.
20176 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20178 * tree-ssa-address.c (struct mem_address): Move to header file.
20179 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
20180 * tree-ssa-address.h (struct mem_address): Move from C file.
20181 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
20183 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20185 * tree-affine.h (aff_combination_type): New interface.
20186 (aff_combination_zero_p): Remove static.
20187 (aff_combination_const_p): New interface.
20188 (aff_combination_singleton_var_p): New interfaces.
20190 2017-05-11 Richard Biener <rguenther@suse.de>
20192 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20193 Skip unreachable blocks and destinations.
20194 (eliminate): Move stmt removal and fixup ...
20195 (fini_eliminate): ... here. Skip inserted exprs.
20196 (pass_pre::execute): Move fini_pre after fini_eliminate.
20197 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
20198 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
20199 PRE to get rid of dead code that has invalid SSA form and
20200 split critical edges again.
20202 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20204 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20206 2017-05-11 Richard Biener <rguenther@suse.de>
20208 * passes.c (execute_function_todo): Verify loops if they are
20209 said to be up-to-date.
20210 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
20211 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
20213 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
20216 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
20217 handle calling assemble_external ourself.
20220 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
20221 modes with zero size. Enhance comment.
20223 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20225 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
20226 built-ins for vec_xl and vec_xst with short and char pointer
20229 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20231 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
20232 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
20233 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
20234 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
20235 (_mm_maskz_min_round_ss): New intrinsics.
20236 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
20237 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
20238 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
20239 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
20240 (__builtin_ia32_minss_mask_round): New builtins.
20241 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20242 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20243 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
20245 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
20246 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
20248 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
20251 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20253 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
20254 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
20255 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
20256 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
20257 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
20258 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
20259 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
20260 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20261 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20262 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
20263 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
20264 (__builtin_ia32_mulss_mask_round): New builtins.
20265 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20266 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20267 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
20269 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
20270 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20272 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20275 2017-05-10 Julia Koval <julia.koval@intel.com>
20277 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
20278 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
20279 (_mm256_setr_m128i): New intrinsics.
20281 2017-05-10 Julia Koval <julia.koval@intel.com>
20283 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
20284 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
20285 (_mm_maskz_rcp14_ss): New intrinsics.
20286 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
20287 (__builtin_ia32_rcp14ss_mask): New builtins.
20288 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
20290 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
20292 PR tree-optimization/51513
20293 * tree-cfg.c (gimple_seq_unreachable_p): New function.
20294 (assert_unreachable_fallthru_edge_p): Use it.
20295 (group_case_labels_stmt): Likewise.
20296 * tree-cfg.h: Prototype it.
20297 * stmt.c: Include cfghooks.h and tree-cfg.h.
20298 (emit_case_dispatch_table) <gap_label>: New local variable.
20299 Use it to fill dispatch table gaps.
20300 Test for default_label before updating probabilities.
20301 (expand_case) <default_label>: Remove unneeded initialization.
20302 Test for unreachable default case statement and remove its edge.
20303 Set default_label accordingly.
20304 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
20306 2017-05-10 Carl Love <cel@us.ibm.com>
20308 * config/rs6000/rs6000-c: Add support for built-in functions
20309 vector signed char vec_neg (vector signed char)
20310 vector signed short int vec_neg (vector short int)
20311 vector signed int vec_neg (vector signed int)
20312 vector signed long long vec_neg (vector signed long long)
20313 vector float vec_neg (vector float)
20314 vector double vec_neg (vector double)
20315 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
20317 * config/rs6000/altivec.h: Add define for vec_neg
20318 * doc/extend.texi: Update the built-in documentation for the
20319 new built-in functions.
20321 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20323 PR tree-optimization/77644
20324 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
20326 2017-05-10 Nathan Sidwell <nathan@acm.org>
20328 * dumpfile.h (TDI_lang_all): New.
20329 (TDF_KIND): New. Renumber others
20330 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
20332 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
20334 (get_dump_file_name): Adjust suffix generation.
20335 (dump_enable_all): Use TDF_KIND.
20336 * doc/invoke.texi (-fdump-lang-all): Document.
20338 * dumpfile.h: Tabify.
20340 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
20343 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
20344 Move member access before delete.
20346 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
20348 * tree-inline.c (expand_call_inline): Split block at stmt
20351 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
20354 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
20355 are now unused after splitting mov{sf,sd}_hardfloat.
20356 (f32_lr2): Likewise.
20357 (f32_lm): Likewise.
20358 (f32_lm2): Likewise.
20359 (f32_li): Likewise.
20360 (f32_li2): Likewise.
20361 (f32_lv): Likewise.
20362 (f32_sr): Likewise.
20363 (f32_sr2): Likewise.
20364 (f32_sm): Likewise.
20365 (f32_sm2): Likewise.
20366 (f32_si): Likewise.
20367 (f32_si2): Likewise.
20368 (f32_sv): Likewise.
20369 (f32_dm): Likewise.
20370 (f32_vsx): Likewise.
20371 (f32_av): Likewise.
20372 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
20373 For movsf, order stores so the VSX stores occur before the GPR
20374 store which encourages the register allocator to use a traditional
20375 FPR instead of a GPR. For movsd, order the stores so that the GPR
20376 store comes before the VSX stores to allow the power6 to work.
20377 This is due to the power6 not having a 32-bit integer store
20378 instruction from a FPR.
20379 (movsf_hardfloat): Likewise.
20380 (movsd_hardfloat): Likewise.
20382 2017-05-09 Martin Sebor <msebor@redhat.com>
20384 PR translation/80280
20385 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
20388 PR translation/80280
20389 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
20390 data member added in r247778.
20391 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
20393 2017-05-09 Nathan Sidwell <nathan@acm.org>
20395 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
20397 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
20400 2017-05-09 Marek Polacek <polacek@redhat.com>
20402 * doc/invoke.texi: Fix typo.
20404 2017-05-09 Richard Biener <rguenther@suse.de>
20406 * tree-vrp.c (vrp_val_is_max): Adjust comment.
20407 (vrp_val_is_min): Likewise.
20408 (set_value_range_to_value): Likewise.
20409 (set_value_range_to_nonnegative): Likewise.
20410 (gimple_assign_nonzero_p): Likewise.
20411 (gimple_stmt_nonzero_p): Likewise.
20412 (vrp_int_const_binop): Likewise. Remove unreachable case.
20413 (adjust_range_with_scev): Adjust comments.
20414 (compare_range_with_value): Likewise.
20415 (extract_range_from_phi_node): Likewise.
20416 (test_for_singularity): Likewise.
20418 2017-05-09 Richard Biener <rguenther@suse.de>
20420 * tree-vrp.c (get_single_symbol): Add assert that we don't
20421 get overflowed constants as invariant part.
20422 (compare_values_warnv): Add comment before the TREE_NO_WARNING
20423 checks. Use wi::cmp instead of recursing for integer constants.
20424 (compare_values): Just ignore whether we assumed undefined
20425 overflow instead of failing the compare.
20426 (extract_range_for_var_from_comparison_expr): Add comment before the
20427 TREE_NO_WARNING sets.
20428 (test_for_singularity): Likewise.
20429 (extract_range_from_comparison): Do not disable optimization
20430 when we assumed undefined overflow.
20431 (extract_range_basic): Remove init of unused var.
20433 2017-05-09 Richard Biener <rguenther@suse.de>
20435 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
20436 (extract_range_from_multiplicative_op_1): Adjust.
20437 (extract_range_from_binary_expr_1): Use int_const_binop.
20439 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20442 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
20443 rs6000_store_data_bypass_p in seven define_bypass directives and
20444 in several comments.
20445 * config/rs6000/rs6000-protos.h: Add prototype for
20446 rs6000_store_data_bypass_p function.
20447 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
20448 function implements slightly different (rs6000-specific) semantics
20449 than store_data_bypass_p, returning false rather than aborting
20450 with assertion error when arguments do not satisfy the
20451 requirements of store data bypass.
20452 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
20453 rs6000_store_data_bypass_p.
20455 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
20457 * config/xtensa/xtensa-protos.h
20458 (xtensa_initial_elimination_offset): New declaration.
20459 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20460 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
20461 macro definition, add case for FRAME_POINTER_REGNUM when
20462 FRAME_GROWS_DOWNWARD.
20463 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
20464 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
20465 xtensa_initial_elimination_offset.
20467 2017-05-08 Nathan Sidwell <nathan@acm.org>
20469 * doc/invoke.texi: Alphabetize -fdump options.
20471 2017-05-08 Martin Sebor <msebor@redhat.com>
20473 PR translation/80280
20474 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
20476 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
20478 * target.def (compute_frame_layout): New optional target hook.
20479 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
20480 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
20481 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
20483 * reload1.c (verify_initial_elim_offsets): Likewise.
20484 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
20485 (use_simple_return_p): Call arm_compute_frame_layout if needed.
20486 (arm_get_frame_offsets): Split up into this ...
20487 (arm_compute_frame_layout): ... and this function.
20489 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
20491 * config/aarch64/constraints.md (Usa): New constraint.
20492 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
20494 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
20496 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
20497 with_multilib_list after it has been checked.
20499 2017-05-08 Richard Biener <rguenther@suse.de>
20501 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
20502 (bitmap_set_subtract_values): Likewise.
20504 2017-05-08 Richard Biener <rguenther@suse.de>
20506 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
20507 (gimple_assign_nonzero): ... this and remove strict_overflow_p
20509 (gimple_stmt_nonzero_warnv_p): Rename to ...
20510 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
20512 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
20513 (extract_range_basic): Adjust, do not disable propagation on
20514 strict overflow sensitive simplification.
20515 (vrp_visit_cond_stmt): Likewise.
20517 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
20519 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
20520 body size unconditionally.
20522 2017-05-07 Jeff Law <law@redhat.com>
20525 2017-05-06 Jeff Law <law@redhat.com>
20526 PR tree-optimization/78496
20527 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
20530 PR tree-optimization/78496
20531 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
20532 (simplify_stmt_using_ranges): Call it.
20533 (vrp_dom_walker::before_dom_children): Extract equivalences
20534 from an ASSERT_EXPR with an equality comparison against a
20537 2017-05-06 Jeff Law <law@redhat.com>
20539 PR tree-optimization/78496
20540 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
20543 PR tree-optimization/78496
20544 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
20545 (simplify_stmt_using_ranges): Call it.
20546 (vrp_dom_walker::before_dom_children): Extract equivalences
20547 from an ASSERT_EXPR with an equality comparison against a
20550 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
20552 * lra-constraints.c (lra_copy_reg_equiv): New function.
20553 (split_reg): Use it to copy equivalence information from the
20554 original register to the spill register.
20556 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
20558 PR rtl-optimization/75964
20559 * simplify-rtx.c (simplify_const_relational_operation): Remove
20560 invalid handling of comparisons of integer ABS.
20562 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
20564 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
20565 initialize to zero.
20566 (init_regs): Remove declaration.
20567 (function_arg_advance_32): Initialize error_p as boolean variable.
20569 2017-05-05 Nathan Sidwell <nathan@acm.org>
20571 * store-motion.c (remove_reachable_equiv_notes): Reformat long
20572 lines. Use for (;;).
20574 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20576 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
20577 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
20578 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
20579 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
20580 VF=2 that require versioning.
20582 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20584 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
20587 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20589 * diagnostic.h (diagnostic_override_option_index): Convert from
20590 macro to inline function.
20592 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20594 * diagnostic.c (last_module_changed_p): New function.
20595 (set_last_module): New function.
20596 (diagnostic_report_current_module): Convert macro usage to
20597 the above functions.
20598 * diagnostic.h (diagnostic_context::last_module): Strengthen
20599 from const line_map * to const line_map_ordinary *.
20600 (diagnostic_last_module_changed): Delete macro.
20601 (diagnostic_set_last_module): Delete macro.
20603 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20605 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
20606 with diagnostic_report_diagnostic.
20607 (diagnostic_n_impl_richloc): Likewise.
20608 * diagnostic.h (report_diagnostic): Delete macro.
20609 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
20610 with diagnostic_report_diagnostic.
20611 * substring-locations.c (format_warning_va): Likewise.
20613 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20615 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
20616 save/restor of format_spec. Move option-printing code to...
20617 (print_option_information): ...this new function, and
20618 reimplement by simply printing to the pretty_printer,
20619 rather than appending to the format string.
20621 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20623 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
20624 handling logic into...
20625 (update_effective_level_from_pragmas): ...this new function.
20627 2017-05-04 Andrew Waterman <andrew@sifive.com>
20629 * config/riscv/riscv.opt (mstrict-align): New option.
20630 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
20631 (SLOW_UNALIGNED_ACCESS): Define.
20632 (riscv_slow_unaligned_access): Declare.
20633 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
20635 (riscv_slow_unaligned_access): New variable.
20636 (rocket_tune_info): Set slow_unaligned_access to true.
20637 (optimize_size_tune_info): Set slow_unaligned_access to false.
20638 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
20639 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
20640 (riscv_option_override): Set riscv_slow_unaligned_access.
20641 * doc/invoke.texi: Add -mstrict-align to RISC-V.
20643 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
20645 * config/riscv/riscv.md: Unify indentation.
20647 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20652 * config/rs6000/rs6000.md (u code attribute): Add FIX and
20654 (extendsi<mode>2): Add support for doing sign extension via
20655 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
20656 don't have ISA 3.0 instructions.
20657 (extendsi<mode>2 splitter): Likewise.
20658 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
20659 generate the normal insns since SImode can now go in vector
20660 registers. Disallow the special UNSPECs needed for previous
20661 machines to hide SImode being used. Add new insns
20662 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
20663 (fix_trunc<mode>si2_stfiwx): Likewise.
20664 (fix_trunc<mode>si2_internal): Likewise.
20665 (fixuns_trunc<mode>si2): Likewise.
20666 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20667 (fctiw<u>z_<mode>_smallint): Likewise.
20668 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
20669 of floating point to 32-bit integer from doing a direct move to
20670 the GPR registers to do a store.
20671 (fctiwz_<mode>): Break long line.
20673 2017-05-05 Bin Cheng <bin.cheng@arm.com>
20675 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
20676 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
20677 (addr_list, addr_offset_valid_p): New.
20678 (split_address_groups): Check offset validity with above function.
20679 (gt-tree-ssa-loop-ivopts.h): Include header file.
20681 2017-05-05 Nathan Sidwell <nathan@acm.org>
20683 * config.gcc (arm*-*-*): Add missing 'fi'.
20685 2017-05-05 Steve Ellcey <sellcey@cavium.com>
20687 * doc/invoke.texi (-fopt-info): Explicitly say order of options
20688 included in -fopt-info does not matter.
20689 * doc/optinfo.texi (-fopt-info): Fix description of default
20690 behavour. Explicitly say order of options included in -fopt-info
20693 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20695 * config.gcc: Allow combinations of aprofile and rmprofile values for
20696 --with-multilib-list.
20697 * config/arm/t-multilib: New file.
20698 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
20699 variables. Remove setting of ISA and floating-point ABI in
20700 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
20701 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
20702 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
20703 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
20705 * config/arm/t-rmprofile: Likewise except for the matches changes.
20706 * doc/install.texi (--with-multilib-list): Document the combination of
20707 aprofile and rmprofile values and warn about pitfalls in doing that.
20709 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20711 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
20712 (movdi_aarch64): Likewise.
20714 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20716 PR tree-optimization/80632
20717 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
20719 (build_arrays): Initialize it for virtual phis.
20720 (fix_phi_nodes): Use it for virtual phis.
20722 PR tree-optimization/80558
20723 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
20724 [x, y] op z into [x op, y op z] for op & or | if conditions
20727 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
20728 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20731 * config/arm/arm.md (use_literal_pool): Remove.
20732 (64-bit immediate split): No longer takes cost into consideration
20733 if arm_disable_literal_pool is enabled.
20734 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
20735 used when arm_disable_literal_pool is enabled.
20736 (arm_max_const_double_inline_cost): Remove use of
20737 arm_disable_literal_pool.
20738 (push_minipool_fix): Add assert.
20739 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
20740 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
20741 (no_literal_pool_sf_immediate): New.
20743 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20745 PR tree-optimization/80613
20746 * tree-ssa-dce.c (propagate_necessity): Remove cases for
20747 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
20749 2017-05-05 Richard Biener <rguenther@suse.de>
20751 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
20753 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
20755 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
20756 of this flag from insn conditions due to removal from r247495.
20758 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20760 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
20762 (arm_early_store_addr_dep_ptr): Likewise.
20763 * config/arm/aarch-common-protos.h
20764 (arm_early_load_addr_dep_ptr): Add prototype.
20765 (arm_early_store_addr_dep_ptr): Likewise.
20766 * config/arm/cortex-a53.md: Add new bypasses.
20768 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20770 * tree.c (next_type_uid): Change type to unsigned.
20771 (type_hash_canon): Decrement back next_type_uid if
20772 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
20773 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
20776 2017-05-04 Martin Sebor <msebor@redhat.com>
20778 * builtins.c: Fix a trivial typo in a comment.
20780 PR middle-end/79234
20781 * builtins.c (check_sizes): Adjust to handle reading past the end.
20782 Avoid printing excessive upper bound of ranges. Use %E to print
20783 tree nodes instead of converting them to %wu.
20784 (expand_builtin_memchr): New function.
20785 (compute_dest_size): Rename...
20786 (compute_objsize): ...to this.
20787 (expand_builtin_memcpy): Adjust.
20788 (expand_builtin_mempcpy): Adjust.
20789 (expand_builtin_strcat): Adjust.
20790 (expand_builtin_strcpy): Adjust.
20791 (check_strncat_sizes): Adjust.
20792 (expand_builtin_strncat): Adjust.
20793 (expand_builtin_strncpy): Adjust and simplify.
20794 (expand_builtin_memset): Adjust.
20795 (expand_builtin_bzero): Adjust.
20796 (expand_builtin_memcmp): Adjust.
20797 (expand_builtin): Handle memcmp.
20798 (maybe_emit_chk_warning): Check strncat just once.
20800 2017-05-04 Martin Sebor <msebor@redhat.com>
20802 PR preprocessor/79214
20803 PR middle-end/79222
20804 PR middle-end/79223
20805 * builtins.c (check_sizes): Add inlining context and issue
20806 warnings even when -Wno-system-headers is set.
20807 (check_strncat_sizes): Same.
20808 (expand_builtin_strncat): Same.
20809 (expand_builtin_memmove): New function.
20810 (expand_builtin_stpncpy): Same.
20811 (expand_builtin): Handle memmove and stpncpy.
20813 2017-05-04 Bin Cheng <bin.cheng@arm.com>
20815 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
20816 which is not used any more.
20818 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20820 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
20822 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20824 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
20825 (cortexa53_tunings): Likewise.
20826 (cortexa57_tunings): Likewise.
20827 (cortexa72_tunings): Likewise.
20828 (cortexa73_tunings): Likewise.
20830 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20832 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
20833 Set loop alignment to 8.
20835 2017-05-04 Martin Sebor <msebor@redhat.com>
20837 PR translation/80280
20838 * builtins.c (expand_builtin_object_size): Add missing quoting to
20839 %D and like directives.
20840 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
20841 (hsa_type_for_tree_type): Same.
20842 (verify_function_arguments): Same.
20843 * symtab.c (symbol_table::change_decl_assembler_name): Same.
20844 * varasm.c (get_section): Same.
20847 2017-05-04 Martin Sebor <msebor@redhat.com>
20849 PR translation/80280
20850 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
20852 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20854 * config/aarch64/aarch64.c (generic_addrcost_table):
20855 Change HI/TI mode setting.
20857 2017-05-04 Martin Jambor <mjambor@suse.cz>
20859 PR tree-optimization/80622
20860 * tree-sra.c (comes_initialized_p): New function.
20861 (build_accesses_from_assign): Only set write lazily when
20862 comes_initialized_p is false.
20863 (analyze_access_subtree): Use comes_initialized_p.
20864 (propagate_subaccesses_across_link): Assert !comes_initialized_p
20865 instead of testing for PARM_DECL.
20867 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20869 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
20870 constraint on operand 0 to allow more general addressing modes.
20871 Adjust output template.
20872 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
20874 * config/aarch64/aarch64-protos.h
20875 (aarch64_address_valid_for_prefetch_p): Declare prototype.
20876 * config/aarch64/constraints.md (Dp): New address constraint.
20877 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
20880 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
20882 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
20883 update use of estimate_ipcp_clone_size_and_time.
20884 (estimate_local_effects): Update use of
20885 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
20886 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
20887 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
20888 Return nonspecialized time.
20890 2017-05-04 Richard Biener <rguenther@suse.de>
20892 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
20893 for the last VUSE which def dominates the PHI. Directly call
20895 (get_continuation_for_phi_1): Remove.
20897 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
20899 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
20900 to explain the use of truncating division. Cap the number of
20901 iterations to the maximum given by nb_iterations_upper_bound,
20904 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20906 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
20907 * configure: Regenerate.
20908 * config.in: Regenerate.
20909 * config/i386/driver-mingw32.c: new file.
20910 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
20911 * config.host: Link driver-mingw32.o on MinGW host.
20912 * doc/install.texi: Document new --enable-mingw-wildcard configure
20915 2017-05-04 Marek Polacek <polacek@redhat.com>
20917 PR tree-optimization/80612
20918 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
20920 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20921 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
20923 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
20924 (movt splitter): Likewise.
20925 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
20926 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
20927 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
20928 block for Thumb-1 with MOVT.
20929 (thumb2_legitimate_address_p): Move code block ...
20930 (can_avoid_literal_pool_for_label_p): ... into this new function.
20931 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
20933 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
20934 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
20935 "M-profile targets with the MOVT instruction".
20937 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20939 * config/arm/arm-builtins.c (arm_init_builtins): Rename
20940 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
20941 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
20943 2017-05-04 Martin Liska <mliska@suse.cz>
20945 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
20946 variable cond_code.
20948 2017-05-04 Richard Biener <rguenther@suse.de>
20950 * tree.c (array_at_struct_end_p): Handle arrays at struct
20951 end with flexarrays more conservatively. Refactor and treat
20952 arrays of arrays or aggregates more strict. Fix
20953 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
20954 * tree.c (array_at_struct_end_p): Adjust prototype.
20955 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
20956 * gimple-fold.c (get_range_strlen): Likewise.
20957 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
20959 2017-05-04 Richard Biener <rguenther@suse.de>
20961 PR tree-optimization/31130
20962 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
20964 (supports_overflow_infinity): Likewise.
20965 (is_negative_overflow_infinity): Likewise.
20966 (is_positive_overflow_infinity): Likewise.
20967 (is_overflow_infinity): Likewise.
20968 (stmt_overflow_infinity): Likewise.
20969 (overflow_infinity_range_p): Likewise.
20970 (usable_range_p): Remove as always returning true.
20971 (make_overflow_infinity): Remove.
20972 (negative_overflow_infinity): Likewise.
20973 (positive_overflow_infinity): Likewise.
20974 (avoid_overflow_infinity): Likewise.
20975 (set_value_range): Adjust accordingly.
20976 (set_value_range_to_nonnegative): Likewise, remove now unused
20977 overflow_infinity arg.
20978 (vrp_operand_equal_p): Adjust.
20979 (update_value_range): Likewise.
20980 (range_int_cst_singleton_p): Likewise.
20981 (operand_less_p): Likewise.
20982 (compare_values_warnv): Likewise.
20983 (extract_range_for_var_from_comparison_expr): Likewise.
20984 (vrp_int_const_binop): Likewise.
20985 (zero_nonzero_bits_from_vr): Likewise.
20986 (extract_range_from_multiplicative_op_1): Likewise.
20987 (extract_range_from_binary_expr_1): Likewise.
20988 (extract_range_from_unary_expr): Likewise.
20989 (extract_range_from_comparison): Likewise.
20990 (extract_range_basic): Likewise.
20991 (adjust_range_with_scev): Likewise.
20992 (compare_ranges): Likewise.
20993 (compare_range_with_value): Likewise.
20994 (dump_value_range): Likewise.
20995 (test_for_singularity): Likewise, remove strict_overflow_p parameter
20997 (simplify_cond_using_ranges): Adjust.
20999 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21001 * brig-builtins.def: Added a builtin for class_f64.
21002 * builtin-types.def: Added a builtin type needed by class_f64.
21004 2017-05-03 Jason Merrill <jason@redhat.com>
21006 * timevar.def: Add TV_CONSTEXPR.
21008 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21010 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
21012 2017-05-03 Martin Jambor <mjambor@suse.cz>
21014 * ipa-prop.c (ipa_update_after_lto_read): Removed.
21015 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
21016 * ipa-cp.c (ipcp_propagate_stage): Do not call
21017 ipa_update_after_lto_read.
21018 * ipa-inline.c (ipa_inline): Likewise.
21020 2017-05-03 Martin Jambor <mjambor@suse.cz>
21022 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
21023 tag. Added a default constructor and a destructor.
21024 (ipa_edge_args_sum_t): New class;
21025 (ipa_edge_args_sum): Declare.
21026 (ipa_edge_args_vector): Remove declaration.
21027 (IPA_EDGE_REF): Use ipa_edge_args_sum.
21028 (ipa_free_edge_args_substructures): Remove declaration.
21029 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
21030 (ipa_edge_args_info_available_for_edge_p): Likewise.
21031 * ipa-prop.c (ipa_edge_args_vector): Removed.
21032 (edge_removal_hook_holder): Likewise.
21033 (edge_duplication_hook_holder): Likewise.
21034 (ipa_edge_args_sum): New variable.
21035 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
21036 ipa_edge_args_vector.
21037 (ipa_free_edge_args_substructures): Likewise.
21038 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
21039 ipa_edge_args_vector.
21040 (ipa_edge_removal_hook): Turned into method
21041 ipa_edge_args_sum_t::remove.
21042 (ipa_edge_duplication_hook): Turned into method
21043 ipa_edge_args_sum_t::duplicate.
21044 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
21045 registering edge hooks.
21046 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
21047 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
21048 ipa_edge_args_sum instead of ipa_edge_args_vector.
21049 * ipa-profile.c (ipa_profile): Likewise.
21051 2017-05-03 Martin Jambor <mjambor@suse.cz>
21053 * symbol-summary.h (function_summary): New method exists.
21054 (function_summary::symtab_removal): Deallocate through release.
21055 (call_summary): New class.
21056 (gt_ggc_mx): New overload.
21057 (gt_pch_nx): Likewise.
21058 (gt_pch_nx): Likewise.
21060 2017-05-03 Jeff Law <law@redhat.com>
21062 PR tree-optimization/78496
21063 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
21064 from simplify_cond_using_ranges. Split off code to walk
21065 backwards through casts into ...
21066 (simplify_cond_using_ranges_2): New function.
21067 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
21068 (execute_vrp): After identifying jump threads, call
21069 simplify_cond_using_ranges_2.
21071 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
21074 * ipa-inline.h (inline_summary): Add ctor.
21075 (create_ggc): Do not use ggc_cleared_alloc.
21077 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
21078 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21080 * gcc.c (handle_braces): Support escaping in switch matching
21082 * doc/invoke.texi (Spec Files): Document it.
21083 Remove superfluous @code markup in items.
21085 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21087 * diagnostic-show-locus.c (struct column_range): New struct.
21088 (get_affected_columns): New function.
21089 (get_printed_columns): New function.
21090 (struct correction): New struct.
21091 (correction::ensure_capacity): New function.
21092 (correction::ensure_terminated): New function.
21093 (struct line_corrections): New struct.
21094 (line_corrections::~line_corrections): New dtor.
21095 (line_corrections::add_hint): New function.
21096 (layout::print_trailing_fixits): Reimplement in terms of the new
21098 (selftest::test_overlapped_fixit_printing): New function.
21099 (selftest::diagnostic_show_locus_c_tests): Call it.
21101 2017-05-03 Nathan Sidwell <nathan@acm.org>
21103 Canonicalize canonical type hashing
21104 * tree.h (type_hash_canon_hash): Declare.
21105 * tree.c (type_hash_list, attribute_hash_list): Move into
21106 type_hash_canon_hash.
21107 (build_type_attribute_qual_variant): Break out hash code calc into
21108 type_hash_canon_hash.
21109 (type_hash_canon_hash): New. Generic type hash computation.
21110 (build_range_type_1, build_array_type_1, build_function_type,
21111 build_method_type_directly, build_offset_type, build_complex_type,
21112 make_vector_type): Call it.
21114 2017-05-03 Richard Biener <rguenther@suse.de>
21116 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
21117 When all DRs have unknown misaligned do not always peel
21118 when there is a store but apply the same costing model as if
21119 there were only loads.
21121 2017-05-03 Richard Biener <rguenther@suse.de>
21124 PR tree-optimization/80492
21125 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21126 compare_base_decls returning dont-know properly.
21128 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
21130 * config/arm/iterators.md (CCSI): New mode iterator.
21131 (arch): New mode attribute.
21132 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
21133 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
21134 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
21135 code iterator for success result mode.
21136 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
21137 the corresponding new insn generators.
21139 2017-05-03 Bin Cheng <bin.cheng@arm.com>
21142 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21143 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21145 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
21147 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
21148 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
21149 (DDR_A): Wrap DDR argument in brackets.
21150 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
21151 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
21152 (DDR_REVERSED_P): Likewise.
21154 2017-05-03 Jakub Jelinek <jakub@redhat.com>
21156 PR tree-optimization/79472
21157 * tree-switch-conversion.c (struct switch_conv_info): Add
21158 contiguous_range and default_case_nonstandard fields.
21159 (collect_switch_conv_info): Compute contiguous_range and
21160 default_case_nonstandard fields, don't clear final_bb if
21161 contiguous_range and only the default case doesn't have the required
21163 (check_all_empty_except_final): Set default_case_nonstandard instead
21164 of failing if contiguous_range and the default case doesn't have empty
21166 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
21167 and only the default case doesn't have the required constants. Skip
21169 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
21170 if default_case_nonstandard.
21171 (build_constructors): Build constant 1 just once. Assert that default
21172 values aren't inserted in between cases if contiguous_range. Skip
21174 (build_arrays): Skip virtual phis.
21175 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
21176 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
21177 Handle virtual phis.
21178 (gen_inbound_check): Handle default_case_nonstandard case.
21179 (process_switch): Adjust check_final_bb caller. Call
21180 gather_default_values with the first non-default case instead of
21181 default case if default_case_nonstandard.
21183 2017-05-02 Nathan Sidwell <nathan@acm.org>
21185 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
21186 check. Fix formatting.
21188 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
21190 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
21191 errors when comparing specialized and unspecialized times.
21193 2017-05-02 David Malcolm <dmalcolm@redhat.com>
21195 * diagnostic-show-locus.c
21196 (layout::should_print_annotation_line_p): Make private.
21197 (layout::print_annotation_line): Make private.
21198 (layout::annotation_line_showed_range_p): Make private.
21199 (layout::show_ruler): Make private.
21200 (layout::print_source_line): Make private. Pass in line and
21201 line_width, rather than calling location_get_source_line. Drop
21203 (layout::print_leading_fixits): New method.
21204 (layout::print_any_fixits): Rename to...
21205 (layout::print_trailing_fixits): ...this, and make private.
21206 Don't print newline fixits.
21207 (diagnostic_show_locus): Move logic for printing one row into...
21208 (layout::print_line): ...this new function. Move the
21209 location_get_source_line call and error-handling from
21210 print_source_line to here. Call print_leading_fixits, and rename
21211 print_any_fixits to print_trailing_fixits.
21212 (selftest::test_fixit_insert_containing_newline): Update now that
21213 newlines are partially supported.
21214 (selftest::test_fixit_insert_containing_newline_2): New test.
21215 (selftest::test_fixit_replace_containing_newline): Update comments.
21216 (selftest::diagnostic_show_locus_c_tests): Call the new test.
21217 * edit-context.c (class added_line): New class.
21218 (class edited_line): Describe newline handling in comment.
21219 (edited_line::actually_edited_p): New method.
21220 (edited_line::print_content): Delete redundant decl.
21221 (edited_line::m_predecessors): New field.
21222 (edited_file::print_content): Call edited_line::print_content.
21223 (edited_file::print_diff): Update to support newlines.
21224 (edited_file::print_diff_hunk): Likewise.
21225 (edited_file::print_run_of_changed_lines): New function.
21226 (edited_file::print_diff_line): Convert to...
21227 (print_diff_line): ...this.
21228 (edited_file::get_effective_line_count): New function.
21229 (edited_line::edited_line): Initialize new field m_predecessors.
21230 (edited_line::~edited_line): Clean up m_predecessors.
21231 (edited_line::apply_fixit): Handle newlines.
21232 (edited_line::get_effective_line_count): New function.
21233 (edited_line::print_content): New function.
21234 (edited_line::print_diff_lines): New function.
21235 (selftest::test_applying_fixits_insert_containing_newline): New
21237 (selftest::test_applying_fixits_replace_containing_newline): New
21239 (selftest::insert_line): New function.
21240 (selftest::test_applying_fixits_multiple_lines): Add example of
21242 (selftest::edit_context_c_tests): Call the new tests.
21244 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21246 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
21247 parameter cand. Update dump information.
21248 (get_computation_cost): Update uses.
21250 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21252 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
21253 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
21254 (get_computation_at, rewrite_use_address): Update use of
21255 get_computation_aff.
21257 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21259 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
21260 (get_computation): Delete.
21261 (get_computation_cost): Implement like get_computation_cost_at.
21262 Use get_computation_at.
21263 (get_computation_cost_at): Delete.
21264 (rewrite_use_nonlinear_expr): Use get_computation_at.
21265 (rewrite_use_compare, remove_unused_ivs): Ditto.
21267 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21269 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
21271 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21273 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
21274 (ivopts_global_cost_for_size): Rename parameter and update uses.
21275 (iv_ca_recount_cost): Update uses.
21276 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
21277 candidates seperately in n_invs and n_cands.
21278 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
21280 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21282 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
21283 (find_inv_vars_cb): New.
21284 (find_depends): Renamed to ...
21285 (find_inv_vars): ... this.
21286 (add_candidate_1, force_var_cost): Call find_inv_vars.
21287 (split_address_cost, determine_group_iv_cost_cond): Ditto.
21289 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21291 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
21292 inv_vars. Add inv_exprs.
21293 (struct iv_cand): Rename depends_on to inv_vars.
21294 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
21295 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
21296 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
21297 (dump_cand): Dump inv_vars.
21298 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
21299 (record_invariant, find_depends, add_candidate_1): Ditto.
21300 (set_group_iv_cost, force_var_cost): Ditto.
21301 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
21302 (get_computation_cost_at, get_computation_cost): Ditto.
21303 (determine_group_iv_cost_generic): Ditto.
21304 (determine_group_iv_cost_address): Ditto.
21305 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
21306 (determine_group_iv_costs): Ditto.
21307 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
21308 (iv_ca_set_remove_invariants): Renamed to ...
21309 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
21310 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
21311 (iv_ca_set_add_invariants): Renamed to ...
21312 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
21313 (iv_ca_set_cp): Use iv_ca_set_add_invs.
21314 (iv_ca_has_deps): Support inv_vars and inv_exprs.
21315 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
21316 (create_new_ivs): Remove useless dump.
21318 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21320 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
21322 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
21323 (iv_ca_set_no_cp, create_new_iv): Ditto.
21325 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21327 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21329 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21331 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
21332 function tree_check2.
21334 2017-05-02 Martin Liska <mliska@suse.cz>
21336 * doc/gcov.texi: Add missing preposition.
21337 * gcov.c (function_info::function_info): Properly fill up
21338 all member variables.
21340 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21342 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
21344 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21346 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
21348 2017-05-02 Martin Liska <mliska@suse.cz>
21351 * lto-streamer-in.c (lto_read_tree_1): Remove
21352 LTO_STREAMER_DEBUG.
21353 * lto-streamer.c (struct tree_hash_entry): Likewise.
21354 (struct tree_entry_hasher): Likewise.
21355 (tree_entry_hasher::hash): Likewise.
21356 (tree_entry_hasher::equal): Likewise.
21357 (lto_streamer_init): Likewise.
21358 (lto_orig_address_map): Likewise.
21359 (lto_orig_address_get): Likewise.
21360 (lto_orig_address_remove): Likewise.
21361 * lto-streamer.h: Likewise.
21362 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
21363 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
21365 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
21367 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
21368 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
21369 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
21370 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
21371 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
21372 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
21373 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
21374 (mm_maskz_sub_ss): New intrinsics.
21375 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21376 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21377 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
21378 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
21379 (__builtin_ia32_subss_mask_round): New builtins.
21380 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21381 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21382 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
21384 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
21385 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21387 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21390 2017-05-02 Martin Jambor <mjambor@suse.cz>
21392 PR tree-optimization/78687
21393 * tree-sra.c (access): New field parent.
21394 (process_subtree_disqualification): New function.
21395 (disqualify_candidate): Call it.
21396 (build_accesses_from_assign): Reset write flag if creating an
21398 (build_access_subtree): Fill in parent field and also prpagate
21399 down grp_write flag.
21400 (create_artificial_child_access): New parameter set_grp_write, set
21401 grp_write to its value.
21402 (propagate_subaccesses_across_link): Also propagate grp_write flag
21404 (propagate_all_subaccesses): Push the closest parent back to work
21405 queue if add_access_to_work_queue returned true.
21407 2017-05-02 Richard Biener <rguenther@suse.de>
21409 * common.opt (fstrict-overflow): Alias negative to fwrapv.
21410 * doc/invoke.texi (fstrict-overflow): Remove all traces of
21411 -fstrict-overflow documentation.
21412 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
21413 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
21414 flag_strict_overflow.
21415 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
21416 * lto-opts.c (lto_write_options): Do not stream it.
21417 * lto-wrapper.c (merge_and_complain): Do not handle it.
21418 * opts.c (default_options_table): Do not set -fstrict-overflow.
21419 (finish_options): Likewise do not clear it when sanitizing.
21420 * simplify-rtx.c (simplify_const_relational_operation): Do not
21421 test flag_strict_overflow.
21423 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21425 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
21426 using enabled attribute.
21427 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
21428 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
21429 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
21430 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
21431 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
21432 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
21433 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
21434 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
21435 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
21436 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
21438 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21440 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
21442 2017-05-02 Richard Biener <rguenther@suse.de>
21444 PR tree-optimization/80591
21446 2017-04-10 Richard Biener <rguenther@suse.de>
21448 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21451 2017-05-02 Richard Biener <rguenther@suse.de>
21453 PR tree-optimization/80549
21454 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
21455 (cleanup_tree_cfg_noloop): Create forwarders to known loop
21456 headers if they do not have a preheader.
21458 2017-05-02 Martin Liska <mliska@suse.cz>
21461 * common.opt: Fix typo.
21462 * doc/invoke.texi: Likewise.
21464 2017-05-01 Jan Beulich <jbeulich@suse.com>
21466 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
21467 swapping, add (x,x,m,x,n) alternative.
21469 2017-05-01 Nathan Sidwell <nathan@acm.org>
21471 * calls.c (combine_pending_stack_adjustment_and_call): Remove
21472 unnecessary unadjusted_alignment check.
21474 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21477 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
21479 * gimplify.c (gimplify_cilk_detach): New function.
21480 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
21481 * tree-core.h: Document EXPR_CILK_SPAWN.
21482 * tree.h (EXPR_CILK_SPAWN): Define.
21484 2017-05-01 David Malcolm <dmalcolm@redhat.com>
21486 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
21487 to use new fixit_hint representation, using the "replace" logic.
21488 (get_line_span_for_fixit_hint): Likewise.
21489 (layout::print_any_fixits): Likewise.
21490 (selftest::test_one_liner_many_fixits): Rename to...
21491 (selftest::test_one_liner_many_fixits_1): ...this, and update
21492 comment and expected output to reflect that the multiple fix-it
21493 hints are now consolidated into one insertion.
21494 (selftest::test_one_liner_many_fixits_2): New test.
21495 (selftest::test_diagnostic_show_locus_one_liner): Update for
21497 (selftest::test_fixit_consolidation): Update for fix-it API
21499 * diagnostic.c (print_parseable_fixits): Likewise.
21500 * edit-context.c (edited_line::m_line_events): Convert from
21501 auto_vec <line_event *> to auto_vec <line_event>.
21502 (class line_event): Convert from abstract base class to a concrete
21503 class, taking over the role of replace_event.
21504 (class insert_event): Delete.
21505 (class replace_event): Rename to class line_event. Convert to
21507 (edit_context::add_fixits): Reimplement.
21508 (edit_context::apply_insert): Delete.
21509 (edit_context::apply_replace): Rename to...
21510 (edit_context::apply_fixit): ...this. Convert to half-open range.
21511 (edited_file::apply_insert): Delete.
21512 (edited_file::apply_replace): Rename to...
21513 (edited_file::apply_fixit): ...this.
21514 (edited_line::~edited_line): Drop deletion of events.
21515 (edited_line::apply_insert): Delete.
21516 (edited_line::apply_replace): Rename to...
21517 (edited_line::apply_fixit): ...this. Convert to half-open range.
21518 Update for change to type of m_line_events.
21519 * edit-context.h (edit_context::apply_insert): Delete.
21520 (edit_context::apply_replace): Rename to...
21521 (edit_context::apply_fixit): ...this.
21523 2017-05-01 Martin Sebor <msebor@redhat.com>
21525 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
21528 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
21531 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
21532 __get_cpuid_max returns 0.
21533 (__get_cpuid_count): Ditto.
21535 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
21537 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
21538 replacement expression is another instance of one of its arguments.
21540 2017-05-01 Jakub Jelinek <jakub@redhat.com>
21543 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
21544 check for stack push/pop autoinc.
21545 * config/i386/i386.c (ix86_agi_dependent): Return false
21546 if the only reason why modified_in_p returned true is that
21547 addr is SP based and set_insn is a push or pop.
21549 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
21551 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
21554 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
21557 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
21558 (account_size_time): Use two predicates - exec_pred and
21560 (evaluate_conditions_for_known_args): Compute both clause and
21562 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
21563 (inline_summary_t::duplicate): Update.
21564 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
21566 (compute_inline_parameters): Likewise.
21567 (estimate_edge_size_and_time): Update caluclation of time.
21568 (estimate_node_size_and_time): Compute both time and nonspecialized
21570 (estimate_ipcp_clone_size_and_time): Update.
21571 (inline_merge_summary): Update.
21572 (do_estimate_edge_time): Update.
21573 (do_estimate_edge_size): Update.
21574 (do_estimate_edge_hints): Update.
21575 (inline_read_section, inline_write_summary): Stream both new predicates.
21576 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
21578 (compute_inlined_call_time): Cleanup.
21579 (big_speedup_p): Update.
21580 (edge_badness): Update.
21581 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
21582 (size_time_entry): Replace predicate by exec_predicate and
21583 nonconst_predicate.
21584 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
21585 (estimate_edge_time): Return also nonspec_time.
21586 (reset_edge_growth_cache): Update.
21588 2017-04-29 Jakub Jelinek <jakub@redhat.com>
21590 PR rtl-optimization/80491
21591 * ifcvt.c (noce_process_if_block): When looking for x setter
21592 with missing else_bb, don't check only the insn right before
21593 cond_earliest, but look for the last insn that x is modified in
21594 within the same bb.
21596 PR rtl-optimization/80491
21597 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
21599 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
21601 PR tree-optimization/80487
21602 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
21604 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21606 PR tree-optimization/79697
21607 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
21608 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
21609 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
21611 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
21612 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
21614 2017-04-28 Martin Sebor <msebor@redhat.com>
21616 PR tree-optimization/80523
21617 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
21618 (init_target_to_host_charmap, target_to_host, target_strtol10): New
21620 (maybe_warn, format_directive, parse_directive): Use new functions.
21621 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
21623 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
21625 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
21627 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
21629 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
21630 target_header_dir): Set correctly.
21631 * configure: Regenerated.
21632 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21633 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21634 instead of SYSTEM_HEADER_DIR.
21636 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
21638 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
21639 (estimate_local_effects): Likewise.
21640 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
21641 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
21642 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
21643 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
21644 do_estimate_edge_time, estimate_edge_time): Likewise.
21645 * ipa-inline-analysis.c (estimate_node_size_and_time,
21646 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
21647 (estimate_time_after_inlining): Remove.
21649 2017-04-28 Martin Liska <mliska@suse.cz>
21651 * doc/gcov.texi: Enhance documentation of gcov.
21653 2017-04-28 Martin Liska <mliska@suse.cz>
21655 * doc/gcov.texi: Sort options in alphabetic order.
21656 * doc/gcov-dump.texi: Likewise.
21657 * doc/gcov-tool.texi: Likewise.
21658 * gcov.c (print_usage): Likewise.
21659 * gcov-dump.c (print_usage): Likewise.
21660 * gcov-tool.c (print_merge_usage_message): Likewise.
21661 (print_rewrite_usage_message): Likewise.
21662 (print_overlap_usage_message): Likewise.
21664 2017-04-28 Martin Liska <mliska@suse.cz>
21666 PR gcov-profile/53915
21667 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
21669 2017-04-28 Martin Liska <mliska@suse.cz>
21671 PR gcov-profile/79891
21672 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
21673 is marked by compiler as living on a line.
21674 (get_cycles_count): Remove usage of the union.
21675 (output_intermediate_file): Likewise.
21676 (find_source): Fix GNU coding style.
21677 (accumulate_line_counts): Remove old non-all block mode.
21678 (output_lines): Remove usage of the union.
21679 * profile.c (output_location): Include all BBs, even if
21680 belonging to a same line (and file) as a previous BB.
21682 2017-04-28 Martin Liska <mliska@suse.cz>
21684 * gcov.c (process_args): Handle new argument 'w'.
21685 (read_graph_file): Assign ID to BBs.
21686 (output_branch_count): Display BB # if verbose flag is set.
21687 (output_lines): Likewise for arcs.
21688 (print_usage): Add '--verbose' option help.
21689 * doc/gcov.texi: Document --verbose (-w) option.
21691 2017-04-28 Martin Liska <mliska@suse.cz>
21693 * gcov.c (struct block_location_info): New struct.
21694 (process_file): Fill up the new structure.
21695 (read_graph_file): Replace usage of encoding by the newly added
21697 (add_line_counts): Likewise.
21698 (accumulate_line_counts): Remove usage of the union.
21699 (function_info::function_info): New function.
21700 (function_info::~function_info): Likewise.
21701 (process_file): Call delete instead of release_function.
21702 (release_function): Release the function.
21703 (release_structures): Call delete instead of release_function.
21704 (solve_flow_graph): Replace usage of num_blocks.
21705 (find_exception_blocks): Likewise.
21706 (output_lines): Fix GNU coding style.
21708 2017-04-28 Martin Liska <mliska@suse.cz>
21711 * coverage.c (coverage_remove_note_file): New function.
21712 * coverage.h: Declare the function.
21713 * toplev.c (finalize): Clean if an error has been seen.
21715 2017-04-28 Martin Liska <mliska@suse.cz>
21717 PR gcov-profile/80031
21718 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
21719 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
21720 * gcov.c (read_graph_file): Read just number of blocks.
21721 * profile.c (branch_prob): Do not stream 0 flags per a basic
21724 2017-04-28 Martin Liska <mliska@suse.cz>
21726 * gcov-dump.c (tag_*): Add new argument to declarations.
21727 (dump_gcov_file): Likewise.
21728 (tag_blocks): Add and use new argument depth.
21729 (tag_arcs): Likewise.
21730 (tag_lines): Likewise.
21731 (tag_counters): Likewise.
21732 (tag_summary): Likewise.
21733 (dump_working_sets): Use depth to do a proper indentation.
21735 2017-04-28 Jakub Jelinek <jakub@redhat.com>
21738 * cgraph.h (symtab_node::debug_symtab): No longer inline.
21739 * symtab.c (symtab_node::debug_symtab): Move definition here.
21741 2017-04-28 Richard Biener <rguenther@suse.de>
21743 * lto-streamer.h (LTO_major_version): Bump to 7.
21745 2017-04-28 Richard Biener <rguenther@suse.de>
21747 * tree-vrp.c (assert_info): New struct.
21748 (add_assert_info): New helper.
21749 (register_edge_assert_for_2): Refactor to add asserts to a vector
21751 (register_edge_assert_for_1): Likewise.
21752 (register_edge_assert_for): Likewise.
21753 (finish_register_edge_assert_for): New helper actually registering
21754 asserts where live on edge.
21755 (find_conditional_asserts): Adjust.
21756 (find_switch_asserts): Likewise.
21757 (evrp_dom_walker::try_find_new_range): Generalize.
21758 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
21760 2017-04-27 Marek Polacek <polacek@redhat.com>
21763 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
21764 arg10 and arg11 to itype.
21766 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
21768 * doc/extend.texi (Object Size Checking): Improve grammar.
21770 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
21773 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
21774 that the logic for permitting reciprocal estimates matches that
21777 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21780 * tree.c (type_cache_hasher::equal): Only compare
21781 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
21782 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
21783 non-aggregate element types.
21784 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
21785 about the flag on ARRAY_TYPEs in the comment, formatting fix.
21787 2017-04-27 Richard Biener <rguenther@suse.de>
21789 PR middle-end/80533
21790 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
21791 stripping ARRAY_REFs from MEM_EXPR make sure we're not
21792 keeping a reference to a trailing array.
21794 2017-04-27 Richard Biener <rguenther@suse.de>
21796 PR middle-end/80539
21797 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
21798 being in loop-closed SSA form conservatively.
21799 (chrec_fold_multiply_poly_poly): Likewise.
21801 2017-04-27 Tamar Christina <tamar.christina@arm.com>
21803 PR middle-end/79665
21804 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
21805 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
21807 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21810 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
21811 (aarch64_function_arg_alignment): Return unsigned int again, but still
21812 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
21813 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21814 Don't emit -Wpsabi note.
21815 (aarch64_function_arg_boundary): Likewise.
21816 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21819 2017-04-26 Nathan Sidwell <nathan@acm.org>
21821 * tree.h (crc32_unsigned_n): Declare.
21822 (crc32_unsigned, crc32_unsigned): Make inline.
21823 * tree.c (crc32_unsigned_bits): Replace with ...
21824 (crc32_unsigned_n): ... this.
21825 (crc32_unsigned, crc32_byte): Remove.
21826 (crc32_string): Remove unnecessary braces.
21828 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21830 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
21831 * ipa-inline-analysis.c (MAX_TIME): Remove.
21832 (account_size_time): Use sreal for time.
21833 (dump_inline_summary): Update.
21834 (estimate_function_body_sizes): Update.
21835 (estimate_edge_size_and_time): Update.
21836 (estimate_calls_size_and_time): Update.
21837 (estimate_node_size_and_time): Update.
21838 (inline_merge_summary): Update.
21839 (inline_update_overall_summary): Update.
21840 (estimate_time_after_inlining): Update.
21841 (inline_read_section): Update.
21842 (inline_write_summary): Update.
21843 * ipa-inline.c (compute_uninlined_call_time): Update.
21844 (compute_inlined_call_time): Update.
21845 (recursive_inlining): Update.
21846 (inline_small_functions): Update.
21847 (dump_overall_stats): Update.
21848 * ipa-inline.h: Include sreal.h.
21849 (size_time_entry): Turn time to sreal.
21850 (inline_summary): Turn self_time nad time to sreal.
21852 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21854 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
21856 (sreal::stream_out, sreal::stream_in): New.
21857 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
21859 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21861 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
21864 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
21867 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
21869 (dimode_scalar_chain::compute_convert_gain): Ditto.
21870 (dimode_scalar_chain::make_vector_copies): Ditto.
21871 (dimode_scalar_chain::convert_reg): Ditto.
21872 (dimode_scalar_chain::convert_insn): Ditto.
21873 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
21874 (VI248_AVX512BW_1): New mode iterator.
21875 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
21876 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
21879 2017-04-25 Martin Sebor <msebor@redhat.com>
21881 PR tree-optimization/80497
21882 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
21883 constants are representable in HOST_WIDE_INT.
21884 (parse_directive): Ditto.
21886 2017-04-25 Martin Sebor <msebor@redhat.com>
21889 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
21890 (new_zero_array): Adjust signature.
21891 (dom_info::dom_init): Used unsigned rather that size_t.
21892 (dom_info::dom_info): Same.
21894 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21895 Jakub Jelinek <jakub@redhat.com>
21898 * config/arm/arm.c: Include gimple.h.
21899 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21900 returns negative, increment ncrn only if it returned positive.
21901 (arm_needs_doubleword_align): Return int instead of bool,
21902 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
21903 members, but if there is any such non-FIELD_DECL
21904 > PARM_BOUNDARY aligned decl, return -1 instead of false.
21905 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21906 returns negative, increment nregs only if it returned positive.
21907 (arm_setup_incoming_varargs): Likewise.
21908 (arm_function_arg_boundary): Emit -Wpsabi note if
21909 arm_needs_doubleword_align returns negative, return
21910 DOUBLEWORD_ALIGNMENT only if it returned positive.
21912 2017-04-25 Marek Polacek <polacek@redhat.com>
21915 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21916 first argument to type.
21918 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
21921 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
21922 type checks to test for compatibility instead of equality.
21924 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21925 Jakub Jelinek <jakub@redhat.com>
21928 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
21930 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
21931 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
21932 the alignment computation, but return their maximum in warn_alignment.
21933 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21934 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
21936 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
21937 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21940 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21942 * config/arc/simdext.md (dmpyh): Fix typo.
21944 2017-04-25 Richard Biener <rguenther@suse.de>
21946 PR tree-optimization/80492
21947 * alias.c (compare_base_decls): Handle registers with asm
21948 specification conservatively.
21949 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21950 compare_base_decls returning dont-know properly.
21952 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21954 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
21955 (legitimate_offset_address_p): New function.
21956 (arc_legitimate_address_p): Use above function.
21958 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21960 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
21962 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21964 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
21965 ACCH registers whenever they are available.
21967 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21969 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
21970 double regs fix when not used.
21972 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21974 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
21976 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
21977 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
21979 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21981 * config/arc/arc.c (arc_output_addsi): Check for h-register class
21982 when emitting short ADD instructions.
21984 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21986 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
21988 (cmpsi_cc_c_insn): Likewise.
21989 (cbranchsi4_scratch): Compute proper instruction length using
21990 compact_hreg_operand.
21991 * config/arc/predicates.md (compact_hreg_operand): New predicate.
21993 2017-04-25 Richard Biener <rguenther@suse.de>
21995 PR middle-end/80509
21996 * passes.c (pass_manager::pass_manager): Initialize
21997 m_name_to_pass_map.
21999 2017-04-25 Richard Biener <rguenther@suse.de>
22001 PR tree-optimization/79201
22002 * tree-ssa-sink.c (statement_sink_location): Handle calls.
22004 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22007 * config/s390/vector.md: Split MEM->GPR vector moves for
22008 non-s_operand addresses.
22010 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22013 * config/s390/predicates.md (reload_const_wide_int_operand): New
22015 * config/s390/s390.md ("movti"): Remove d/P alternative.
22016 ("movti_bigconst"): New pattern definition.
22018 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22021 * s390-protos.h (s390_expand_cs_hqi): Removed.
22022 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
22023 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
22024 modes as well as CCZ1mode and CCZmode.
22025 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
22026 signature of s390_emit_compare_and_swap.
22027 (s390_expand_cs_hqi): Likewise, make static.
22028 (s390_expand_cs_tdsi): Generate an explicit compare before trying
22029 compare-and-swap, in some cases.
22030 (s390_expand_cs): Wrapper function.
22031 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
22033 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
22034 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
22035 patterns for small and large integers. Forbid symref memory operands.
22036 Move expander to s390.c. Require cc register.
22037 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
22038 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
22039 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
22040 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
22041 symref memory operands. Remove CC mode and call s390_match_ccmode
22043 ("atomic_exchange<mode>"): Allow and implement all integer modes.
22045 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22047 * config/s390/s390.md (define_peephole2): New peephole to help
22048 combining the load-and-test pattern with volatile memory.
22050 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22052 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
22053 with CCZmode for TARGET_Z196.
22055 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22057 PR rtl-optimization/80501
22058 * combine.c (make_compound_operation_int): Set subreg_code to SET
22059 even for AND with mask of the sign bit of mode.
22061 PR rtl-optimization/80500
22062 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
22063 sum's initial value.
22065 2017-04-25 Julian Brown <julian@codesourcery.com>
22066 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22068 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
22070 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
22072 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
22074 2017-04-25 Julian Brown <julian@codesourcery.com>
22075 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22077 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
22078 (thunderx2t99_sha): New Reservation.
22080 2017-04-25 Julian Brown <julian@codesourcery.com>
22081 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22083 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
22084 type for 1-element load.
22086 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
22088 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
22090 2017-04-24 Martin Jambor <mjambor@suse.cz>
22092 PR tree-optimization/80293
22093 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
22094 char arrays not totally scalarizable if it is false.
22095 (analyze_all_variable_accesses): Pass correct value in the new
22096 parameter. Add a statistics counter.
22098 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
22100 PR middle-end/79931
22101 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
22103 2017-04-24 Richard Biener <rguenther@suse.de>
22105 PR tree-optimization/80494
22106 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
22107 out for complex types.
22109 2017-04-24 Richard Biener <rguenther@suse.de>
22111 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
22112 * tree-ssa-sccvn.c (print_scc): Print SCC size.
22113 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
22114 (DFS): Adjust and never fail.
22115 (sccvn_dom_walker::fail): Remove.
22116 (sccvn_dom_walker::before_dom_children): Adjust.
22117 (run_scc_vn): Likewise and never fail.
22118 * tree-ssa-pre.c (pass_pre::execute): Adjust.
22119 (pass_fre::execute): Likewise.
22121 2017-04-24 Richard Biener <rguenther@suse.de>
22123 PR tree-optimization/79725
22124 * tree-ssa-sink.c (statement_sink_location): Return whether
22125 failure reason was zero uses. Move that check later.
22126 (sink_code_in_bb): Deal with zero uses by removing the stmt
22129 2017-04-24 Richard Biener <rguenther@suse.de>
22132 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
22133 pointer-based references.
22135 2017-04-24 Richard Biener <rguenther@suse.de>
22138 * pass_manager.h (pass_manager::operator new): Remove.
22139 (pass_manager::operator delete): Likewise.
22140 * passes.c (pass_manager::operator new): Remove.
22141 (pass_manager::operator delete): Likewise.
22142 (pass_manager::pass_manager): Zero individual pass members.
22144 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
22147 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
22148 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
22149 Check "XEXP (src, 1)" operand here.
22150 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
22151 Check "XEXP (src, 1)" operand here.
22152 (dimode_scalar_chain::make_vector_copies): Detect count register
22153 of a shift instruction. Zero extend count register from QImode
22154 to DImode to satisfy vector shift pattern count operand predicate.
22155 Substitute vector shift count operand with a DImode copy.
22156 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
22159 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
22161 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
22162 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22163 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
22164 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22165 (UNSPEC_NOREX_MEM): Remove definition.
22167 2017-04-21 Richard Biener <rguenther@suse.de>
22169 PR tree-optimization/79547
22170 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22171 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
22172 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
22173 without any constraints.
22175 2017-04-21 Richard Biener <rguenther@suse.de>
22177 PR tree-optimization/78847
22178 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
22180 2017-04-21 Richard Biener <rguenther@suse.de>
22182 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
22183 (build_distinct_type_copy): Likewise.
22184 (build_variant_type_copy): Likewise.
22185 * tree.c (build_qualified_type): Pass down mem-stat info.
22186 (build_distinct_type_copy): Likewise.
22187 (build_variant_type_copy): Likewise.
22189 2017-04-21 Richard Biener <rguenther@suse.de>
22191 PR tree-optimization/80237
22192 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
22194 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
22195 for a simplified result.
22197 2016-04-21 Richard Biener <rguenther@suse.de>
22199 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
22200 sth as strict as a simple_iv but a chrec without symbols and an
22201 operand defined in the loop we are peeling (and not some subloop).
22202 (propagate_constants_for_unrolling): Propagate all constants.
22204 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22207 * config/i386/i386.c (print_reg): Remove assert for disalowed
22208 regno values, call output_operand_lossage instead.
22210 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22213 * config/i386/constraints.md (Yc): New register constraint.
22214 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
22215 Use Yc constraint for alternative 2 of operand 0. Remove
22216 preferred_for_speed attribute.
22218 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22220 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
22221 lastprivate clauses in SIMT case.
22223 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
22225 * doc/invoke.texi (-Wextra-semi): Document new warning option.
22227 2017-04-20 Richard Biener <rguenther@suse.de>
22229 PR tree-optimization/57796
22230 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
22231 as N scalar stores.
22232 (vect_model_load_cost): Cost gathers as N scalar loads.
22234 2017-04-20 Richard Biener <rguenther@suse.de>
22236 * ggc-page.c (ggc_allocated_p): Rename to ...
22237 (safe_lookup_page_table_entry): ... this and return the lookup
22239 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
22241 2017-04-20 Richard Biener <rguenther@suse.de>
22243 PR tree-optimization/80453
22244 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
22245 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
22246 from the conditions.
22247 (vn_phi_eq): Pass them down.
22248 (vn_phi_lookup): Record them.
22249 (vn_phi_insert): Likewise.
22251 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
22253 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
22254 uninitialized variable warning to avoid buffer overrun.
22256 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22259 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
22260 is suppressed for '{ 0 }' in C.
22262 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22264 * BASE-VER: Set to 8.0.0.
22266 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
22268 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
22269 priority .init_array and .fini_array section with SECTION_NOTYPE
22272 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22274 PR middle-end/80423
22275 * tree.h (build_array_type): Add typeless_storage default argument.
22276 * tree.c (type_cache_hasher::equal): Also compare
22277 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
22278 (build_array_type): Add typeless_storage argument, set
22279 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
22281 (build_nonshared_array_type): Adjust build_array_type_1 caller.
22282 (build_array_type): Likewise. Add typeless_storage argument.
22284 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22285 Jakub Jelinek <jakub@redhat.com>
22287 PR tree-optimization/80426
22288 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
22289 operation on symbolic operands, also compute the overflow for the
22290 invariant part when the operation degenerates into a negation.
22292 2017-04-19 Jakub Jelinek <jakub@redhat.com>
22295 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
22296 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
22299 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
22301 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
22304 * config/avr/avr.c (tree.h): Include it.
22305 (cgraph.h): Include it.
22306 (avr_encode_section_info): Don't warn for uninitialized progmem
22307 variable if it's just an alias.
22309 2017-04-19 Richard Biener <rguenther@suse.de>
22312 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
22313 when needed by AutoPGO.
22315 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
22318 * doc/lto.texi: Remove an extra 'that'.
22320 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
22322 PR rtl-optimization/80429
22323 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
22324 are only used in debug insns.
22326 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22327 Vladimir Makarov <vmakarov@redhat.com>
22329 * config/sparc/predicates.md (input_operand): Add comment. Return
22330 true for any memory operand when LRA is in progress.
22331 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
22333 2017-04-18 Jeff Law <law@redhat.com>
22336 * mips.md ({return,simple_return}_internal): Do not overwrite
22339 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22341 PR tree-optimization/80443
22342 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
22343 instead of adding 1, subtract -1 and similarly instead of subtracting
22346 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
22348 PR rtl-optimization/80357
22349 * haifa-sched.c (tmp_bitmap): New variable.
22350 (model_recompute): Handle duplicate use records.
22351 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
22352 (free_global_sched_pressure_data): Free it.
22354 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
22357 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22358 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22359 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22360 instead of SYSTEM_HEADER_DIR.
22362 2017-04-18 Jeff Law <law@redhat.com>
22364 PR middle-end/80422
22365 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
22366 predecessors after walking up the insn chain.
22368 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22371 * dwarf2out.c (modified_type_die): Try harder not to emit internal
22372 sizetype type into debug info.
22374 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22377 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
22378 unneeded test for TARGET_UPPER_REGS_SF.
22379 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
22381 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22384 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
22385 instead of gsi_after_labels.
22387 2017-04-18 Jeff Law <law@redhat.com>
22389 * regcprop.c (maybe_mode_change): Avoid creating copies of the
22393 2017-04-13 Jeff Law <law@redhat.com>
22394 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22395 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22397 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
22400 * config/avr/avr.c (intl.h): Include it.
22401 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
22403 2017-04-18 Martin Liska <mliska@suse.cz>
22405 PR gcov-profile/78783
22406 * gcov-tool.c (gcov_output_files): Validate that destination
22407 file is either removed by the tool or by a user.
22409 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22410 Guy Benyei <guybe@mellanox.com>
22412 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
22413 block, and do not negate it, the stored id is already negative.
22415 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22417 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
22419 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22422 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
22423 masks of options that should be turned off if the VSX vector
22424 options are turned off.
22425 (OTHER_P8_VECTOR_MASKS): Likewise.
22426 (OTHER_VSX_VECTOR_MASKS): Likewise.
22427 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
22428 rs6000_disable_incompatible_switches to validate no type switches
22430 (rs6000_incompatible_switch): New function to disallow turning on
22431 other vector options if -mno-vsx, -mno-power8-vector, or
22432 -mno-power9-vector are specified.
22434 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22436 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
22438 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22440 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
22441 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
22442 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
22443 (ARG_POINTER_CFA_OFFSET): Likewise.
22445 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22447 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
22448 conditions to take advantage of various optimizations.
22450 2017-04-13 Jeff Law <law@redhat.com>
22452 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22453 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22454 (zero_extendsidi2_dext): Likewise.
22456 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22459 * fold-const.c (fold_ternary_loc): Revert
22460 use op0 instead of fold_convert_loc (loc, type, arg0) part of
22463 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
22465 PR rtl-optimization/80343
22466 * lra-remat.c (update_scratch_ops): Assign original hard reg to
22467 new scratch pseudo.
22469 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
22472 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
22473 to ubsan_encode_value.
22475 2017-04-13 Jeff Law <law@redhat.com>
22477 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
22478 appearing in DEBUG_INSNs.
22480 2017-04-13 Martin Liska <mliska@suse.cz>
22482 PR gcov-profile/80413
22483 * gcov-io.c (gcov_write_string): Copy to buffer just when
22484 allocated size is greater than zero.
22486 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22489 * dwarf2out.c (decls_for_scope): Ignore declarations of
22490 current_function_decl in BLOCK_NONLOCALIZED_VARS.
22492 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
22495 * ipa-visibility.c (non_local_p): Fix typos.
22496 (localize_node): When localizing symbol in same comdat group,
22497 dissolve the group only when we know external symbols are going
22499 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
22501 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22503 PR tree-optimization/79390
22504 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
22505 order does not result in usable sequence, retry with reversed operand
22511 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
22512 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
22513 op0 instead of fold_convert_loc (loc, type, arg0).
22515 2017-04-12 Jeff Law <law@redhat.com>
22517 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
22518 has a delay slot in the generated code.
22520 * config/cris/cris.md (cris_preferred_reload_class): Return
22521 GENNONACR_REGS rather than GENERAL_REGS.
22523 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22526 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
22527 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
22528 signedness of the result type.
22530 2017-04-12 Richard Biener <rguenther@suse.de>
22531 Jeff Law <law@redhat.com>
22533 PR tree-optimization/80359
22534 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
22535 trim stores to TARGET_MEM_REFs.
22537 2017-04-12 Richard Biener <rguenther@suse.de>
22539 PR tree-optimization/79390
22540 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
22541 threading case even more.
22543 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
22546 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
22547 for quad_address_p for TImode, instead of just not indexed_address.
22549 2017-04-12 Richard Biener <rguenther@suse.de>
22550 Bernd Edlinger <bernd.edlinger@hotmail.de>
22552 PR middle-end/79671
22553 * alias.c (component_uses_parent_alias_set_from): Handle
22554 TYPE_TYPELESS_STORAGE.
22555 (get_alias_set): Likewise.
22556 * tree-core.h (tree_type_common): Add typeless_storage flag.
22557 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
22558 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
22559 for types containing members with TYPE_TYPELESS_STORAGE.
22560 (place_field): Likewise.
22561 (layout_type): Likewise for ARRAY_TYPE.
22562 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
22563 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
22564 TYPE_TYPELESS_STORAGE.
22565 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
22567 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22570 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22571 first argument to type.
22573 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22577 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
22578 CONST0_RTX (mode) rather than const0_rtx where appropriate.
22579 (rs6000_expand_binop_builtin): Likewise.
22580 (rs6000_expand_ternop_builtin): Likewise; also add missing
22581 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
22582 vshasigma built-ins.
22583 * doc/extend.texi: Document that vec_xxpermdi's third argument
22584 must be a constant.
22586 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
22588 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
22589 Use shift_const cost parameter when calculating gain of STV shifts.
22591 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
22593 PR rtl-optimization/70478
22594 * lra-constraints.c (process_alt_operands): Check memory for
22595 disfavoring memory insn operand.
22597 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22599 PR middle-end/80100
22600 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
22601 left shift in unsigned HOST_WIDE_INT type.
22603 PR rtl-optimization/80385
22604 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
22605 (not (neg X)) into (plus X -1) for complex or non-integral modes.
22608 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
22609 if they have any depend clauses.
22611 2017-04-11 Martin Liska <mliska@suse.cz>
22614 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
22615 * ipa-split.c (split_function): Create a local comdat symbol
22616 if caller is in a comdat group.
22618 2017-04-11 Martin Liska <mliska@suse.cz>
22621 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
22624 2017-04-11 Martin Sebor <msebor@redhat.com>
22626 PR middle-end/80364
22627 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
22628 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
22630 (directive::set_width, directive::set_precision, format_character):
22632 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
22635 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
22638 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
22639 conflict, set target->arch_name instead of target->cpu_name.
22641 2017-04-11 Richard Biener <rguenther@suse.de>
22643 PR tree-optimization/80374
22644 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
22645 build_zero_cst, remove fold_convertible_p check again.
22647 2017-04-11 Martin Liska <mliska@suse.cz>
22650 * ubsan.c (instrument_object_size): Do not instrument register
22653 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22656 * config/i386/i386-builtin-types.def
22657 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
22658 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
22659 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
22660 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
22661 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
22662 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
22663 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
22664 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
22665 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
22666 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
22667 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
22668 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
22669 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
22670 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
22671 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
22672 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
22673 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
22674 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
22675 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
22676 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
22677 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
22678 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
22679 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
22680 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
22681 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
22682 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
22683 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
22684 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
22685 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
22686 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
22687 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
22688 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
22689 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
22690 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
22691 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
22692 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
22693 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
22694 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
22695 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
22696 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
22697 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
22698 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
22699 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
22700 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
22701 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
22702 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
22704 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
22705 flag to second_arg_count, handle 4 argument function type _COUNT
22706 aliases, handle second_arg_count on second argument rather than last.
22708 2017-04-10 Jeff Law <law@redhat.com>
22710 PR tree-optimization/80374
22711 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
22712 record anything if we can not convert integer_zero_node to the
22715 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22718 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22719 Enhance special handling given to the TARGET_P9_MINMAX option in
22720 relation to certain other options.
22722 2017-04-10 Bin Cheng <bin.cheng@arm.com>
22724 PR tree-optimization/80153
22725 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
22726 remove POINTER_PLUS_EXPR's base part directly, rather than through
22729 2017-04-10 Richard Biener <rguenther@suse.de>
22730 Bin Cheng <bin.cheng@arm.com>
22732 PR tree-optimization/80153
22733 * tree-affine.c (aff_combination_to_tree): Get base pointer from
22734 the first element of pointer type aff_tree. Build result expr in
22736 (add_elt_to_tree): Convert to type unconditionally. Remove other
22737 fold_convert calls.
22738 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
22739 (rewrite_use_nonlinear_expr): Check invariant using iv information.
22741 2017-04-10 Richard Biener <rguenther@suse.de>
22743 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22746 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
22748 PR rtl-optimization/70478
22749 * lra-constraints.c (curr_small_class_check): New.
22750 (update_and_check_small_class_inputs): New.
22751 (process_alt_operands): Update curr_small_class_check. Disfavor
22752 alternative insn memory operands. Check available regs for small
22755 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22758 * config/mips/mips.opt (-mvirt): Update description.
22759 * doc/invoke.texi (-mvirt): Likewise.
22761 2017-04-10 Richard Biener <rguenther@suse.de>
22763 PR middle-end/80362
22764 * fold-const.c (fold_binary_loc): Look at unstripped ops when
22765 looking for NEGATE_EXPR in -A / -B to A / B folding.
22767 2017-04-10 Martin Liska <mliska@suse.cz>
22769 PR gcov-profile/80224
22770 * gcov.c (print_usage): Fix usage string.
22771 (get_gcov_intermediate_filename): Remove.
22772 (output_gcov_file): Use both for normal and intermediate format.
22773 (generate_results): Do not initialize special file for
22774 intermediate format.
22776 2017-04-10 Richard Biener <rguenther@suse.de>
22778 PR tree-optimization/80304
22779 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
22782 2017-04-10 Nathan Sidwell <nathan@acm.org>
22785 * config/rs6000/rs6000.c (rs6000_vector_type): New.
22786 (rs6000_init_builtins): Use it.
22788 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22790 * config/arm/arm.md (<mrc>): Add mode to SET source.
22791 (<mrrc>): Likewise.
22793 2017-04-10 Richard Biener <rguenther@suse.de>
22795 PR middle-end/80344
22796 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
22798 2017-04-10 Jakub Jelinek <jakub@redhat.com>
22801 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
22802 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
22803 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
22804 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
22805 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
22806 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
22807 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
22808 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
22809 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
22810 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
22811 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
22812 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
22813 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
22814 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
22815 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
22816 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
22817 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
22818 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
22819 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
22820 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
22821 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
22822 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
22823 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
22825 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
22827 PR rtl-optimization/70478
22828 * lra-constraints.c: Reverse the last patch.
22830 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
22832 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
22833 Add comment for WCHAR_T.
22835 2017-04-08 Martin Liska <mliska@suse.cz>
22838 2017-04-07 Martin Liska <mliska@suse.cz>
22841 * ipa-split.c (split_function): Add function part to a same comdat
22844 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22847 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
22849 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
22851 * rs6000/rs6000.c (vec_load_pendulum): Rename...
22852 (vec_pairing): ...to this.
22853 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
22854 (rs6000_sched_init): Adjust for name change.
22855 (struct rs6000_sched_context): Likewise.
22856 (rs6000_init_sched_context): Likewise.
22857 (rs6000_set_sched_context): Likewise.
22859 2017-04-07 Jakub Jelinek <jakub@redhat.com>
22865 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
22867 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
22868 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
22869 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
22871 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
22873 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
22875 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22877 PR rtl-optimization/70703
22878 * ira-color.c (update_conflict_hard_regno_costs): Use
22879 int64_t instead of HOST_WIDE_INT.
22881 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22883 PR rtl-optimization/70478
22884 * lra-constraints.c (process_alt_operands): Disfavor alternative
22885 insn memory operands.
22887 2017-04-07 Jeff Law <law@redhat.com>
22889 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
22890 CALL and NOTE_INSN_CALL_ARG_LOCATION.
22892 2017-04-07 Martin Liska <mliska@suse.cz>
22895 * config/aarch64/aarch64.c (aarch64_process_target_attr):
22896 Show error message instead of an ICE.
22898 2017-04-07 Martin Liska <mliska@suse.cz>
22901 * ipa-split.c (split_function): Add function part to a same comdat
22904 2017-04-07 Richard Biener <rguenther@suse.de>
22906 PR middle-end/80341
22907 * tree.c (get_unwidened): Also handle ! for_type case for
22909 * convert.c (do_narrow): Split out from ...
22910 (convert_to_integer_1): ... here. Do not pass final truncation
22911 type to get_unwidened for TRUNC_DIV_EXPR.
22913 2017-04-07 Richard Biener <rguenther@suse.de>
22915 * tree-affine.c (wide_int_ext_for_comb): Take type rather
22917 (aff_combination_const): Adjust.
22918 (aff_combination_scale): Likewise.
22919 (aff_combination_add_elt): Likewise.
22920 (aff_combination_add_cst): Likewise.
22921 (aff_combination_convert): Likewise.
22922 (add_elt_to_tree): Likewise. Remove unused argument.
22923 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
22925 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
22927 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
22929 * config/arm/arm.c (arm_default_short_enums): Use
22930 ARM_DEFAULT_SHORT_ENUMS.
22931 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
22933 2017-04-06 Jakub Jelinek <jakub@redhat.com>
22936 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
22937 members with redundant out-of-class redeclaration.
22939 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22942 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
22943 * config/i386/i386.md (*zero_extendsidi2):
22944 Add (?*x,*x) and (?*v,*v) alternatives.
22946 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22949 * config/i386/i386.c (ix86_expand_builtin)
22950 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
22951 mode from insn data. Convert operands to insn operand mode.
22952 Copy operands that don't satisfy insn predicate to a register.
22954 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
22956 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
22959 2017-04-06 Richard Biener <rguenther@suse.de>
22961 PR tree-optimization/80334
22962 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
22963 preserve alignment of accesses.
22965 2017-04-06 Richard Biener <rguenther@suse.de>
22967 PR tree-optimization/80262
22968 * tree-sra.c (build_ref_for_offset): Preserve address-space
22970 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
22971 Drop useless address-space information on MEM_REF offsets.
22973 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
22975 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
22977 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22979 PR rtl-optimization/70703
22980 * ira-color.c (update_conflict_hard_regno_costs): Use
22981 HOST_WIDE_INT instead of long.
22983 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
22986 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
22987 not defined for x86_64 target. Add -mmmx target option when __SSE2__
22989 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
22990 for x86_64 target. Handle -m3dnowa option.
22992 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22994 PR rtl-optimization/70703
22995 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
22996 (update_conflict_hard_regno_costs): Use long instead of unsigned
22997 arithmetic for cost calculation.
22999 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23000 Bernd Edlinger <bernd.edlinger@hotmail.de>
23003 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
23006 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
23009 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
23010 ptr_mode with Pmode throughout.
23011 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
23012 into probe_stack_range and use DImode.
23014 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
23017 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
23018 call_eh_return is true.
23020 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23022 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
23023 Initialize last_match_fntype_index.
23025 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23028 * tree-nvr.c: Include internal-fn.h.
23029 (pass_return_slot::execute): Ignore internal calls without
23032 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23033 Richard Biener <rguenther@suse.de>
23036 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
23037 captures used multiple times, except for the last use.
23038 * generic-match-head.c: Include gimplify.h.
23040 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23042 PR tree-optimization/79390
23043 * target.h (struct noce_if_info): Declare.
23044 * targhooks.h (default_noce_conversion_profitable_p): Declare.
23045 * target.def (noce_conversion_profitable_p): New target hook.
23046 * ifcvt.h (struct noce_if_info): New type, moved from ...
23047 * ifcvt.c (struct noce_if_info): ... here.
23048 (noce_conversion_profitable_p): Renamed to ...
23049 (default_noce_conversion_profitable_p): ... this. No longer
23051 (noce_try_store_flag_constants, noce_try_addcc,
23052 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
23053 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
23054 instead of noce_conversion_profitable_p.
23055 * config/i386/i386.c: Include ifcvt.h.
23056 (ix86_option_override_internal): Don't override
23057 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
23058 (ix86_noce_conversion_profitable_p): New function.
23059 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
23060 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
23061 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
23062 * doc/tm.texi: Regenerated.
23064 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23066 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
23069 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23072 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
23073 instructions for small multiply cores.
23075 2017-04-04 Jeff Law <law@redhat.com>
23077 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
23079 (mips_expand_vec_perm_const): Initialize elements in orig_perm
23080 that are not set by the loop over the elements.
23082 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23085 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
23086 int mode, convert_modes it to mode as unsigned, otherwise use
23087 lowpart_subreg to mode rather than SImode.
23088 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
23089 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
23090 Use DImode instead of SImode for the shift count operand.
23091 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
23094 2017-04-04 Richard Biener <rguenther@suse.de>
23096 PR middle-end/80281
23097 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
23098 arithmetic done for the negate or the plus. Simplify.
23099 (A - (-B) -> A + B): Likewise.
23100 * fold-const.c (split_tree): Make sure to not negate pointers.
23102 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
23104 PR rtl-optimization/60818
23105 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
23106 a compare of comparisons with the thing compared if this results
23107 in a different machine mode.
23109 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
23111 * alias.c (base_alias_check): Fix typo in comment.
23112 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
23113 * cgraphunit.c (symbol_table::compile): Likewise.
23114 * collect2.c (maybe_run_lto_and_relink): Likewise.
23115 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
23116 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
23117 * config/avr/avr.c (avr_map_op_t): Likewise.
23118 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
23119 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
23120 * config/epiphany/epiphany.md (movcc): Likewise.
23121 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
23122 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
23124 * config/mips/mips.c (mips_save_restore_reg): Likewise.
23125 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
23126 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
23127 * config/sh/sh.c (sh_rtx_costs): Likewise.
23128 * fold-const.c (fold_truth_andor): Likewise.
23129 * genautomata.c (collapse_flag): Likewise.
23130 * gengtype.h (struct type::u::s): Likewise.
23131 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
23132 * input.c (FORMAT_AMOUNT): Likewise.
23133 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
23134 (known_aggs_to_agg_replacement_list): Likewise.
23135 * ipa-inline-analysis.c: Likewise.
23136 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
23137 * ipa-polymorphic-call.c
23138 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
23139 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
23140 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
23142 * modulo-sched.c (apply_reg_moves): Likewise.
23143 * omp-expand.c (build_omp_regions_1): Likewise.
23144 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
23145 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
23146 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
23147 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
23148 * value-prof.c: Likewise.
23149 * var-tracking.c (val_reset): Likewise.
23151 2017-04-03 Richard Biener <rguenther@suse.de>
23153 PR tree-optimization/80275
23154 * fold-const.c (split_address_to_core_and_offset): Handle
23157 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
23159 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
23160 descriptors is at least equal to that of functions.
23162 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23164 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
23166 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23169 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
23170 (mov<IMOD4:mode>): New expander.
23171 (*mov<IMOD4:mode>_internal): New insn and split pattern.
23173 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
23175 PR rtl-optimization/79405
23176 * fwprop.c (propagations_left): New variable.
23177 (forward_propagate_into): Decrement it.
23178 (fwprop_init): Initialize it.
23179 (fw_prop): If the variable has reached zero, stop propagating.
23180 (fwprop_addr): Ditto.
23182 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23185 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
23186 a FUNCTION_DECL, pass it as decl instead of origin to
23189 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
23191 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
23194 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
23197 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
23198 TARGET_VSX_SMALL_INTEGER.
23200 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23202 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23203 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
23205 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23207 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
23208 extraction from odd-numbered MSA register.
23210 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23212 PR middle-end/80173
23213 * expmed.c (store_bit_field_1): Don't attempt to create
23214 a word subreg out of hard registers wider than word if they
23215 have HARD_REGNO_NREGS of 1 for their mode.
23217 PR middle-end/80163
23218 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
23219 conversions to integer types wider than word and pointer.
23222 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
23223 (rtx_equal_for_cselib_p): Pass 0 to it.
23224 * cselib.c (cselib_hasher::equal): Likewise.
23225 (rtx_equal_for_cselib_1): Add depth argument. If depth
23226 is 128, don't look up VALUE locs and punt. Increment
23227 depth in recursive calls when walking VALUE locs.
23229 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
23231 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
23232 (make_gcov_file_name): Use the canonical path name for generating
23234 (read_line): Fix handling of files with ascii null bytes.
23236 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
23238 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
23239 to initialise a vector register instead
23240 of using a const_int.
23242 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23244 PR translation/80189
23245 * gimplify.c (omp_default_clause): Use %qs instead of %s in
23246 diagnostic messages.
23248 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
23251 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
23252 (dfp_diex_<mode>): Update mode of operand 1.
23253 * doc/extend.texi (dxex, dxexq): Document change to return type.
23254 (diex, diexq): Document change to argument type.
23256 2017-03-30 Martin Jambor <mjambor@suse.cz>
23259 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
23260 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
23261 it reflects the signature changes performed at the callee side.
23262 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
23263 to cgraph_build_function_type_skip_args.
23264 (build_function_decl_skip_args): Adjust call to the above function.
23266 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23269 * config/i386/sse.md
23270 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
23271 register as dest whenever it is a MEM not rtx_equal_p to the
23272 corresponding dup operand, and when forcing into reg move the
23273 reg into the memory afterwards.
23274 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
23275 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
23276 for the force_reg mode.
23277 (avx512vl_vextractf128<mode>): Use register as dest either
23278 always when a MEM, or when it is a MEM not rtx_equal_p to the
23279 corresponding dup operand, or even not when it is a CONST_VECTOR
23280 depending on the mode and lo vs. hi.
23281 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
23283 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
23284 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
23285 Likewise. Require that operands[2] is even.
23286 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
23287 Remove extraneous parens. Require that operands[2] is a multiple
23289 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
23290 operands[0] is a MEM if <mask_applied>, the predicates/constraints
23291 disallow memory then.
23293 2017-03-30 Richard Biener <rguenther@suse.de>
23295 PR tree-optimization/77498
23296 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
23297 to non-constants over backedges.
23299 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
23301 PR rtl-optimization/80233
23302 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
23303 as last_combined_insn. Do not test for BARRIER_P separately.
23305 2017-03-29 Andreas Schwab <schwab@suse.de>
23308 * calls.c (prepare_call_address): Convert funexp to Pmode before
23309 copying to temp reg.
23311 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23313 PR tree-optimization/80158
23314 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
23315 Handle possible future case of more than one alternate
23317 (replace_rhs_if_not_dup): Likewise.
23318 (replace_one_candidate): Likewise.
23320 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
23322 PR rtl-optimization/80193
23323 * ira.c (ira): Do not check allocation for LRA.
23325 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
23327 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
23328 (nvptx_output_simt_exit): Declare.
23329 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
23330 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
23331 (init_softstack_frame): Move initialization of crtl->is_leaf to...
23332 (nvptx_declare_function_name): ...here. Emit declaration of local
23333 memory space buffer for omp_simt_enter insn.
23334 (nvptx_output_unisimt_switch): New.
23335 (nvptx_output_softstack_switch): New.
23336 (nvptx_output_simt_enter): New.
23337 (nvptx_output_simt_exit): New.
23338 * config/nvptx/nvptx.h (struct machine_function): New fields
23339 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
23340 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
23341 (UNSPECV_SIMT_EXIT): Ditto.
23342 (omp_simt_enter_insn): New insn.
23343 (omp_simt_enter): New expansion.
23344 (omp_simt_exit): New insn.
23345 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
23347 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
23348 (expand_GOMP_SIMT_ENTER_ALLOC): New.
23349 (expand_GOMP_SIMT_EXIT): New.
23350 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
23351 (GOMP_SIMT_ENTER_ALLOC): Ditto.
23352 (GOMP_SIMT_EXIT): Ditto.
23353 * target-insns.def (omp_simt_enter): New insn.
23354 (omp_simt_exit): Ditto.
23355 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
23357 (lower_rec_simd_input_clauses): Implement SIMT privatization.
23358 (lower_rec_input_clauses): Likewise.
23359 (lower_lastprivate_clauses): Handle SIMT privatization.
23361 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
23362 (ompdevlow_adjust_simt_enter): New.
23363 (find_simtpriv_var_op): New.
23364 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
23365 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
23367 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
23368 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
23369 (copy_decl_for_dup_finish): Ditto.
23371 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
23373 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
23376 * config/i386/i386.c (ix86_option_override_internal): Always
23377 allow -mpreferred-stack-boundary=3 for 64-bit targets.
23379 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23381 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
23383 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23385 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
23386 mark new edge's irreducible flag accordign to it.
23387 (vect_do_peeling): Check loop preheader edge's irreducible flag
23388 and pass it to function slpeel_add_loop_guard.
23390 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
23392 PR tree-optimization/80218
23393 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23394 Update block frequencies and counts.
23396 2017-03-28 Richard Biener <rguenther@suse.de>
23398 PR tree-optimization/78644
23399 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
23400 of a simplification result we may not use it at all.
23402 2017-03-28 Richard Biener <rguenther@suse.de>
23405 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
23406 without arguments, generate default definition of a SSA name.
23408 2017-03-28 Richard Biener <rguenther@suse.de>
23410 PR middle-end/80222
23411 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
23412 TYPE_REF_CAN_ALIAS_ALL references.
23413 * fold-const.c (fold_indirect_ref_1): Likewise.
23415 2017-03-28 Martin Liska <mliska@suse.cz>
23418 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
23419 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
23421 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23422 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
23424 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
23425 (EXTRA_SPECS): Define.
23426 (SUBTARGET_EXTRA_SPECS): Likewise.
23427 (SUBTARGET_CPP_SPEC): Likewise.
23428 * config/arc/elf.h (EXTRA_SPECS): Renamed to
23429 SUBTARGET_EXTRA_SPECS.
23430 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
23432 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23434 * config/arc/simdext.md (vst64_insn): Update pattern.
23435 (vld32wh_insn): Likewise.
23436 (vld32wl_insn): Likewise.
23437 (vld64_insn): Likewise.
23438 (vld32_insn): Likewise.
23440 2017-03-28 Marek Polacek <polacek@redhat.com>
23443 * fold-const.c (fold_comparison): Use protected_set_expr_location
23444 instead of SET_EXPR_LOCATION.
23446 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
23448 * tree.c (add_expr): Avoid name lookup warning.
23450 2017-03-27 Jeff Law <law@redhat.com>
23452 PR tree-optimization/80216
23453 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
23454 function name. Limit recursion depth.
23455 (record_temporary_equivalences): Corresponding changes.
23457 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
23459 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
23462 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23465 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
23467 * cfgcleanup.c (reg_note_cfa_p): New array.
23468 (insns_have_identical_cfa_notes): New function.
23469 (old_insns_match_p): Don't cross-jump in between /f
23470 and non-/f instructions. If both i1 and i2 are frame related,
23471 verify all CFA notes, their order and content.
23473 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
23476 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
23477 HImode and SImode with zero extend to DImode to one insn.
23478 (bswap<mode>2_extenddi): Likewise.
23479 (bswapsi2_extenddi): Likewise.
23480 (bswaphi2_extendsi): Likewise.
23481 (bswaphi2): Combine bswap HImode and SImode into one insn.
23482 Separate memory insns from swapping register.
23483 (bswapsi2): Likewise.
23484 (bswap<mode>2): Likewise.
23485 (bswaphi2_internal): Delete, no longer used.
23486 (bswapsi2_internal): Likewise.
23487 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
23488 store, and gpr<-gpr swap insns.
23489 (bswap<mode>2_store): Likewise.
23490 (bswaphi2_reg): Register only splitter, combine with the splitter.
23491 (bswaphi2 splitter): Likewise.
23492 (bswapsi2_reg): Likewise.
23493 (bswapsi2 splitter): Likewise.
23494 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
23495 the insns into load, store, and register/register insns.
23496 (bswapdi2_ldbrx): Likewise.
23497 (bswapdi2_load): Likewise.
23498 (bswapdi2_store): Likewise.
23499 (bswapdi2_reg): Likewise.
23501 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
23503 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
23504 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
23506 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
23509 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
23511 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23512 special handling for target option conflicts between dform
23513 options (-mpower9-dform, -mpower9-dform-vector,
23514 -mpower9-dform-scalar) and -mno-direct-move.
23516 2017-03-27 Richard Biener <rguenther@suse.de>
23518 PR tree-optimization/80181
23519 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
23521 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23523 * config/arc/predicates.md (move_double_src_operand): Replace the
23524 call to move_double_src_operand with a call to address_operand.
23526 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23528 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
23529 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
23530 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
23532 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23534 * config/arc/predicates.md (long_immediate_loadstore_operand):
23535 Consider scaled addresses cases.
23537 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23539 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
23540 restored when in interrupt.
23541 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
23542 doesn't have delay slot.
23544 2017-03-27 Richard Biener <rguenther@suse.de>
23547 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
23548 inlined thunk clones.
23550 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23553 * asan.c (instrument_derefs): Copy over last operand from
23554 original COMPONENT_REF to the new COMPONENT_REF with
23555 DECL_BIT_FIELD_REPRESENTATIVE.
23556 * ubsan.c (instrument_object_size): Likewise.
23558 2017-03-27 Richard Biener <rguenther@suse.de>
23560 PR tree-optimization/80170
23561 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
23562 sure DR/SCEV didnt fold in constants we do not see when looking
23563 at the reference base alignment.
23565 2017-03-27 Richard Biener <rguenther@suse.de>
23567 PR middle-end/80171
23568 * gimple-fold.c (fold_ctor_reference): Properly guard against
23569 NULL return value from canonicalize_constructor_val.
23571 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
23574 * config/i386/i386.c (ix86_expand_builtin)
23575 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
23576 flags reg setting and flags reg using instructions.
23577 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
23578 clobbering instructions to zero extend op2.
23580 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
23582 * doc/install.texi (Configuration) <--with-aix-soname>:
23583 Update link to AIX ld.
23585 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
23587 PR rtl-optimization/80160
23588 PR rtl-optimization/80159
23589 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
23590 reg_alternate_class into account.
23592 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
23595 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
23596 to consider in curr_insn_transform.
23598 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23600 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
23601 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
23602 and emit_mode_inner.
23604 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23606 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
23607 argument to the overloaded builtin variants. Use the new flag to
23608 deprecate certain builtin variants.
23609 * config/s390/s390-builtin-types.def: Add new builtin types.
23610 * config/s390/s390-builtins.h: Support new flags field for
23611 overloaded builtins.
23612 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
23613 (s390_macro_to_expand): Enable vector float data type.
23614 (s390_cpu_cpp_builtins_internal): Indicate support of the new
23615 builtins by incrementing the __VEC__ version number.
23616 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
23618 (s390_resolve_overloaded_builtin): Emit error messages depending
23619 on the builtin flags.
23620 * config/s390/s390.c (s390_expand_builtin): Support additional
23621 flags argument. Change error message to match the messages
23622 emitted in s390-c.c.
23623 * config/s390/s390.md: New UNSPEC_* constants.
23624 (op_type): Add new instruction types.
23625 * config/s390/vecintrin.h: Add new builtins and test data class
23627 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
23628 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
23629 (VEC_INEXACT, VEC_NOINEXACT): New constants.
23630 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
23631 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
23632 ("vec_mergel<mode>"): V_HW -> VEC_HW.
23634 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
23635 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
23636 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
23637 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
23639 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
23640 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
23641 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
23642 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
23644 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
23645 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
23646 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
23647 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
23648 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
23649 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
23650 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
23652 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
23653 ("vec_scatter_element<V_HW_4:mode>_DI")
23654 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
23655 ("vec_fpint<mode>", "vflls")
23656 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
23657 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
23658 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
23659 ("*vec_cmphe<mode>_cc"): ... these.
23661 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
23662 mode constant instead of magic value.
23664 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23666 * config/s390/s390.c (s390_expand_vec_compare): Support other
23667 vector floating point modes than just V2DF.
23668 (s390_expand_vcond): Likewise.
23669 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
23670 (s390_cannot_change_mode_class): Prevent mode changes between TF
23671 and V1TF in vector registers.
23672 * config/s390/s390.md (DF, SF): New mode attributes.
23673 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
23674 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
23675 SFmode support for VRs.
23676 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
23678 (VFT, VF_HW): New mode iterators.
23679 (vw, sdx): New mode attributes.
23680 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
23681 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
23682 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
23683 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
23684 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
23685 also the new vector floating point modes. Renaming to ...
23687 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
23688 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
23689 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
23690 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
23691 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
23692 ("vec_unordered<mode>"): ... these.
23694 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
23695 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
23696 ("*vec_extendv2df"): New insn definitions.
23698 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23700 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
23701 ("mulditi3_2", "*muldi3_sign"): New patterns.
23702 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
23703 rename the pattern definition.
23705 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23707 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
23709 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
23711 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23713 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
23714 instruction if possible.
23715 * config/s390/vector.md (vec_halfnumelts): New mode
23717 ("*vec_vllezlf<mode>"): New pattern.
23719 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23721 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
23722 ("popcountv4si2", "popcountv2di2"): Rename to ...
23723 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
23724 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
23726 ("popcount<mode>2_vxe"): New pattern.
23728 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23730 * common/config/s390/s390-common.c (processor_flags_table): Add
23732 * config.gcc: Add arch12.
23733 * config/s390/driver-native.c (s390_host_detect_local_cpu):
23734 Default to arch12 for unknown CPU model numbers.
23735 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
23736 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
23737 PROCESSOR_max sanity check.
23738 * config/s390/s390-opts.h (enum processor_type): Add
23740 * config/s390/s390.c (processor_table): Add arch12.
23741 (s390_expand_builtin): Add check for B_VXE flag.
23742 (s390_issue_rate): Add PROCESSOR_ARCH12.
23743 (s390_get_sched_attrmask): Likewise.
23744 (s390_get_unit_mask): Likewise.
23745 (s390_sched_score): Enable z13 scheduling for arch12.
23746 (s390_sched_reorder): Likewise.
23747 (s390_sched_variable_issue): Likewise.
23748 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
23750 (s390_tune_attr): Use z13 scheduling also for arch12.
23751 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
23752 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
23753 (TARGET_VXE_P): New macros.
23754 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
23755 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
23756 * config/s390/s390.opt: Add arch12 as processor_type.
23758 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23760 * config/s390/s390.md
23761 ("fixuns_truncdddi2", "fixuns_trunctddi2")
23762 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
23763 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
23765 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
23766 Rename expanders to ...
23768 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
23769 ("fixuns_truncdddi2_emu"): ... these.
23771 ("fixuns_trunc<mode>si2_emu"): New expander.
23773 ("*fixuns_truncdfdi2_z13"): Rename to ...
23774 ("*fixuns_truncdfdi2_vx"): ... this.
23776 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23778 * config/s390/2964.md: Remove the single element vector compare
23779 instructions which are no longer used.
23780 * config/s390/s390.c (s390_select_ccmode): Remove handling of
23782 (s390_canonicalize_comparison): Remove handling of DFmode
23784 (s390_expand_vec_compare_scalar): Remove function.
23785 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
23786 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
23788 ("*cmp<mode>_ccs"): Add wfcdb instruction.
23790 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23792 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
23794 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
23795 will anyway by matched by mov<mode>_64dfp.
23797 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23799 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
23800 vlef/vstef. Add missing operand to vleif.
23802 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23804 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
23805 pair for all vector types with 64 bit elements.
23806 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
23807 * config/s390/vector.md (V_HW_64): ... here.
23808 (V_128_NOSINGLE): New mode iterator.
23809 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
23810 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
23811 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
23812 ("*vec_load_pairv2di"): Change to ...
23813 ("*vec_load_pair<mode>"): ... this one.
23815 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23817 * config/s390/constraints.md: Add comments.
23818 (jKK): Reject element sizes > 8 bytes.
23819 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
23821 * config/s390/s390.md: Add the s_operand checks formerly in
23822 s390_split_ok_p to various splitters where they are still
23824 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
23825 for 128 bit vectors. Plus two splitters.
23827 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23829 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
23832 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23835 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
23836 error if the boundary argument is not constant.
23838 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23840 PR rtl-optimization/80112
23841 * loop-doloop.c (doloop_condition_get): Don't check condition
23842 if cmp isn't SET with IF_THEN_ELSE src.
23844 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23846 PR tree-optimization/80158
23847 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
23848 replacing a candidate statement, also replace it for the
23849 candidate's alternate interpretation.
23850 (replace_rhs_if_not_dup): Likewise.
23851 (replace_one_candidate): Likewise.
23853 2017-03-24 Richard Biener <rguenther@suse.de>
23855 PR tree-optimization/80167
23856 * graphite-isl-ast-to-gimple.c
23857 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
23859 (translate_isl_ast_to_gimple::get_rename): Likewise.
23861 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23863 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
23864 handling of certain combinations of target options, including the
23865 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
23866 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
23868 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23871 * config/arm/arm.md (*load_multiple): Add reload_completed to
23872 matching condition.
23874 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23875 Richard Biener <rguenth@suse.de>
23877 PR tree-optimization/79908
23878 PR tree-optimization/80136
23879 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23880 been cast away, gimplify_and_add suffices.
23882 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
23884 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
23886 2017-03-23 Richard Biener <rguenther@suse.de>
23888 PR tree-optimization/80032
23889 * gimplify.c (gimple_push_cleanup): Forced unconditional
23890 cleanups still have to go to the conditional_cleanups
23893 2017-03-22 Jakub Jelinek <jakub@redhat.com>
23895 PR tree-optimization/80072
23896 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
23898 (next_operand_entry_id): Change type to unsigned int.
23899 (sort_by_operand_rank): Make sure to return the right return value
23900 even if unsigned fields are bigger than INT_MAX.
23901 (struct oecount): Change cnt and id type to unsigned int.
23902 (oecount_hasher::equal): Formatting fix.
23903 (oecount_cmp): Make sure to return the right return value
23904 even if unsigned fields are bigger than INT_MAX.
23905 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
23908 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
23909 TREE_READONLY on result if writing it more than once.
23912 * doc/invoke.texi (-fsanitize=thread): Document that with
23913 -fnon-call-exceptions atomics are not able to throw
23917 * tsan.c: Include tree-eh.h.
23918 (instrument_builtin_call): Call maybe_clean_eh_stmt or
23919 maybe_clean_or_replace_eh_stmt where needed.
23920 (instrument_memory_accesses): Add cfg_changed argument.
23921 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
23922 if it returned true.
23923 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
23925 PR rtl-optimization/63191
23926 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
23927 wrapper function, moved the whole old content into ...
23928 (ix86_delegitimize_address_1): ... this. New inline function.
23929 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
23930 true as last argument instead of ix86_delegitimize_address.
23932 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23934 * config/aarch64/aarch64.c (generic_branch_cost): Copy
23935 cortexa57_branch_cost.
23937 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23939 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
23941 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23944 * doc/md.texi (Constraints): Document wA constraint.
23945 * config/rs6000/constraints.md (wA): New.
23946 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
23947 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
23948 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
23949 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
23951 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
23954 * gimplify.c (is_oacc_declared): New function.
23955 (oacc_default_clause): Use it to set default flags for acc declared
23956 variables inside parallel regions.
23957 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
23958 declared variables.
23959 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
23960 declare attribute to any decl as necessary.
23962 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
23965 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
23966 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
23967 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
23968 (arm_arch_lpae): This.
23969 * config/arm/arm.c (arm_arch7ve): Rename into ...
23970 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
23971 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
23974 2017-03-22 Martin Liska <mliska@suse.cz>
23977 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
23978 error message instead of an ICE.
23980 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23982 * doc/extend.texi (6.11 Additional Floating Types): Revise.
23984 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
23986 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23988 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23991 2017-03-21 Martin Sebor <msebor@redhat.com>
23993 * doc/extend.texi: Use "cannot" instead of "can't."
23994 * doc/hostconfig.texi: Same.
23995 * doc/install.texi: Same.
23996 * doc/invoke.texi: Same.
23997 * doc/loop.texi: Same.
23998 * doc/md.texi: Same.
23999 * doc/objc.texi: Same.
24000 * doc/rtl.texi: Same.
24001 * doc/tm.texi: Same.
24002 * doc/tm.texi.in: Same.
24003 * doc/trouble.texi: Same.
24005 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
24008 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
24009 (collect_checksum_attributes): Set it.
24010 (die_checksum_ordered): Use it.
24012 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24014 PR tree-optimization/79908
24015 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
24016 change: For a VA_ARG whose LHS has been cast away, use
24017 force_gimple_operand to construct the side effects.
24019 2017-03-21 David Malcolm <dmalcolm@redhat.com>
24021 PR translation/80001
24022 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
24023 more amenable to translation.
24024 (oacc_loop_auto_partitions): Likewise.
24026 2017-03-21 Marek Polacek <polacek@redhat.com>
24027 Martin Sebor <msebor@redhat.com>
24029 PR tree-optimization/80109
24030 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
24031 on INTEGRAL_TYPE_P.
24033 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24034 Segher Boessenkool <segher@kernel.crashing.org>
24037 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
24038 check reg_used_between_p between insn and one of succ or succ2
24039 depending on if succ is artificial insn not inserted into insn
24042 2017-03-21 Martin Liska <mliska@suse.cz>
24044 PR gcov-profile/80081
24045 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
24046 * doc/gcc.texi: Include gcov-dump stuff.
24047 * doc/gcov-dump.texi: New file.
24049 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
24051 PR rtl-optimization/79150
24052 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
24053 conditional jump, if the jump is the last insn of the loop.
24055 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24056 Richard Biener <rguenth@suse.de>
24058 PR tree-optimization/79908
24059 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24060 been cast away, use force_gimple_operand to construct the side
24063 2017-03-21 Martin Liska <mliska@suse.cz>
24065 PR libfortran/79956
24066 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
24069 2017-03-21 Brad Spengler <spender@grsecurity.net>
24072 * plugin.c (htab_hash_plugin): New function.
24073 (add_new_plugin): Use it and adjust.
24074 (parse_plugin_arg_opt): Adjust.
24075 (init_one_plugin): Likewise.
24077 2017-03-21 Richard Biener <rguenther@suse.de>
24079 PR tree-optimization/80032
24080 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
24081 if set force the cleanup to happen unconditionally.
24082 (gimplify_target_expr): Push inserted clobbers with force_uncond
24083 to avoid them being removed by control-dependent DCE.
24085 2017-03-21 Richard Biener <rguenther@suse.de>
24087 PR tree-optimization/80122
24088 * tree-inline.c (copy_bb): Do not expans va-arg packs or
24089 va_arg_pack_len when the inlined call stmt requires pack
24091 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
24093 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24096 * tsan.c (instrument_builtin_call): If the memory model argument
24097 is not a constant, assume it is valid.
24100 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
24103 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24105 PR rtl-optimization/79910
24106 * combine.c (can_combine_p): Do not allow combining an I0 or I1
24107 if its dest is used by an insn before I2 (other than the combined
24108 insns themselves, which are properly handled already).
24110 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24113 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24115 * combine.c (record_used_regs): New static function.
24116 (try_combine): Handle situations where there is an additional
24117 instruction between I2 and I3 which needs to have a LOG_LINK
24121 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24123 * combine.c (try_combine): Delete redundant i1 test. Call
24124 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24126 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24129 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
24130 alternatives 13/14.
24132 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24134 PR tree-optimization/80054
24135 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
24136 the optimization if a PHI or any of its arguments is not dominated
24137 by the candidate's basis. Use gphi* rather than gimple* as
24139 (replace_profitable_candidates): Clean up a gimple* variable that
24140 should be a gphi* variable.
24142 2017-03-20 Martin Sebor <msebor@redhat.com>
24145 * doc/extend.texi (attribute constructor): Document present limitation.
24147 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
24150 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
24151 __POWER9_VECTOR__ #ifdef control, change template definition to
24152 use Power9-specific built-in function.
24153 (vec_any_eq): Likewise.
24154 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
24155 to control outcomes from this test.
24156 (vector_ae_<mode>p): For VEC_F modes, likewise.
24158 2017-03-20 Ian Lance Taylor <iant@google.com>
24160 * config/i386/i386.c (ix86_function_regparm): Save an extra
24161 register for -fsplit-stack with DECL_STATIC_CHAIN.
24163 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24166 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
24167 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
24169 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24171 * config/riscv/riscv.c (riscv_print_operand): Use "fence
24173 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
24176 2017-03-20 Marek Polacek <polacek@redhat.com>
24179 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
24181 2017-03-20 Richard Biener <rguenther@suse.de>
24183 PR tree-optimization/80113
24184 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
24185 allocate extra SSA name for PHI def.
24186 (add_close_phis_to_outer_loops): Likewise.
24187 (add_close_phis_to_merge_points): Likewise.
24188 (copy_loop_close_phi_args): Likewise.
24189 (copy_cond_phi_nodes): Likewise.
24191 2017-03-20 Martin Liska <mliska@suse.cz>
24193 PR middle-end/79753
24194 * tree-chkp.c (chkp_build_returned_bound): Do not build
24195 returned bounds for a LHS that's not a BOUNDED_P type.
24197 2017-03-20 Martin Liska <mliska@suse.cz>
24201 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
24202 COMPLEX_CST and VECTOR_CST.
24204 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24207 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
24208 target operand. A new splitter adds the clobber statement in case
24209 the target operand is dead anyway.
24211 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
24213 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
24214 to age-old versions of binutils and glibc.
24216 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
24218 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
24220 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24222 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
24224 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24226 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
24227 requirement for binutils 2.13.
24229 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24231 * combine.c (try_combine): Delete redundant i1 test. Call
24232 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24234 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
24236 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
24237 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
24239 <riscv64-*-elf>: Re-arrange section
24240 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
24241 <riscv32-*-linux>: Likewise.
24242 <riscv64-*-elf>: Likewise
24243 <riscv64-*-linux>: Likewise.
24245 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
24248 * aarch64.opt(verbose-cost-dump): Fix typo.
24250 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
24253 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
24254 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
24256 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24258 * reload.c (find_reloads): When reloading a nonoffsettable address,
24259 use RELOAD_OTHER for it and its address reloads.
24261 PR rtl-optimization/79910
24262 * combine.c (record_used_regs): New static function.
24263 (try_combine): Handle situations where there is an additional
24264 instruction between I2 and I3 which needs to have a LOG_LINK
24267 2017-03-17 Jeff Law <law@redhat.com>
24269 PR tree-optimization/71437
24270 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
24271 conditional in the hash table first.
24272 (vrp_dom_walker::before_dom_children): Extract condition from
24273 ASSERT_EXPR. Record condition, its inverion and any implied
24274 conditions as well.
24276 2017-03-17 Marek Polacek <polacek@redhat.com>
24277 Markus Trippelsdorf <markus@trippelsdorf.de>
24279 PR tree-optimization/80079
24280 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
24283 2017-03-17 Richard Biener <rguenther@suse.de>
24285 PR middle-end/80075
24286 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
24287 Properly verify the LHS before the RHS possibly claims to be
24289 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
24292 2017-03-17 Martin Jambor <mjambor@suse.cz>
24294 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
24295 (List of -O2 options): Likewise.
24296 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
24299 2017-03-17 Tom de Vries <tom@codesourcery.com>
24301 * gcov-dump.c (print_usage): Print bug_report_url.
24303 2017-03-17 Richard Biener <rguenther@suse.de>
24305 PR middle-end/80050
24306 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
24307 (parser::peek): Likewise.
24309 2017-03-17 Richard Biener <rguenther@suse.de>
24311 PR tree-optimization/80048
24312 * sese.c (free_sese_info): Properly release rename_map and
24313 copied_bb_map elements.
24315 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
24317 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
24318 Add linked-list forward and backlinks. Insert on
24319 construction, remove on destruction.
24320 (class pass_store_merging): Add m_stores_head field.
24321 (pass_store_merging::terminate_and_process_all_chains):
24322 Iterate over m_stores_head list.
24323 (pass_store_merging::terminate_all_aliasing_chains):
24325 (pass_store_merging::execute): Check for debug stmts first.
24326 Push new chains onto the m_stores_head stack.
24328 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24331 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
24332 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
24333 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
24335 2017-03-16 Jeff Law <law@redhat.com>
24337 PR tree-optimization/71437
24338 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
24339 member function. Implementation moved into after_dom_children
24340 member function and into the threader's thread_outgoing_edges
24342 (dom_opt_dom_walker::after_dom_children): Simplify by moving
24343 some code into new thread_outgoing_edges.
24344 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
24345 definition. Simplify marker handling (do it here). Assume we always
24346 have the available expression and the const/copies tables.
24347 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
24349 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
24350 * tree-vrp.c (equiv_stack): No longer file scoped.
24351 (vrp_dom_walker): New class.
24352 (vrp_dom_walker::before_dom_children): New member function.
24353 (vrp_dom_walker::after_dom_children): Likewise.
24354 (identify_jump_threads): Setup domwalker. Use it rather than
24355 walking edges in a random order by hand. Simplify setup/finalization.
24356 (finalize_jump_threads): Remove.
24357 (vrp_finalize): Do not call identify_jump_threads here.
24358 (execute_vrp): Do it here instead and call thread_through_all_blocks
24361 PR tree-optimization/71437
24362 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
24364 (simplify_stmt_for_jump_threading): Add basic_block argument. All
24366 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
24367 (dom_opt_dom_walker::thread_across_edge): Remove
24368 handle_dominating_asserts argument. All callers changed.
24369 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
24370 changes. Remove calls to lhs_of_dominating_assert. Other
24371 uses of handle_dominating_asserts turn into unconditional code
24372 (simplify_control_stmt_condition_1): Likewise.
24373 (simplify_control_stmt_condition): Likewise.
24374 (thread_through_normal_block, thread_across_edge): Likewise.
24375 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
24376 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
24377 object if it is not an SSA_NAME.
24378 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
24379 before calling into the VRP specific simplifiers.
24380 (identify_jump_threads): Remove handle_dominating_asserts
24383 2017-03-16 Jakub Jelinek <jakub@redhat.com>
24386 * tree-diagnostic.c (default_tree_printer): No longer static.
24387 * tree-diagnostic.h (default_tree_printer): New prototype.
24389 2017-03-16 Tamar Christina <tamar.christina@arm.com>
24391 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
24392 Change ins into fmov.
24394 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24396 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
24397 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
24398 Use h_con constraint for operand 1.
24399 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
24400 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
24402 2017-03-15 Jeff Law <law@redhat.com>
24404 PR tree-optimization/71437
24405 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
24406 (record_temporary_equivalences): Use it.
24408 PR tree-optimization/71437
24409 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
24410 tree-ssa-scopedtables.
24411 (lookup_avail_expr, build_and_record_new_cond): Likewise.
24412 (record_conditions, record_cond, vuse_eq): Likewise.
24413 (record_edge_info): Adjust to API tweak of record_conditions.
24414 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
24415 (record_temporary_equivalences, optimize_stmt): Likewise.
24416 (eliminate_redundant_computations): Likewise.
24417 (record_equivalences_from_stmt): Likewise.
24418 * tree-ssa-scopedtables.c: Include options.h and params.h.
24419 (vuse_eq): New function, moved from tree-ssa-dom.c
24420 (build_and_record_new_cond): Likewise.
24421 (record_conditions): Likewise. Accept vector of conditions rather
24422 than edge_equivalence structure for first argument.
24423 for the first argument.
24424 (avail_exprs_stack::lookup_avail_expr): New member function, moved
24425 from tree-ssa-dom.c.
24426 (avail_exprs_stack::record_cond): Likewise.
24427 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
24428 from tree-ssa-dom.c.
24429 (avail_exprs_stack): Add new member functions lookup_avail_expr
24431 (record_conditions): Declare.
24433 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
24436 * lra-constraints.c (process_alt_operands): Increase reject for
24437 reloading an input/output operand.
24439 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
24442 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
24443 insns to convert from signed/unsigned char/short to IEEE 128-bit
24445 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
24447 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
24450 * config/i386/i386.c (ix86_vector_duplicate_value): Create
24451 subreg of inner mode for values already in registers.
24453 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
24455 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
24456 iteration reg is used after the loop.
24458 2017-03-14 Martin Sebor <msebor@redhat.com>
24460 PR tree-optimization/79800
24461 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
24462 precision in negative-positive range.
24463 (format_floating): Call non-const overload with adjusted precision.
24465 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24468 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
24471 2017-03-14 Martin Sebor <msebor@redhat.com>
24473 PR middle-end/80020
24474 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
24475 * builtins.def (aligned_alloc): Use it.
24478 * Makefile.in (GTFILES): Add calls.c.
24479 * calls.c: Include "gt-calls.h".
24481 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
24483 PR rtl-optimization/79728
24484 * regs.h (struct target_regs): New field
24485 x_contains_allocatable_regs_of_mode.
24486 (contains_allocatable_regs_of_mode): New macro.
24487 * reginfo.c (init_reg_sets_1): Initialize it, and change
24488 contains_reg_of_mode so it includes global regs as well.
24489 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
24490 rather than contains_regs_of_mode.
24492 2017-03-14 Martin Liska <mliska@suse.cz>
24494 * doc/invoke.texi: Document options that can't be combined with
24495 -fcheck-pointer-bounds.
24497 2017-03-14 Martin Liska <mliska@suse.cz>
24499 PR middle-end/79831
24500 * doc/invoke.texi (-Wchkp): Document the option.
24502 2017-03-14 Martin Liska <mliska@suse.cz>
24504 * Makefile.in: Install gcov-dump.
24506 2017-03-14 Martin Liska <mliska@suse.cz>
24508 * multiple_target.c (expand_target_clones): Bail out for
24509 an invalid attribute.
24511 2017-03-14 Richard Biener <rguenther@suse.de>
24513 * alias.c (struct alias_set_entry): Pack properly.
24514 * cfgloop.h (struct loop): Likewise.
24515 * cse.c (struct set): Likewise.
24516 * ipa-utils.c (struct searchc_env): Likewise.
24517 * loop-invariant.c (struct invariant): Likewise.
24518 * lra-remat.c (struct cand): Likewise.
24519 * recog.c (struct change_t): Likewise.
24520 * rtl.h (struct address_info): Likewise.
24521 * symbol-summary.h (function_summary): Likewise.
24522 * tree-loop-distribution.c (struct partition): Likewise.
24523 * tree-object-size.c (struct object_size_info): Likewise.
24524 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
24525 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
24526 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
24527 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
24528 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
24529 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
24530 (struct _stmt_vec_info): Likewise.
24532 2017-03-14 Martin Liska <mliska@suse.cz>
24535 * multiple_target.c (create_dispatcher_calls): Check that
24536 a target can create a function dispatcher.
24538 2017-03-14 Martin Liska <mliska@suse.cz>
24541 * multiple_target.c (expand_target_clones): Drop local.local
24542 flag for default implementation.
24544 2017-03-14 Richard Biener <rguenther@suse.de>
24546 PR tree-optimization/80030
24547 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
24549 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
24551 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
24552 gcc_fallthrough() instead of __attribute__((fallthrough));
24554 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24556 * doc/gcc.texi: Remove "up" link to (DIR).
24557 * doc/gccint.texi: Ditto.
24559 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24561 * doc/install.texi (Specific) <avr>: Remove reference to
24564 2017-03-13 Jeff Law <law@redhat.com>
24566 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
24567 attribute rather than comments.
24569 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
24570 match_scratch operand is highest.
24572 2017-03-13 Martin Liska <mliska@suse.cz>
24574 PR middle-end/78339
24575 * ipa-pure-const.c (warn_function_noreturn): If the declarations
24576 is a CHKP clone, use original declaration.
24578 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24580 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
24581 (arc_conditional_register_usage): Use a different allocation order
24582 when optimizing for size.
24583 * common/config/arc/arc-common.c (arc_option_optimization_table):
24584 Section anchors default on when optimizing for size.
24586 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24588 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
24590 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24592 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
24593 * config/arc/arc.md (cpu_facility): Add cd variant.
24594 (*movqi_insn): Add code density variant.
24595 (*movhi_insn): Likewise.
24596 (*movqi_insn): Likewise.
24597 (*addsi3_mixed): Likewise.
24598 (subsi3_insn): Likewise.
24600 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24602 * config/arc/arc.md (movsi_cond_exec): Update constraint.
24604 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24606 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
24607 expressions with MINUS and UNARY ops.
24609 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24612 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
24614 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
24615 between vec_select and vector argument.
24616 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
24617 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
24618 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
24619 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
24620 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
24621 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
24623 2017-03-13 Richard Biener <rguenther@suse.de>
24626 * params.def (vect-max-peeling-for-alignment): Fix typo.
24628 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24630 * doc/install.texi (Specific) <mips-*-*>: Remove description of
24631 issue that only occurred with binutils below 2.18.
24633 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24635 * doc/install.texi (Specific) <cris-axis-elf>: No longer
24636 refer to binutils 2.11/2.12 minimum.
24638 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24640 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
24641 ftp.kernel.org and simplify binutils requirement.
24643 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
24645 * doc/invoke.texi (Warning Options): Fix spelling of link-time
24647 (Optimize Options): Ditto. Also remove redundancy.
24649 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24651 PR translation/79848
24652 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
24654 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
24655 to G_ to avoid double translation.
24657 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24659 PR translation/79923
24660 * auto-profile.c (get_combined_location): Convert leading
24661 character of diagnostics to lower case and remove trailing period.
24662 (read_profile): Likewise for various diagnostics.
24663 * config/arm/arm.c (arm_option_override): Remove trailing period
24664 from various diagnostics.
24665 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
24666 (msp430_expand_delay_cycles): Likewise.
24668 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24671 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
24672 full command-line argument, rather than just "str".
24673 (aarch64_validate_march): Likewise.
24674 (aarch64_validate_mtune): Likewise.
24676 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
24678 PR rtl-optimization/78911
24679 * lra-assigns.c (must_not_spill_p): New function.
24680 (spill_for): Use it.
24682 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24684 PR tree-optimization/79981
24685 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
24686 ATOMIC_COMPARE_EXCHANGE ifn result.
24687 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
24688 IFN_ATOMIC_COMPARE_EXCHANGE.
24690 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24693 * opts.c (parse_sanitizer_options): Add missing question mark to
24694 "did you mean" message.
24696 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24698 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
24700 (VMULEUH_UNS): Likewise.
24701 (VMULOUB_UNS): Likewise.
24702 (VMULOUH_UNS): Likewise.
24703 * config/rs6000/rs6000.c (builtin_function_type): Remove
24704 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
24706 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24709 * read-rtl-function.c (function_reader::read_rtx_operand): Update
24710 x with result of extra_parsing_for_operand_code_0.
24711 (function_reader::extra_parsing_for_operand_code_0): Convert
24712 return type from void to rtx, returning x. When reading
24713 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
24714 larger size containing struct block_symbol.
24716 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
24718 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
24719 -mfloat128-hardware without -m64.
24721 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
24724 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
24725 entries to the case statement that marks unsigned arguments to
24726 overloaded functions.
24728 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24730 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
24731 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
24733 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
24736 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
24737 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
24739 2017-03-10 Martin Liska <mliska@suse.cz>
24743 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
24744 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
24747 2017-03-10 Olivier Hainque <hainque@adacore.com>
24749 * tree-switch-conversion (array_value_type): Start by resetting
24750 candidate type to it's main variant.
24752 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24754 PR rtl-optimization/79909
24755 * combine.c (try_combine): Use simplify_replace_rtx on individual
24756 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
24757 of the whole CALL_INSN_FUNCTION_USAGE.
24759 PR tree-optimization/79972
24760 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
24761 get_range_info on SSA_NAMEs. Formatting fixes.
24763 2017-03-10 Richard Biener <rguenther@suse.de>
24764 Jakub Jelinek <jakub@redhat.com>
24766 PR tree-optimization/77975
24767 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
24768 edge to be constant.
24769 (get_val_for): For constant x return it. Formatting fix.
24770 (loop_niter_by_eval): Avoid pointless looping if the next iteration
24771 would use the same bases as the current one.
24773 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24775 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
24776 instead of vec_select for V1TImode.
24777 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
24779 (VSX_LE_128): Add V1TI to this mode iterator.
24780 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
24781 (*vsx_le_perm_store_<mode>): Likewise.
24782 (pre-reload splitter for VSX stores): Likewise.
24783 (post-reload splitter for VSX stores): Likewise.
24784 (*vsx_xxpermdi2_le_<mode>): Likewise.
24785 (*vsx_lxvd2x2_le_<mode>): Likewise.
24786 (*vsx_stxvd2x2_le_<mode>): Likewise.
24788 2017-03-09 Michael Eager <eager@eagercon.com>
24790 Correct failures with --enable-checking=yes,rtl.
24792 * config/microblaze/microblaze.c (microblaze_expand_shift):
24793 Replace GET_CODE test with CONST_INT_P and INTVAL test with
24794 test for const0_rtx.
24795 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
24796 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
24798 2017-03-09 Richard Biener <rguenther@suse.de>
24800 PR tree-optimization/79977
24801 * graphite-scop-detection.c (scop_detection::merge_sese):
24802 Handle the case of extra exits to blocks dominating the entry.
24804 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
24806 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24809 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
24811 PR rtl-optimization/79949
24812 * lra-constraints.c (process_alt_operands): Check memory when
24813 trying to predict a cycle. Print about the overall increase.
24815 2017-03-09 Richard Biener <rguenther@suse.de>
24817 PR middle-end/79971
24818 * gimple-expr.c (useless_type_conversion_p): Preserve
24819 TYPE_SATURATING for fixed-point types.
24821 2017-03-09 Richard Biener <rguenther@suse.de>
24824 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
24825 alignment of BLKmode params.
24827 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24830 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
24831 (VALL_NO_V2Q): Likewise.
24833 * config/aarch64/aarch64-simd.md
24834 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
24836 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
24837 VALL_NO_V2Q mode iterator.
24838 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
24840 2017-03-09 Martin Liska <mliska@suse.cz>
24842 PR tree-optimization/79631
24843 * tree-chkp-opt.c (chkp_is_constant_addr): Call
24844 tree_int_cst_sign_bit just for INTEGER constants.
24846 2017-03-09 Martin Liska <mliska@suse.cz>
24850 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
24853 2017-03-09 Marek Polacek <polacek@redhat.com>
24856 * tree.c (inchash::add_expr): Handle TREE_VEC.
24858 2017-03-09 Martin Liska <mliska@suse.cz>
24861 (chkp_narrow_size_and_offset): New function.
24862 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
24863 (void chkp_parse_bit_field_ref): New function.
24864 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
24865 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
24867 2017-03-09 Martin Liska <mliska@suse.cz>
24870 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
24871 (chkp_find_bounds_1): Remove gcc_unreachable.
24873 2017-03-09 Jakub Jelinek <jakub@redhat.com>
24876 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
24877 BUILT_IN_SYNC*, determine the access type from the size suffix and
24878 always build a MEM_REF with that type. Handle forgotten
24879 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
24882 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
24883 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
24884 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
24885 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
24886 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
24887 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
24888 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
24889 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
24890 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
24891 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
24892 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
24893 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
24894 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
24895 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
24896 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
24897 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
24898 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
24899 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
24900 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
24901 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
24902 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
24903 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
24904 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
24905 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
24906 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
24907 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
24908 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
24909 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
24910 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
24911 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
24912 definitions outside of __OPTIMIZE__ guarded section.
24915 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
24916 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
24917 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
24918 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
24921 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24923 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
24924 ("vfenez<mode>"): Add missing constraints.
24926 2017-03-08 Martin Sebor <msebor@redhat.com>
24929 * config/nds32/nds32.c (nds32_option_override):
24930 Fix misspelled diagnostic.
24932 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24935 * gimplify.c (gimplify_omp_for): Replace index var in outer
24936 taskloop statement with an artificial variable and add
24937 OMP_CLAUSE_PRIVATE clause for it.
24939 2017-03-08 Richard Biener <rguenther@suse.de>
24941 PR tree-optimization/79955
24942 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
24943 for accesses that are completely outside of the variable.
24945 2017-03-08 Andrew Haley <aph@redhat.com>
24947 PR tree-optimization/79943
24948 * tree-ssa-loop-split.c (compute_new_first_bound): When
24949 calculating the new upper bound, (END-BEG) should be added, not
24952 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24954 * config/avr/avr.md (setmemhi): Make sure match_dup
24955 operand number comes before match_scratch.
24957 2017-03-08 Richard Biener <rguenther@suse.de>
24959 PR tree-optimization/79920
24960 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
24961 with ncopies == 1 to ...
24962 (vect_transform_slp_perm_load): ... here. Properly compute
24963 all element loads by iterating VF times over the group. Do
24964 not handle ncopies (computed in a broken way) in
24965 vect_create_mask_and_perm.
24967 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24970 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
24971 is a uniform vector, use uniform_vector_p return value instead of
24972 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
24974 2017-03-07 Marek Polacek <polacek@redhat.com>
24976 PR middle-end/79809
24977 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
24978 (alloca_call_type): Likewise.
24980 2017-03-07 Martin Liska <mliska@suse.cz>
24982 * gcov.c (process_args): Put comment to correct location.
24984 2017-03-07 Martin Liska <mliska@suse.cz>
24986 PR middle-end/68270
24987 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
24988 Use array_at_struct_end_p instead of DECL_CHAIN (field).
24989 (chkp_narrow_bounds_for_field): Likewise.
24990 (chkp_parse_array_and_component_ref): Pass one more argument to
24993 2017-03-07 Richard Biener <rguenther@suse.de>
24995 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
24998 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25000 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
25001 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
25003 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25006 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
25007 to end of description.
25008 (PARAM_MAX_STORES_TO_MERGE): Likewise.
25010 2017-03-07 Jakub Jelinek <jakub@redhat.com>
25012 PR rtl-optimization/79901
25013 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
25015 (*avx512f_<code><mode>3<mask_name>): ... this.
25016 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
25017 iterator instead of VI8_AVX2_AVX512BW.
25019 PR rtl-optimization/79901
25020 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
25021 min/max expander, expand it using expand_vec_cond_expr.
25024 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
25027 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25030 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
25031 to void * for PCH reasons.
25032 * dwarf2out.c (output_loc_operands, output_die): Cast
25033 v.val_vec.array to unsigned char *.
25035 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
25038 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
25041 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
25043 PR rtl-optimization/79571
25044 * lra-constraints.c (process_alt_operands): Calculate static
25045 reject and subtract it from overall when only addresses will be
25048 2017-03-06 Julia Koval <julia.koval@intel.com>
25051 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
25052 incoming stack boundary to 128 for 64-bit targets.
25054 2017-03-06 Richard Biener <rguenther@suse.de>
25056 PR tree-optimization/79894
25057 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
25058 to NULL after folding it.
25060 2017-03-06 Richard Biener <rguenther@suse.de>
25062 PR tree-optimization/79824
25063 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
25064 check disabling peeling for gaps.
25066 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
25068 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25069 attributes): Document gettimeofday.
25071 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25073 * config/s390/s390.c (s390_option_override_internal): Set
25074 PARAM_MIN_VECT_LOOP_BOUND
25076 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25078 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
25079 * config/s390/s390.md: Likewise.
25081 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25084 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
25085 (<avx2_avx512>_perm<mode>): Rename to ...
25086 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
25088 (<avx512>_perm<mode>_mask): Rename to ...
25089 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
25091 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
25092 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
25093 instead of VI8F_256_512.
25094 (avx512f_perm<mode>): New define_expand.
25095 (avx512f_perm<mode>_mask): Likewise.
25096 (avx512f_perm<mode>_1<mask_name>): New define_insn.
25097 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
25099 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25101 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
25102 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
25104 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
25106 2017-03-06 Martin Liska <mliska@suse.cz>
25109 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
25110 when having a SSA NAME w/o VAR_DECL assigned to it.
25112 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25114 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
25115 msa_dpsub_<su>_d): Fix MODE for vec_select.
25117 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25119 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
25121 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
25123 2017-03-06 Richard Biener <rguenther@suse.de>
25125 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
25126 * plugin.c (register_plugin_info): Likewise.
25127 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
25129 2017-03-05 Jakub Jelinek <jakub@redhat.com>
25131 * config/i386/sse.md (sse_storehps, sse_storelps,
25132 avx_<castmode><avxsizesuffix>_<castmode>,
25133 avx512f_<castmode><avxsizesuffix>_<castmode>,
25134 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
25135 in condition that at least one operand is not a MEM.
25137 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25139 PR middle-end/79805
25140 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
25141 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
25143 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
25144 gimple_call_nothrow_p flag based on whether original builtin can throw.
25145 If it can, emit following stmts on the fallthrough edge.
25146 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
25147 don't create new bb if inserting just debug stmts on the edge, try to
25148 insert them on the fallthru bb or just reset debug stmts.
25150 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
25153 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
25154 restore recog_data (including the operand rtxes inside it) around
25155 the call to get_insn_template.
25157 2017-03-03 Martin Sebor <msebor@redhat.com>
25159 PR tree-optimization/79699
25160 * context.c (context::~context): Free MPFR caches to avoid
25161 a memory leak on program exit.
25163 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25165 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
25166 Use wide_int::ulow () instead of .elt (0).
25168 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25170 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
25171 (*pushxf): Limit oF constraint to 32bit targets and add oC
25172 constraint for 64bit targets.
25173 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
25174 (*pushdf): Change rmF constraint to rmC.
25176 2017-03-03 Martin Liska <mliska@suse.cz>
25178 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
25179 Remove unused variable.
25181 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25184 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
25185 is a memory operand, increase num_memory.
25186 (ix86_expand_args_builtin): Likewise.
25188 2017-03-03 Jan Hubicka <jh@suse.cz>
25191 * ipa-devirt.c (maybe_record_node): Properly handle
25192 __cxa_pure_virtual visibility.
25194 2017-03-03 Martin Liska <mliska@suse.cz>
25196 PR tree-optimization/79803
25197 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
25199 (pass_loop_prefetch::execute): Disabled optimization if an
25200 assumption about L1 cache size is not met.
25202 2017-03-03 Martin Liska <mliska@suse.cz>
25204 PR rtl-optimization/79574
25205 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
25206 (hash_scan_set): Likewise.
25207 (dump_hash_table): Likewise.
25208 (hoist_code): Likewise.
25210 2017-03-03 Richard Biener <rguenther@suse.de>
25212 * fixed-value.c (fixed_from_string): Restore use of elt (1)
25213 in place of uhigh ().
25214 (fixed_convert_from_real): Likewise.
25216 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25219 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
25221 2017-03-03 Richard Biener <rguenther@suse.de>
25223 PR middle-end/79818
25224 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
25225 TYPE_OVERFLOW_UNDEFINED check.
25227 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25229 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
25231 (vector_ae_<mode>_p): Likewise.
25232 (vector_nez_<mode>_p): Likewise.
25233 (vector_ne_v2di_p): Likewise.
25234 (vector_ae_v2di_p): Likewise.
25235 (vector_ne_<mode>_p): Likewise.
25236 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
25238 (vsx_tsqrt<mode>2_fe): Likewise.
25240 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
25243 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
25245 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25247 PR rtl-optimization/79780
25248 * cprop.c (one_cprop_pass): When second and further conditional trap
25249 in a single basic block is turned into an unconditional trap, turn it
25250 into a deleted note to avoid RTL verification failures.
25252 2017-03-02 Richard Biener <rguenther@suse.de>
25254 * fold-const.c (const_binop): Use ulow () instead of elt (0).
25256 2017-03-02 Richard Biener <rguenther@suse.de>
25258 PR tree-optimization/79345
25260 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
25261 param and abort the walk, returning -1 if it is hit.
25262 (walk_aliased_vdefs): Take a limit param and pass it on.
25263 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
25264 defaulting to 0 and return a signed int.
25265 * tree-ssa-uninit.c (struct check_defs_data): New struct.
25266 (check_defs): New helper.
25267 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
25268 about uninitialized memory.
25269 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
25270 bogus uninitialized warning.
25271 (fixed_convert_from_real): Likewise.
25273 2017-03-02 Bin Cheng <bin.cheng@arm.com>
25275 PR tree-optimization/66768
25276 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
25277 iv_use if base object can't be determined.
25279 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25281 PR tree-optimization/79345
25282 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
25283 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
25284 (get_pattern_stats): Initialize it.
25285 * genemit.c (gen_expand): Verify match_scratch numbers come after
25286 match_operand/match_dup numbers.
25287 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
25288 match_scratch numbers.
25289 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
25291 * config/s390/s390.md (trunctdsd2): Likewise.
25293 2017-03-02 Richard Biener <rguenther@suse.de>
25295 * wide-int.h (wide_int_storage::operator=): Implement in terms
25298 2017-03-02 Richard Biener <rguenther@suse.de>
25300 PR tree-optimization/79777
25301 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
25302 the to insert expression to sth existing.
25304 2017-03-01 Martin Sebor <msebor@redhat.com>
25306 PR middle-end/79692
25307 * gimple-ssa-sprintf.c
25308 (directive::known_width_and_precision): New function.
25309 (format_integer): Use it.
25310 (get_mpfr_format_length): Consider the full range of precision
25311 when computing %g output with the # flag. Set the likely byte
25312 count to 3 rather than 1 when precision is indeterminate.
25313 (format_floating): Correct the lower bound of precision.
25315 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25317 * doc/invoke.texi: Document default code model for 64-bit Linux.
25319 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25322 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
25323 udiv rather than div since input pattern is unsigned.
25325 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
25327 * config/i386/i386.c (print_reg): Warn for values of
25328 unsupported size in integer register.
25330 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
25333 * config/rs6000/predicates.md (current_file_function_operand): Do
25334 not allow self calls to be local if the function is replaceable.
25336 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
25339 * config/rs6000/altivec.h (vec_ctz and others): Change the
25340 preprocessor macro that controls conditional compilation from
25341 _ARCH_PWR9 to __POWER9_VECTOR__.
25342 (vec_all_ne): Change parameterization of __altivec_scalar_pred
25343 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
25344 control (instead of _ARCH_PWR9 control) so that template
25345 definition uses power9-specific function.
25346 (vec_any_eq): Likewise.
25347 (vec_all_ne): Change macro definition to use a power9-specific
25348 expansion under #ifdef __POWER9_VECTOR__ control (instead of
25349 _ARCH_PWR9 control).
25350 (vec_any_eq) Likewise.
25351 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
25352 expansion for CMPNEF to remove support for xvcmpnesp instruction.
25353 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
25354 support for xvcmpnedp instruction.
25355 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
25356 macro expansion so that Power9 implementation of vec_all_ne does
25357 not use the AltiVec predicate framework.
25358 (VCMPNEH_P): Likewise.
25359 (VCMPNEW_P): Likewise.
25360 (VCMPNED_P): Likewise.
25361 (VCMPNEFP_P): Likewise.
25362 (VCMPNEDP_P): Likewise.
25363 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
25364 implementation of vec_any_eq to not use AltiVec predicate
25366 (VCMPAEH_P): Likewise.
25367 (VCMPAEW_P): Likewise.
25368 (VCMPAED_P): Likewise.
25369 (VCMPAEFP_P): Likewise.
25370 (VCMPAEDP_P): Likewise.
25371 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
25372 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
25373 not use the AltiVec predicate framework.
25374 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
25375 of vec_any_eq to not use AltiVec predicate framework.
25376 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25377 support for predefined __POWER9_VECTOR__ macro to indicate that
25378 Power9 instruction selection is enabled.
25379 (altivec_overloaded_builtins): Remove extraneous
25380 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
25381 function argument types RS6000_BTI_bool_V16QI and
25382 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
25383 entry for overloaded function argument types RS6000_BTI_bool_V4SI
25384 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
25385 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
25386 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
25387 Power9 for implementations of vec_cmpne. Change the signature for
25388 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
25389 (representing vec_all_ne) to remove the previously described first
25390 argument of type RS6000_BTI_INTSI, as this was an artifact of
25391 reliance on the AltiVec predicate framework, which is no longer
25392 used in the implementation of these functions. Add
25393 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
25394 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
25395 since, unlike the AltiVec predicate framework implementation, we
25396 do not share function descriptors between vec_alle and vec_anyeq.
25397 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
25398 set of modes that receive special treatment even when
25399 TARGET_P9_VECTOR is true. The special treatment emits code that
25400 does not depend on Power9 instructions.
25401 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
25402 define_expand to not rely on AltiVec predicate framework.
25403 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25405 (vector_ne_v2di_p): Change this define_expand to not rely on
25406 AltiVec predicate framework.
25407 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
25409 (vector_ne_<mode>_p): Change this define_expand to not rely on
25410 AltiVec predicate framework.
25411 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25413 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
25414 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
25415 define_insn pattern.
25416 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
25417 define_insn pattern because the xvcmpne<VSs>. instruction is not
25419 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
25420 instruction is not supported.
25422 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25424 * config/nvptx/nvptx.c: Include intl.h.
25426 2017-03-01 Martin Jambor <mjambor@suse.cz>
25429 * ipa-prop.h (ipa_bits): Removed field known.
25430 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
25431 to pointers. Adjusted their comments to warn about their sharing.
25432 (ipcp_transformation_summary): Change bits to a vector of pointers.
25433 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
25434 (ipa_get_ipa_bits_for_value): Declare.
25435 * tree-vrp.h (value_range): Mark as GTY((for_user)).
25436 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
25437 (ipa_bits_hash_table): Likewise.
25438 (ipa_vr_ggc_hash_traits): Likewise.
25439 (ipa_vr_hash_table): Likewise.
25440 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
25441 being pointers and vr_known being removed.
25442 (ipa_set_jf_unknown): Likewise.
25443 (ipa_get_ipa_bits_for_value): New function.
25444 (ipa_set_jfunc_bits): Likewise.
25445 (ipa_get_value_range): New overloaded functions.
25446 (ipa_set_jfunc_vr): Likewise.
25447 (ipa_compute_jump_functions_for_edge): Use the above functions to
25448 construct bits and vr parts of jump functions.
25449 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
25450 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25452 (ipcp_grow_transformations_if_necessary): Also allocate
25453 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25455 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
25456 them. Fix too long lines.
25457 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
25458 vr_known being removed.
25459 (ipa_read_jump_function): Use new setter functions to construct bits
25460 and vr parts of jump functions or set them to NULL.
25461 (write_ipcp_transformation_info): Adjust for bits being pointers.
25462 (read_ipcp_transformation_info): Likewise.
25463 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
25465 Include gt-ipa-prop.h.
25466 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
25468 (ipcp_store_bits_results): Likewise.
25469 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
25470 Do not write to existing jump functions but use a temporary instead.
25472 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25475 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
25476 attempt to use its first operand as BIT_FIELD_REF base.
25478 2017-03-01 Richard Biener <rguenther@suse.de>
25480 PR middle-end/79721
25481 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
25482 interpolating formula in wrapping arithmetic.
25483 (chrec_apply): Convert chrec_evaluate return value to wanted type.
25485 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25487 PR tree-optimization/79734
25488 * tree-vect-generic.c (expand_vector_condition): Optimize
25489 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
25490 Handle VEC_COND_EXPR where comparison has different inner width from
25491 type's inner width.
25493 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
25495 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
25496 markup, and similar issues. Remove @opindex entries for things
25497 that aren't options. Add missing -mmpy-option entries.
25499 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25501 PR tree-optimization/79737
25502 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
25503 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
25504 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
25505 instead of byte_size. Formatting fix.
25506 (shift_bytes_in_array_right): Formatting fix.
25508 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
25511 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
25512 condition on optimize for the leaf function test.
25514 2017-02-28 Martin Liska <mliska@suse.cz>
25517 * read-rtl-function.c (function_reader::handle_unknown_directive):
25518 Bail out when one uses -flto.
25520 2017-02-28 Martin Liska <mliska@suse.cz>
25522 * common.opt: Replace space with tabular for options of <number>
25524 * config/i386/i386.opt: Show <number> value for
25525 -mlarge-data-threshold.
25526 * opts.c (print_filtered_help): Do not display number in hexadecimal
25529 2017-02-28 Martin Liska <mliska@suse.cz>
25531 * common.opt: Fix --help=option -Q for options which are of
25534 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
25536 * config/i386/i386.c (print_reg): Error out for values
25537 of 8-bit size in invalid integer register.
25539 2017-02-28 Martin Sebor <msebor@redhat.com>
25541 PR tree-optimization/79691
25542 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
25544 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25547 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
25548 gcc_unreachable with output_operand_lossage.
25550 2017-02-28 Richard Biener <rguenther@suse.de>
25552 PR tree-optimization/79740
25553 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
25555 (visit_nary_op): Insert the nary into the hashtable if we
25556 pattern-matched sth.
25557 * tree-ssa-pre.c (eliminate_insert): Robustify.
25559 2017-02-28 Richard Biener <rguenther@suse.de>
25561 PR middle-end/79731
25562 * fold-const.c (decode_field_reference): Reject out-of-bound
25565 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25567 * config/i386/i386.c: Include intl.h.
25568 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
25569 instead of just cond ? "..." : "...".
25570 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
25571 * coverage.c (read_counts_file): Likewise.
25572 * omp-offload.c: Include intl.h.
25573 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
25574 of just cond ? "..." : "...".
25575 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
25576 of just cond ? "..." : "...".
25578 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
25581 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
25583 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
25584 'tune for' CPU name.
25585 * config/arm/arm-cpu-data.h: Regenerated.
25587 2017-02-28 Richard Biener <rguenther@suse.de>
25589 PR tree-optimization/79732
25590 * tree-inline.c (expand_call_inline): Do not shadow var.
25592 2017-02-28 Richard Biener <rguenther@suse.de>
25594 PR tree-optimization/79723
25595 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
25596 address-space properly.
25598 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
25600 * doc/optinfo.texi (Optimization groups): Fix option used for
25602 * doc/invoke.texi (-fopt-info): Document "omp".
25603 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
25604 (OPTGROUP_ALL): Add OPTGROUP_OMP.
25605 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
25606 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
25607 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
25609 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
25611 * dumpfile.c (optgroup_options): Instead of "openmp", associate
25612 OPTGROUP_OMP with "omp".
25614 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
25617 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
25618 for arithmetic shift of unsigned V2DI.
25620 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
25622 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
25623 arc/linux.h headers.
25624 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
25625 (LINK_SPEC): Likewise.
25626 (ARC_TLS_EXTRA_START_SPEC): Likewise.
25627 (EXTRA_SPECS): Likewise.
25628 (STARTFILE_SPEC): Likewise.
25629 (ENDFILE_SPEC): Likewise.
25630 (LIB_SPEC): Likewise.
25631 (TARGET_SDATA_DEFAULT): Likewise.
25632 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
25633 (MULTILIB_DEFAULTS): Likewise.
25634 (DWARF2_UNWIND_INFO): Likewise.
25635 * config/arc/big.h: New file.
25636 * config/arc/elf.h: Likewise.
25637 * config/arc/linux.h: Likewise.
25638 * config/arc/t-uClibc: Remove.
25640 2017-02-27 Bin Cheng <bin.cheng@arm.com>
25642 PR tree-optimization/77536
25643 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
25644 (tree_transform_and_unroll_loop): Use above function to compute the
25645 estimated niter of unrolled loop and use it when scaling profile.
25646 Also use count info rather than frequency if it's non-zero.
25647 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
25648 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
25649 (vect_transform_loop): Call above function.
25651 2017-02-27 Richard Biener <rguenther@suse.de>
25653 PR tree-optimization/45397
25654 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
25655 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
25656 (visit_nary_op): Add pattern matching for CSEing sign-changed
25657 or truncated operations with wider ones.
25659 2017-02-27 Richard Biener <rguenther@suse.de>
25661 PR tree-optimization/79690
25662 * tree-vect-stmts.c (vectorizable_store): Use vector type
25663 built from the DR with address-space.
25665 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
25667 * doc/invoke.texi (Optimize Options): Refine the description
25668 of asan-use-after-return.
25670 2017-02-25 Alan Modra <amodra@gmail.com>
25672 PR rtl-optimization/79584
25673 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
25674 base, not ad->base_term, the reg within base. Remove assertion
25675 that ad->base == ad->base_term. Replace gen_int_mode using
25676 bogus mode with const0_rtx.
25678 2017-02-25 Jakub Jelinek <jakub@redhat.com>
25680 PR middle-end/79396
25681 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
25682 FMA_EXPR like tcc_binary or tcc_unary.
25684 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
25687 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
25689 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
25690 (output_loc_operands): Handle DW_OP_call_ref and
25691 DW_OP_GNU_variable_value.
25692 (struct variable_value_struct): New type.
25693 (struct variable_value_hasher): Likewise.
25694 (variable_value_hash): New variable.
25695 (string_types): Remove.
25696 (copy_loc_descr): New function.
25697 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
25698 (prepend_loc_descr_to_each): New function.
25699 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
25700 instead of add_loc_descr_to_each if the first argument is single
25701 location list and the second has multiple.
25702 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
25703 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
25704 when looking for variable value which doesn't have other location info.
25705 (loc_list_from_tree): Formatting fix.
25706 (gen_array_type_die): Simplify DW_AT_string_length handling.
25707 (adjust_string_types): Remove.
25708 (gen_subprogram_die): Don't call adjust_string_types nor test/set
25709 string_types. Call resolve_variable_values.
25710 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
25711 (resolve_addr_in_expr): Likewise. Add A argument.
25712 (copy_deref_exprloc): Remove deref argument. Adjust for the
25713 original expression being DW_OP_GNU_variable_value with optionally
25714 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
25715 optionally after it.
25716 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
25717 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
25718 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
25719 (variable_value_hasher::hash, variable_value_hasher::equal): New
25721 (resolve_variable_value_in_expr, resolve_variable_value,
25722 resolve_variable_values, note_variable_value_in_expr,
25723 note_variable_value): New functions.
25724 (dwarf2out_early_finish): Call note_variable_value on all toplevel
25727 2017-02-24 Jakub Jelinek <jakub@redhat.com>
25730 * opts.h (handle_generated_option): Add GENERATED_P argument.
25731 * opts-common.c (handle_option): Adjust function comment.
25732 (handle_generated_option): Add GENERATED_P argument, pass it to
25734 (control_warning_option): Pass false to handle_generated_option
25736 * opts.c (maybe_default_option): Pass true to handle_generated_option
25738 * optc-gen.awk: Likewise.
25740 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25742 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
25743 a REG, look at the REG it is a SUBREG of.
25744 (splitter for cmpeqsi_t): Ditto.
25746 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25748 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
25749 the special USEs with the pattern of the insn, not the insn itself.
25751 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
25754 * doc/invoke.texi: Document -mload-store-pairs.
25756 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25757 Sandra Loosemore <sandra@codesourcery.com>
25759 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
25760 argument isn't a CONST_INT.
25761 (nios2_alternate_compare_const): Assert op is a CONST_INT.
25762 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
25763 (nios2_validate_compare): Bypass alternate compare logic if *op2
25764 is not a CONST_INT.
25765 (ldstwm_operation_p): Return false if first_base is not a REG or
25766 if first_offset is not a CONST_INT.
25768 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25770 * config/cris/cris.md: Use correct operand in a define_peephole2.
25772 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25774 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
25776 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25778 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
25779 this_insn if it is an INSN or JUMP_INSN.
25780 (force_offsettable): Look at base, not at addr.
25781 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
25782 on things that aren't necessarily CONST_INTs.
25784 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
25786 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
25787 -mfpmath=sse is the default also for x86-32 targets with SSE2
25788 instruction set when @option{-ffast-math} is enabled
25790 2017-02-24 Jeff Law <law@redhat.com>
25792 PR rtl-optimizatoin/79286
25793 * ira.c (update_equiv_regs): Drop may_trap_p exception to
25796 2017-02-24 Richard Biener <rguenther@suse.de>
25798 PR tree-optimization/79389
25799 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
25802 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
25804 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
25805 function comment to reflect reality.
25806 (loop_exits_before_overflow): Fix typo in function description.
25808 2017-02-24 Richard Biener <rguenther@suse.de>
25810 PR tree-optimization/79389
25811 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
25812 properly that a threading opportunity exists. Detect conditional
25813 copy/constant propagation opportunities.
25815 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
25817 * config/visium/visium.md (type): Add trap.
25818 (b): New mode attribute.
25819 (*btst): Rename into...
25820 (*btst<mode>): ...this and adjust.
25821 (*cbranchsi4_btst_insn): Rename into...
25822 (*cbranch<mode>4_btst_insn): ...this and adjust.
25823 (trap): New define_insn.
25825 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25827 PR tree-optimization/79389
25828 * ifcvt.c (struct noce_if_info): Add rev_cond field.
25829 (noce_reversed_cond_code): New function.
25830 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
25831 reversed_comparison_code. Formatting fix.
25832 (noce_try_store_flag): Test rev_cond != NULL in addition to
25833 reversed_comparison_code.
25834 (noce_try_store_flag_constants): Likewise.
25835 (noce_try_store_flag_mask): Likewise.
25836 (noce_try_addcc): Use rev_cond if non-NULL instead of
25837 reversed_comparison_code.
25838 (noce_try_cmove_arith): Likewise. Formatting fixes.
25839 (noce_try_minmax, noce_try_abs): Clear rev_cond.
25840 (noce_find_if_block): Initialize rev_cond.
25841 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
25842 instead of false as last argument never attempt to reverse it
25845 2017-02-23 Bin Cheng <bin.cheng@arm.com>
25847 PR tree-optimization/79663
25848 * tree-predcom.c (combine_chains): Process refs in reverse order
25849 only for ZERO length chains, and add explaining comment.
25851 2017-02-23 Jeff Law <law@redhat.com>
25853 PR tree-optimization/79578
25854 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
25855 in call to operand_equal_p.
25857 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
25860 * config/i386/cpuid.h: Fix another undefined behavior.
25862 2017-02-23 Richard Biener <rguenther@suse.de>
25864 PR tree-optimization/79683
25865 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
25866 vector types for data-refs.
25868 2017-02-23 Martin Liska <mliska@suse.cz>
25870 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
25872 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25874 PR middle-end/79665
25875 * internal-fn.c (get_range_pos_neg): Moved to ...
25876 * tree.c (get_range_pos_neg): ... here. No longer static.
25877 * tree.h (get_range_pos_neg): New prototype.
25878 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
25879 are known to be in between 0 and signed maximum inclusive, try to
25880 expand both unsigned and signed divmod and use the cheaper one from
25883 2017-02-22 Jeff Law <law@redhat.com>
25885 PR tree-optimization/79578
25886 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
25887 to compare base operands.
25889 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
25892 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
25893 gpc_reg_operand instead of fpr_reg_operand.
25895 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
25897 * config/mips/mips.c (mips_return_in_memory): Force FP
25898 vector types to be returned in memory for o32 ABI.
25900 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25902 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
25903 instead of DW_TAG_member for static data member declarations and don't
25904 set no_linkage_name for static inline data members.
25905 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
25908 2017-02-22 Martin Liska <mliska@suse.cz>
25910 * doc/invoke.texi: Replace inequality signs with square brackets
25913 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25915 PR tree-optimization/68644
25916 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
25918 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
25921 * lra-constraints.c (simplify_operand_subreg): Handle
25922 WORD_REGISTER_OPERATIONS targets.
25924 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25927 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
25928 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
25929 elimination by swapping fld*.
25931 2017-02-22 Richard Biener <rguenther@suse.de>
25933 PR tree-optimization/79673
25934 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
25935 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
25936 irrelevant address-space qualifiers and avoiding a
25937 ADDR_SPACE_CONVERT_EXPR from fold_convert.
25939 2017-02-22 Richard Biener <rguenther@suse.de>
25941 PR tree-optimization/79666
25942 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
25943 to not symbolically negate if that may introduce undefined
25946 2017-02-22 Martin Liska <mliska@suse.cz>
25949 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
25950 * data-streamer-out.c (streamer_write_gcov_count_stream):
25952 * value-prof.c (stream_out_histogram_value): Make assert more
25953 precise based on type of counter.
25955 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
25958 * config/i386/i386.md (standard_x87sse_constant_load splitter):
25959 Use nonimmediate_operand instead of memory_operand for operand 1.
25960 (float-extend standard_x87sse_constant_load splitter): Ditto.
25962 2017-02-21 Jeff Law <law@redhat.com>
25964 PR tree-optimization/79621
25965 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
25966 blocks with edges to themselves.
25968 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25971 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
25972 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
25973 Use gimple_call_builtin_p.
25976 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
25977 on temporarily removed DEBUG_INSNs.
25979 PR tree-optimization/79649
25980 * tree-loop-distribution.c (classify_partition): Give up on
25981 non-generic address space loads/stores.
25983 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
25985 * doc/loop.texi (Loop manipulation): Remove nonexistent
25986 tree_ssa_loop_version from the documentation.
25987 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
25989 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25992 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
25993 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
25994 * config/rs6000/rs6000.c: Include except.h.
25995 (rs6000_expand_split_stack_prologue): Call
25996 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
25998 2017-02-21 Martin Jambor <mjambor@suse.cz>
26001 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
26002 have been analyzed.
26004 2017-02-21 Martin Jambor <mjambor@suse.cz>
26006 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
26007 for backward compatibility only.
26008 * doc/invoke.texi (Option Summary): Remove all references to
26009 -fipa-cp-alignment.
26011 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
26015 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26017 * lra-constraints.c (curr_insn_transform): Handle
26018 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26020 2017-02-21 Martin Liska <mliska@suse.cz>
26022 * config/i386/i386.opt: Replace -masm-dialect with -masm.
26024 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
26026 PR translation/79638
26027 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
26029 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
26032 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
26033 (arm_function_ok_for_sibcall): Return false for an indirect call by
26034 descriptor if all the argument registers are used.
26035 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
26036 alignment of the function.
26038 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26040 PR tree-optimization/61441
26041 * simplify-rtx.c (simplify_const_unary_operation): For
26042 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
26043 the sNaN unmodified.
26045 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26047 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26048 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26049 instead of SYSTEM_HEADER_DIR.
26051 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
26052 Martin Liška <mliska@suse.cz>
26054 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
26055 Fix typos and grammar, use active voice, and clarify.
26057 2017-02-20 Marek Polacek <polacek@redhat.com>
26059 PR middle-end/79537
26060 * gimplify.c (gimplify_expr): Handle unused *&&L;.
26063 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
26065 2017-02-20 Jakub Jelinek <jakub@redhat.com>
26068 * config/i386/i386.c (ix86_expand_builtin): Handle
26069 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
26070 ix86_builtins_isa[fcode].isa as a requirement of those
26071 flags and any other flag in the bitmask.
26072 (ix86_init_mmx_sse_builtins): Use 0 instead of
26073 ~OPTION_MASK_ISA_64BIT as mask.
26074 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
26075 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
26076 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
26077 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
26079 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26082 * lra-constraints.c (split_reg): Check requested split mode
26083 is supported by the register.
26085 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26087 * lra-constraints.c (simplify_operand_subreg): Remove early
26090 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26093 * lra-constraints.c (curr_insn_transform): Tighten condition
26094 for converting SUBREG reloads from OP_OUT to OP_INOUT.
26096 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26099 * lra-constraints.c (curr_insn_transform): Handle
26100 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26102 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
26105 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
26107 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
26109 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
26112 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
26115 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26117 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
26120 2017-02-18 Jakub Jelinek <jakub@redhat.com>
26122 * final.c (last_columnnum, override_columnnum): New variables.
26123 (final_start_function): Set last_columnnum, pass it to begin_prologue
26124 hook and pass 0 to dwarf2out_begin_prologue.
26125 (final_scan_insn): Update override_columnnum. Pass last_columnnum
26126 to source_line debug hook.
26127 (notice_source_line): Compute last_columnnum and for debug_column_info
26128 return true on column changes.
26129 * debug.h (struct gcc_debug_hooks): Add column argument to
26130 source_line and begin_prologue hooks.
26131 (debug_nothing_int_charstar_int_bool): Remove prototype.
26132 (debug_nothing_int_int_charstar,
26133 debug_nothing_int_int_charstar_int_bool): New prototypes.
26134 (dwarf2out_begin_prologue): Add column argument.
26135 * debug.c (do_nothing_debug_hooks): Adjust source_line and
26136 begin_prologue hooks.
26137 (debug_nothing_int_charstar_int_bool): Remove.
26138 (debug_nothing_int_int_charstar,
26139 debug_nothing_int_int_charstar_int_bool): New functions.
26140 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
26141 through to dwarf2out_source_line.
26142 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
26143 (dwarf2out_source_line): Add column argument, emit it if requested.
26144 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
26146 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26147 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26148 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
26149 through to dwarf2out_begin_prologue.
26150 (vmsdbgout_source_line): Add column argument, pass it through to
26151 dwarf2out_source_line.
26152 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
26153 dbxout_source_line caller.
26154 (dbxout_source_line): Add column argument.
26156 * common.opt (gno-column-info, gcolumn-info): New options.
26157 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
26158 (check_die): Also test for multiple DW_AT_decl_column attributes.
26159 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
26160 DW_AT_decl_column if requested.
26161 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
26163 (gen_variable_die): Likewise.
26164 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
26165 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
26168 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
26169 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
26170 (ix86_handle_option): Handle OPT_m3dnowa.
26171 * doc/invoke.texi (-m3dnowa): Document.
26172 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
26173 -m3dnowa instead of -m3dnow -march=athlon.
26176 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
26177 instead of gcc_assert for K, r and R code checks. Formatting fixes.
26179 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26182 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
26183 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
26184 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
26185 generator for vsx_xxpermdi_<mode>_be.
26186 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
26187 force big-endian semantics.
26188 (vsx_xxpermdi_<mode>_be): New define_expand with same
26189 implementation as previous version of vsx_xxpermdi_<mode>.
26191 2017-02-17 Jakub Jelinek <jakub@redhat.com>
26193 PR tree-optimization/79327
26194 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
26195 variable, its initialization and use.
26197 2017-02-17 Julia Koval <julia.koval@intel.com>
26199 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
26200 (OPTION_MASK_ISA_PKU_UNSET): New.
26201 (ix86_handle_option): Handle -mrdpid.
26202 * config/i386/cpuid.h (bit_RDPID): New.
26203 * config/i386/driver-i386.c (host_detect_local_cpu):
26204 Detect RDPID feature.
26205 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
26206 * config/i386/i386-c.c (ix86_target_macros_internal):
26208 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
26209 (ix86_valid_target_attribute_inner_p): Add "rdpid".
26210 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26211 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
26212 * config/i386/i386.md (define_insn "rdpid"): New.
26213 * config/i386/i386.opt Add -mrdpid.
26214 * config/i386/immintrin.h (_rdpid_u32): New.
26216 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
26218 PR rtl-optimization/79541
26219 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
26220 instead of transforming it into USE.
26222 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
26224 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
26225 If HONOR_SNANS (SFmode) force the input to a register.
26226 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
26227 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
26228 an frsp or similar insn.
26230 2017-02-17 Martin Liska <mliska@suse.cz>
26232 PR rtl-optimization/79577
26233 * params.def (selsched-max-sched-times): Increase minimum to 1.
26235 2017-02-17 Martin Liska <mliska@suse.cz>
26237 PR rtl-optimization/79574
26238 * gcse.c (want_to_gcse_p): Prevent integer overflow.
26240 2017-02-17 Martin Liska <mliska@suse.cz>
26242 PR tree-optimization/79529
26243 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
26244 ssa_defined_default_def_p to handle cases which are implicitly
26246 * tree-ssa.c (ssa_defined_default_def_p): New function.
26247 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
26248 which are implicitly defined.
26249 * tree-ssa.h (ssa_defined_default_def_p): Declare.
26251 2017-02-17 Richard Biener <rguenther@suse.de>
26253 PR middle-end/79576
26254 * params.def (max-ssa-name-query-depth): Limit to 10.
26256 2017-02-17 Richard Biener <rguenther@suse.de>
26258 PR tree-optimization/79552
26259 * tree-ssa-structalias.c (visit_loadstore): Properly verify
26262 2017-02-17 Richard Biener <rguenther@suse.de>
26265 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
26267 2017-02-17 Marek Polacek <polacek@redhat.com>
26269 PR middle-end/79536
26270 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
26271 (fold_negate_expr): New wrapper.
26273 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
26275 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
26276 Correct terminology and de-emphasize pre-standard behavior.
26278 2017-02-16 Alan Modra <amodra@gmail.com>
26280 PR rtl-optimization/79286
26281 * ira.c (def_dominates_uses): New function.
26282 (update_equiv_regs): Don't create an equivalence for insns that
26283 may trap where the register def does not dominate the use.
26285 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
26287 PR rtl-optimization/78127
26288 * lra.c (lra): Call lra_eliminate before finish the loop after
26291 2017-02-16 Richard Biener <rguenther@suse.de>
26293 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
26295 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
26296 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
26297 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
26298 (isl_val_int_from_wi): New function.
26299 (extract_affine_gmp): Rename to ...
26300 (extract_affine_wi): ... this, take a widest_int.
26301 (extract_affine_int): Just wrap extract_affine_wi.
26302 (add_param_constraints): Use isl_val_int_from_wi.
26303 (add_loop_constraints): Likewise, and extract_affine_wi.
26305 2017-02-15 Jeff Law <law@redhat.com>
26307 PR middle-end/79521
26308 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
26309 ira_init_register_move_cost_if_necessary.
26311 2017-02-15 Martin Sebor <msebor@redhat.com>
26313 PR middle-end/32003
26314 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
26315 removed in a prior commit.
26317 2017-02-15 Bin Cheng <bin.cheng@arm.com>
26319 PR tree-optimization/79347
26320 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
26321 counters during peeling.
26323 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
26325 * Makefile.in (site.exp): Remove "set ISLVER".
26327 2017-02-15 Jakub Jelinek <jakub@redhat.com>
26330 * real.c (real_from_integer): Call real_convert even for decimal.
26332 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
26335 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
26337 2017-02-14 Andrew Pinski <apinski@cavium.com>
26339 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
26340 cores and change the partno/implementer to be correct.
26341 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
26342 the 'B" as the implementer.
26343 * config/aarch64/aarch64-tune.md: Regenerate.
26345 2017-02-14 Carl Love <cel@us.ibm.com>
26347 * config/rs6000/rs6000.c: Add case statement entry to make the
26348 xvcvuxdsp built-in argument unsigned.
26349 * config/rs6000/vsx.md: Fix the source and return operand types so they
26350 match the instruction definitions from the ISA document. Fix typo
26351 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
26354 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
26357 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
26358 member early_clobber_alts.
26359 * lra-lives.c (reg_early_clobber_p): New.
26360 (process_bb_lives): Use it.
26361 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
26362 (debug_operand_data): Initialize early_clobber_alts.
26363 (setup_operand_alternative): Set up early_clobber_alts.
26364 (collect_non_operand_hard_regs): Ditto. Pass early clobber
26365 alternatives to new_insn_reg.
26366 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
26368 (lra_update_insn_regno_info): Pass the new arg.
26370 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26372 PR middle-end/79505
26373 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
26374 (new_oacc_loop_raw): Don't clear already cleared fields.
26377 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
26378 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
26379 _mm512_prefetch_i64gather_ps): New inline functions and macros.
26381 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
26384 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
26386 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
26389 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
26390 the extra instruction to the right place to store 128-bit constant
26393 2017-02-14 Martin Sebor <msebor@redhat.com>
26395 PR middle-end/79448
26396 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
26397 warning for strings of unknown length.
26399 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26401 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
26403 2017-02-14 Jeff Law <law@redhat.com>
26406 * ira-costs.c (scan_one_insn): Initialize register move costs
26407 for pseudos seen in USE/CLOBBER insns.
26409 PR tree-optimization/79095
26410 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
26411 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
26412 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
26413 if the operands are known to be not equal, then the resulting range
26415 (intersect_ranges): If the new range is ~[0,0] and the old range is
26416 wide, then prefer ~[0,0].
26417 * tree-vrp.c (overflow_comparison_p_1): New function.
26418 (overflow_comparison_p): New function.
26419 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
26420 if NAME is used in an overflow test.
26421 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
26422 overflow check that can be expressed as an equality test, then adjust
26423 ops to be that equality test.
26425 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26427 * config/s390/s390-builtin-types.def: Remove flags argument.
26428 * config/s390/s390.c (s390_init_builtins): Likewise.
26430 2017-02-14 Martin Liska <mliska@suse.cz>
26432 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
26433 vector. Fix trailing white spaces.
26435 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
26437 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
26440 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26442 PR rtl-optimization/68664
26443 * config/arm/arm.c (arm_sched_can_speculate_insn):
26444 New function. Declare prototype.
26445 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26447 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26449 PR rtl-optimization/68664
26450 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
26452 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26454 2017-02-14 Amit Pawar <amit.pawar@amd.com>
26456 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
26457 max skip bytes for function, loop and jump.
26459 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26461 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
26462 ABS_EXPR for gimple dump.
26464 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26467 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
26469 PR tree-optimization/79408
26470 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
26471 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
26472 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
26473 also if rhs1 is INTEGER_CST.
26475 2017-02-14 Richard Biener <rguenther@suse.de>
26477 PR middle-end/79432
26478 * tree-into-ssa.c (insert_phi_nodes): When the function can
26479 have abnormal edges rewrite SSA names with broken use-def
26480 dominance out of SSA and register them for PHI insertion.
26482 2017-02-13 Martin Sebor <msebor@redhat.com>
26484 PR middle-end/79496
26485 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
26486 clearing info.nowrite flag when snprintf size argument is a range.
26488 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26490 * cprop.c (cprop_jump): Add missing space in string literal.
26491 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
26492 (get_constraint_for_component_ref): Likewise.
26493 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
26494 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
26495 * lra-constraints.c (process_alt_operands): Likewise.
26496 * ipa-inline.c (inline_small_functions): Likewise.
26497 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
26498 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26499 * trans-mem.c (diagnose_tm_1_op): Likewise.
26500 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
26501 (grid_parallel_clauses_gridifiable): Likewise.
26503 * config/nvptx/mkoffload.c (process): Add space in between
26506 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
26507 "MOD4_SSE_REGS" and "ALL_REGS".
26509 * spellcheck.c (test_data): Add , in between "foo" and "food".
26511 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26514 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
26515 boundary crossing check and subsequent code generation agree.
26517 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26519 * config/aarch64/aarch64.c (has_memory_op): Delete.
26520 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
26523 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26525 PR rtl-optimization/79388
26526 PR rtl-optimization/79450
26527 * combine.c (distribute_notes): When removing TEM_INSN for which
26528 corresponding dest has last value recorded, invalidate that last
26531 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26533 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
26534 of explicit '@'. Add missing assembly comment marker on branch costs
26537 2017-02-13 Nathan Sidwell <nathan@acm.org>
26539 * gengtype-lex.l (<in_struct>): Add '/'.
26541 2017-02-13 Martin Liska <mliska@suse.cz>
26544 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
26546 2017-02-13 Richard Biener <rguenther@suse.de>
26548 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
26550 * configure: Re-generate.
26551 * config.in: Likewise.
26552 * graphite-dependences.c: Simplify as if
26553 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
26554 * graphite-isl-ast-to-gimple.c: Likewise.
26555 * graphite-optimize-isl.c: Likewise.
26556 * graphite-poly.c: Likewise.
26557 * graphite-sese-to-poly.c: Likewise.
26558 * graphite.h: Likewise.
26559 * toplev.c: Include isl/version.h and use isl_version () for
26560 printing the ISL version.
26561 * doc/install.texi: Update ISL requirement.
26563 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
26565 * doc/standards.texi (Standards): Update reference to
26568 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
26570 * doc/extend.texi (Named Address Spaces): sourceware.org now
26572 * doc/install.texi (Binaries): Ditto.
26575 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26577 * doc/cpp.texi: Replace "stringify"/"stringification" with C
26578 standard terminology "stringize"/"stringizing" throughout.
26579 * doc/cppinternals.texi: Likewise.
26581 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26583 * doc/extend.texi: Fix some spelling mistakes and typos.
26584 * doc/invoke.texi: Likewise.
26586 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26589 * params.def (inline-min-speedup) Change from 10 to 8.
26591 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26593 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
26596 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26599 * ipa-inline-analysis.c (get_minimal_bb): New function.
26600 (record_modified): Use it.
26601 (remap_edge_change_prob): Handle also ancestor functions.
26603 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
26605 * doc/contrib.texi (Contributors): Remove broken link into
26606 the Mauve CVS repository.
26608 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26610 PR middle-end/79454
26611 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
26612 result computation whenever lhs doesn't have vector mode, not
26613 just when it has BLKmode.
26615 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26617 * doc/makefile.texi (profiledbootstrap): Refer to the
26618 installation instructions only in textual form.
26620 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26623 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
26625 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26627 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
26628 (Specific): Update mingw-w64 reference.
26630 (Specific): Remove broken link to Renesas RX processor.
26632 2017-02-10 Richard Biener <rguenther@suse.de>
26634 * toplev.c (process_options): Do not mention obsolete graphite
26635 options when printing sorry message about missing graphite support.
26636 Mention -floop-nest-optimize.
26638 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
26640 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
26641 (vtst_p16): Likewise.
26642 (vtstq_p8): Likewise.
26643 (vtstq_p16): Likewise.
26645 (vtstq_p64): Likewise.
26646 * config/arm/arm_neon.h (vgetq_lane_p64): New.
26647 (vset_lane_p64): New.
26648 (vsetq_lane_p64): New.
26650 2017-02-10 Jakub Jelinek <jakub@redhat.com>
26652 PR tree-optimization/79411
26653 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
26654 stmt operands are SSA_NAMEs used in abnormal phis.
26655 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
26658 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26661 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
26664 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26666 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
26668 2017-02-09 Jakub Jelinek <jakub@redhat.com>
26670 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
26674 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
26675 not arbitrary TREE_CONSTANT.
26678 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
26679 "omp declare target link" attribute unless is_global_var.
26680 * omp-offload.c (find_link_var_op): Likewise.
26682 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
26683 Chung-Lin Tang <cltang@codesourcery.com>
26685 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
26687 (gimplify_adjust_omp_clauses): Don't delete TILE.
26688 (gimplify_omp_for): Deal with TILE.
26689 * internal-fn.c (expand_GOACC_TILE): New function.
26690 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
26692 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
26693 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
26695 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
26696 avoid DIV for outermost collapse var.
26697 (expand_oacc_for): Insert tile element loop as needed. Adjust.
26698 Remove out of date comments, fix whitespace.
26699 * omp-general.c (omp_extract_for_data): Deal with tiling.
26700 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
26701 adjust OLF_DIM_BASE value.
26702 (struct omp_for_data): Add tiling field.
26703 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
26704 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
26705 for auto loops. Remove default auto determining, moved to
26706 oacc_loop_fixed_partitions.
26707 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
26708 stmts, add e_mask field.
26709 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
26710 (oacc_thread_numbers): Use oacc_dim_call.
26711 (oacc_xform_tile): New.
26712 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
26713 (finish_oacc_loop): Adjust for ifns vector.
26714 (oacc_loop_discover_walk): Append loop abstraction sites to list,
26715 add case for GOACC_TILE fns.
26716 (oacc_loop_xform_loop): Delete.
26717 (oacc_loop_process): Iterate over call list directly, and add
26718 handling for GOACC_TILE fns.
26719 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
26721 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
26722 vector partitioning to outer loops. Assign 2 partitions to loops
26723 when available. Add TILE handling.
26724 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
26725 (execite_oacc_device_lower): Process GOACC_TILE fns,
26726 ignore unknown specs.
26727 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
26728 * tree.c (omp_clause_num_ops): Adjust TILE ops.
26729 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
26731 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
26733 * configure.ac (ACX_BUGURL): Update.
26734 * configure: Regenerate.
26736 2017-02-09 Richard Biener <rguenther@suse.de>
26738 PR tree-optimization/69823
26739 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
26740 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
26742 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26744 * config/arc/arc-c.def: Add __NPS400__ definition.
26745 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
26746 (TARGET_NPS400): Define.
26748 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26750 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
26752 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
26753 pointer, arch_info.
26754 (arc_cpu_types): Fill the arch_info field with a pointer into the
26755 arc_arch_types table.
26756 (arc_selected_cpu): Declare.
26757 * config/arc/arc.c (arc_selected_cpu): Make global.
26758 (arc_selected_arch): Delete.
26759 (arc_base_cpu): Delete.
26760 (arc_override_options): Remove references to deleted variables,
26761 update access to arch information.
26762 (ARC_OPT): Update access to arch information.
26763 (ARC_OPTX): Likewise.
26764 * config/arc/arc.h (arc_base_cpu): Remove declaration.
26765 (TARGET_ARC600): Update access to arch information.
26766 (TARGET_ARC601): Likewise.
26767 (TARGET_ARC700): Likewise.
26768 (TARGET_EM): Likewise.
26769 (TARGET_HS): Likewise.
26770 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
26773 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
26776 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
26777 condition/operands for integer GE/LE/GEU/LEU operations.
26779 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
26781 PR translation/79397
26782 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
26785 2017-02-08 Martin Jambor <mjambor@suse.cz>
26788 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
26789 whether allocation happened.
26790 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
26791 nothing was allocated.
26793 2017-02-08 Jakub Jelinek <jakub@redhat.com>
26795 PR tree-optimization/79408
26796 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
26797 constant, but SSA_NAME with a known integer range, use the minimum
26798 of that range instead of op1 to determine if modulo can be replaced
26799 with its first operand.
26801 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26803 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
26805 2017-02-08 Richard Biener <rguenther@suse.de>
26807 PR tree-optimization/71824
26808 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26809 Check all loops contained in the merged region.
26811 2017-02-07 Andrew Pinski <apinski@cavium.com>
26813 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
26815 2017-02-07 Andrew Pinski <apinski@cavium.com>
26817 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
26818 (thunderxt88): Likewise.
26819 (thunderxt81): Disable LSE and change v8.1 to v8.
26820 (thunderxt83): Likewise.
26822 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26823 Richard Biener <rguenther@suse.de>
26825 PR middle-end/79399
26826 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
26827 type from int to size_t.
26828 * ira-costs.c (struct_costs_size): Change type from int to size_t.
26830 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26832 PR rtl-optimization/79386
26833 * cprop.c (bypass_conditional_jumps): Initialize
26834 bypass_last_basic_block already before splitting bbs after
26835 unconditional traps...
26836 (bypass_conditional_jumps): ... rather than here.
26839 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
26840 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
26841 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
26842 fix -masm=intel patterns.
26844 2017-02-07 Richard Biener <rguenther@suse.de>
26846 PR tree-optimization/79256
26847 PR middle-end/79278
26848 * builtins.c (get_object_alignment_2): Use min_align_of_type
26849 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
26850 and ADJUST_FIELD_ALIGN.
26852 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
26854 * doc/tm.texi: Regenerate.
26855 * stor-layout.c (layout_decl): Adjust.
26856 (update_alignment_for_field): Likewise.
26857 (place_field): Likewise.
26858 (min_align_of_type): Likewise.
26859 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
26860 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
26861 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
26862 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
26863 * config/frv/frv.c (frv_adjust_field_align): Likewise.
26864 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
26865 * config/i386/i386.c (x86_field_alignment): Likewise.
26866 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
26867 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
26868 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
26869 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
26870 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
26871 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
26875 2017-01-30 Richard Biener <rguenther@suse.de>
26877 PR tree-optimization/79256
26878 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26879 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26882 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
26884 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
26885 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
26886 builtins to SImode and emit a zero-extend, if necessary.
26888 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26890 * docs/invoke.texi (RISC-V Options): Alphabetize.
26892 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26894 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
26897 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26899 * config/riscv/riscv.c: New file.
26900 * common/config/riscv/riscv-common.c: Likewise.
26901 * config.gcc: Likewise.
26902 * config/riscv/constraints.md: Likewise.
26903 * config/riscv/elf.h: Likewise.
26904 * config/riscv/generic.md: Likewise.
26905 * config/riscv/linux.h: Likewise.
26906 * config/riscv/multilib-generator: Likewise.
26907 * config/riscv/peephole.md: Likewise.
26908 * config/riscv/pic.md: Likewise.
26909 * config/riscv/predicates.md: Likewise.
26910 * config/riscv/riscv-builtins.c: Likewise.
26911 * config/riscv/riscv-c.c: Likewise.
26912 * config/riscv/riscv-ftypes.def: Likewise.
26913 * config/riscv/riscv-modes.def: Likewise.
26914 * config/riscv/riscv-opts.h: Likewise.
26915 * config/riscv/riscv-protos.h: Likewise.
26916 * config/riscv/riscv.h: Likewise.
26917 * config/riscv/riscv.md: Likewise.
26918 * config/riscv/riscv.opt: Likewise.
26919 * config/riscv/sync.md: Likewise.
26920 * config/riscv/t-elf-multilib: Likewise.
26921 * config/riscv/t-linux: Likewise.
26922 * config/riscv/t-linux-multilib: Likewise.
26923 * config/riscv/t-riscv: Likewise.
26924 * configure.ac: Likewise.
26925 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
26926 Waterman as RISC-V maintainers.
26927 * doc/install.texi: Add RISC-V entries.
26928 * doc/invoke.texi: Add RISC-V options section.
26929 * doc/md.texi: Add RISC-V constraints section.
26930 * configure: Regenerated.
26932 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
26935 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
26936 false values to be constant vectors with all 0 or all 1 bits set.
26937 (vcondu<mode><mode>): Likewise.
26938 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
26940 (fpmask_comparison_operator): Update comment.
26941 (vecint_comparison_operator): New predicate.
26942 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
26943 vector conditionals when the true and false values are constant
26944 vectors with all 0 bits or all 1 bits set.
26946 2017-02-06 Martin Sebor <msebor@redhat.com>
26948 PR tree-optimization/79376
26949 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
26951 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
26953 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
26954 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
26955 to simplify split condition.
26957 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26959 * omp-expand.c (oxpand_omp_atomic_fetch_op,
26960 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
26963 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
26965 PR rtl-optimization/68664
26966 * target.def (can_speculate_insn): New hook.
26967 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
26968 * doc/tm.texi: Regenerate.
26969 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
26970 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
26971 (rs6000_sched_can_speculate_insn): New function.
26973 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26975 PR tree-optimization/79284
26976 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
26977 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
26978 vectorizable_mask_load_store, vectorizable_operation,
26979 vect_is_simple_cond, get_same_sized_vectype): Use it instead
26980 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
26981 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
26982 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
26983 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
26984 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
26985 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
26986 is_gimple_assign (stmt). Replace another such test with
26987 is_gimple_assign (stmt).
26989 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26992 * config/avr/avr.c (rtl-iter.h): Include it.
26993 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
26994 (avr_legitimate_combined_insn): ...and implementation.
26996 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
26998 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
26999 * config/s390/s390.c (s390_const_operand_ok)
27000 (s390_canonicalize_comparison, s390_extract_part)
27001 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
27002 (s390_contiguous_bitmask_p, s390_rtx_costs)
27003 (legitimize_pic_address): Likewise.
27004 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
27005 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
27006 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
27007 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
27008 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
27010 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27012 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
27013 REGNO($0) == REGNO($1).
27015 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27017 * config/s390/linux.h(SIZE_TYPE): Add comment.
27019 2017-02-06 Julian Brown <julian@codesourcery.com>
27020 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27021 Virendra Pathak <virendra.pathak@broadcom.com>
27023 * config/aarch64/aarch64-cores.def: Change the scheduler
27025 * config/aarch64/aarch64.md: Include thunderx2t99.md.
27026 * config/aarch64/thunderx2t99.md: New file.
27028 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27030 * doc/standards.texi (Go Language): Update link to language
27033 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27035 * tree-eh.c (lower_resx): Sanitize profile.
27036 (cleanup_empty_eh_move_lp): Likewise.
27038 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27041 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
27043 * cfgloopmanip.h (loop_version): Update prototype.
27044 * modulo-sched.c (sms_schedule): Update call of loop_version.
27045 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
27046 * tree-parloops.c (gen_parallel_loop): Likewise.
27047 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
27048 * tree-ssa-loop-split.c (split_loop): Likewise.
27049 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
27050 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
27052 2017-02-05 Martin Liska <mliska@suse.cz>
27055 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
27057 (print_operand_address): Initialize a struct to zero.
27059 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27061 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
27062 garbage collector only in textual form.
27064 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27066 * doc/extend.texi (x86 specific memory model extensions for
27067 transactional memory): Simplify a phrase.
27069 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
27072 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
27073 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
27074 (atomic_storedi_1): Likewise.
27076 2017-02-04 Jakub Jelinek <jakub@redhat.com>
27078 PR tree-optimization/79338
27079 * tree-parloops.c (gather_scalar_reductions): Don't call
27080 vect_analyze_loop_form for loop->inner before destroying loop's
27083 2017-02-03 Martin Sebor <msebor@redhat.com>
27085 PR tree-optimization/79327
27086 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
27087 when precision has resulted in leading zeros.
27088 (format_integer): Adjust the likely counter to assume an unknown
27089 argument that may be zero is non-zero.
27091 2017-02-03 Jason Merrill <jason@redhat.com>
27094 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
27095 avoid copying non-taken branch.
27097 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27099 PR tree-optimization/79340
27100 * tree-vect-loop.c (vectorizable_reduction): Release
27101 vec_defs elements after safe_splicing them into other vectors.
27104 PR tree-optimization/79327
27105 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
27106 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
27108 (format_integer): Use wide_int_to_tree instead of build_int_cst
27109 + to_?hwi. If argmin is NULL, just set argmin and argmax to
27110 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
27111 of shortest and longest sequence.
27113 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
27115 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
27116 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
27118 2017-02-03 Walter Lee <walt@tilera.com>
27121 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
27122 after initial stackframe link reg save.
27123 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
27125 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27128 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
27129 wu for stxssp alternative.
27131 2017-02-03 Martin Sebor <msebor@redhat.com>
27133 PR tree-optimization/79352
27134 * gimple-fold.c (get_range_strlen): Add argument.
27135 (get_range_strlen): Change return type to bool.
27136 (get_maxval_strlen): Pass in a dummy argument.
27137 * gimple-fold.h (get_range_strlen): Change return type to bool.
27138 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
27139 * tree.h (array_at_struct_end_p): Add argument.
27140 * tree.c (array_at_struct_end_p): Handle it.
27142 2017-02-03 Martin Liska <mliska@suse.cz>
27145 * multiple_target.c (create_dispatcher_calls): Redirect edge
27146 from a caller of a dispatcher.
27147 (expand_target_clones): Make the clones local.
27148 (ipa_target_clone): Do both target clones and resolvers.
27149 (ipa_dispatcher_calls): Remove the pass.
27150 (pass_dispatcher_calls::gate): Likewise.
27151 (make_pass_dispatcher_calls): Likewise.
27152 * passes.def (pass_target_clone): Put as very first IPA early
27155 2017-02-03 Martin Liska <mliska@suse.cz>
27157 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
27158 in case of a function with ifunc attribute.
27160 2017-02-03 Martin Liska <mliska@suse.cz>
27162 * cgraph.c (cgraph_node::dump): Dump function version info.
27163 * symtab.c (symtab_node::dump_base): Add missing new line.
27165 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27167 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
27168 (ifcombine_ifandif): Use it.
27170 2017-02-03 Martin Liska <mliska@suse.cz>
27172 * doc/invoke.texi: Document default value for
27173 use-after-scope-direct-emission-threshold.
27175 2017-02-03 Martin Liska <mliska@suse.cz>
27177 PR tree-optimization/79339
27178 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
27179 (format_floating): Likewise.
27181 2017-02-03 Martin Liska <mliska@suse.cz>
27184 * ipa-prop.c (ipa_node_params_t::insert): Remove current
27186 (ipa_node_params_t::remove): Likewise.
27187 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
27188 initialization from removed ipa_node_params_t::insert.
27189 (ipa_node_params::~ipa_node_params): Move from removed
27190 ipa_node_params_t::release.
27191 * symbol-summary.h (symbol_summary::m_released): New member.
27192 Do not release a summary twice. Do not allow to call finalizer
27193 for types of a summary that live in GGC memory.
27195 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27197 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
27200 2017-02-02 Martin Sebor <msebor@redhat.com>
27202 PR middle-end/79275
27203 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
27204 (format_string): Tighten up the range of output for non-constant
27205 strings and correct the expected range for wide non-constant strings.
27207 2017-02-02 Martin Sebor <msebor@redhat.com>
27209 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
27211 PR middle-end/32003
27212 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
27214 (-fdump-tree-@var): Add to index and document how to come up
27215 with pass-specific option and dump file names.
27216 (-fdump-passes): Clarify where to look for output.
27218 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27220 PR middle-end/77445
27221 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
27222 statistics of the analyzed path; allow threading for speed when
27223 any of BBs along the path are optimized for speed.
27225 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
27227 PR middle-end/78468
27228 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
27229 settings of the virtual registers.
27232 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
27234 * explow.c (get_dynamic_stack_size): Take known alignment of stack
27235 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
27238 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27240 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
27241 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
27243 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27245 * config/s390/s390.md: Add missing comments with the expanded
27247 * config/s390/vector.md: Likewise.
27248 * config/s390/vx-builtins.md: Likewise.
27250 2017-02-02 Jakub Jelinek <jakub@redhat.com>
27253 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
27254 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
27255 conditions on a single line.
27257 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27259 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27260 __S390_VX__ to __VX__.
27262 2017-02-01 Andrew Pinski <apinski@cavium.com>
27264 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
27265 stmt_info to record_stmt_cost.
27266 (vect_get_known_peeling_cost): Pass stmt_info if known to
27268 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
27269 cpu_vector_cost field into
27270 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
27271 field into vec_int_stmt_cost and vec_fp_stmt_cost.
27272 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
27273 splitting of scalar_stmt_cost and vec_stmt_cost.
27274 (thunderx_vector_cost): Likewise.
27275 (cortexa57_vector_cost): LIkewise.
27276 (exynosm1_vector_cost): Likewise.
27277 (xgene1_vector_cost): Likewise.
27278 (thunderx2t99_vector_cost): Improve after the splitting of the two
27280 (aarch64_builtin_vectorization_cost): Update for the splitting of
27281 scalar_stmt_cost and vec_stmt_cost.
27283 2017-02-01 Torvald Riegel <triegel@redhat.com>
27284 Richard Henderson <rth@redhat.com>
27286 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
27287 conditional on existance of a fast atomic load.
27288 * optabs-query.c (can_atomic_load_p): New function.
27289 * optabs-query.h (can_atomic_load_p): Declare it.
27290 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
27291 no fast atomic load is available for the particular size of access.
27292 (expand_atomic_compare_and_swap): Likewise.
27293 (expand_atomic_load): Likewise.
27294 (expand_atomic_store): Likewise.
27295 (expand_atomic_fetch_op): Likewise.
27296 * testsuite/lib/target-supports.exp
27297 (check_effective_target_sync_int_128): Remove x86 because it provides
27298 no fast atomic load.
27299 (check_effective_target_sync_int_128_runtime): Likewise.
27301 2017-02-01 Richard Biener <rguenther@suse.de>
27303 * graphite.c: Include tree-vectorizer.h for find_loop_location.
27304 (graphite_transform_loops): Provide opt-info for optimized nests.
27305 * tree-parloop.c (parallelize_loops): Provide opt-info for
27306 parallelized loops.
27308 2017-02-01 Richard Biener <rguenther@suse.de>
27310 PR middle-end/79315
27311 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
27312 was not set before.
27314 2017-02-01 Richard Biener <rguenther@suse.de>
27316 PR tree-optimization/71824
27317 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27318 Verify the loops are valid in the merged SESE region.
27319 (scop_detection::can_represent_loop_1): Check analyzing the
27320 evolution of the number of iterations in the region succeeds.
27322 2017-01-31 Ian Lance Taylor <iant@golang.org>
27324 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
27325 REG_ARGS_SIZE note to 32-bit push insns and call insn.
27327 2017-01-31 David Malcolm <dmalcolm@redhat.com>
27329 PR preprocessor/79210
27330 * input.c (get_substring_ranges_for_loc): Replace line_width
27331 assertion with error-handling.
27333 2017-01-31 Richard Biener <rguenther@suse.de>
27335 PR tree-optimization/77318
27336 * graphite-sese-to-poly.c (extract_affine): Fix assert.
27337 (create_pw_aff_from_tree): Take loop parameter.
27338 (add_condition_to_pbb): Pass loop of the condition to
27339 create_pw_aff_from_tree.
27341 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27343 * config/s390/s390.c (s390_asan_shadow_offset): New function.
27344 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
27346 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
27350 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
27352 (convert_int_to_float128): Likewise.
27353 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
27354 (convert_int_to_float128): Likewise.
27355 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
27356 (UNSPEC_IEEE128_CONVERT): Likewise.
27357 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
27358 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
27359 Use local variables for IBM extended format.
27360 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
27361 (fix_trunc<mode>si2_fprs): Likewise.
27362 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
27363 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
27364 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
27365 to know that we can now have integers of all sizes in vector
27367 (fix<uns>_<mode>di2_hw): Likewise.
27368 (float<uns>_<mode>si2_hw): Likewise.
27369 (fix_<mode>si2_hw): Likewise.
27370 (fixuns_<mode>si2_hw): Likewise.
27371 (float<uns>_<mode>di2_hw): Likewise.
27372 (float_<mode>di2_hw): Likewise.
27373 (float_<mode>si2_hw): Likewise.
27374 (floatuns_<mode>di2_hw): Likewise.
27375 (floatuns_<mode>si2_hw): Likewise.
27376 (xscvqp<su>wz_<mode>): Delete, no longer used.
27377 (xscvqp<su>dz_<mode>): Likewise.
27378 (xscv<su>dqp_<mode>): Likewise.
27379 (ieee128_mfvsrd_64bit): Likewise.
27380 (ieee128_mfvsrd_32bit): Likewise.
27381 (ieee128_mfvsrwz): Likewise.
27382 (ieee128_mtvsrw): Likewise.
27383 (ieee128_mtvsrd_64bit): Likewise.
27384 (ieee128_mtvsrd_32bit): Likewise.
27386 2017-01-31 Martin Liska <mliska@suse.cz>
27389 * ipa-prop.c (ipa_free_all_node_params): Call release method
27390 instead of ~sumbol_summary to not to trigger double times
27393 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
27395 PR tree-optimization/71691
27396 * bitmap.h (class auto_bitmap): New.
27397 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
27398 is_maybe_undefined instead of ssa_undefined_value_p.
27400 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27402 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27403 __S390_ARCH_LEVEL__ to __ARCH__.
27405 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27407 PR tree-optimization/79267
27408 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
27409 if should_remove_lhs_p is true.
27411 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
27414 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
27415 (add_alignment_attribute): New.
27416 (base_type_die): Add alignment attribute.
27417 (subrange_type_die): Likewise.
27418 (modified_type_die): Likewise.
27419 (gen_array_type_die): Likewise.
27420 (gen_descr_array_type_die: Likewise.
27421 (gen_enumeration_type_die): Likewise.
27422 (gen_subprogram_die): Likewise.
27423 (gen_variable_die): Likewise.
27424 (gen_field_die): Likewise.
27425 (gen_ptr_to_mbr_type_die): Likewise.
27426 (gen_struct_or_union_type_die): Likewise.
27427 (gen_subroutine_type_die): Likewise.
27428 (gen_typedef_die): Likewise.
27429 (base_type_cmp): Compare alignment attribute.
27431 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27434 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
27435 (setb_unsigned) New pattern for setb with CCUNS.
27436 * config/rs6000/rs6000.c (expand_block_compare): Use a different
27437 subfc./subfe sequence to avoid overflow problems. Generate a
27438 shorter sequence with cmpld/setb for power9.
27439 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
27440 for generating subfc. instruction.
27441 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
27442 now uses this instruction.
27444 2017-01-30 Ian Lance Taylor <iant@google.com>
27447 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
27448 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
27450 2017-01-30 Martin Sebor <msebor@redhat.com>
27452 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
27453 Move constant to the right of a relational operator.
27454 (get_mpfr_format_length, format_character, format_string): Ditto.
27455 (should_warn_p, maybe_warn): Same.
27457 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
27459 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
27462 * asan.c (get_translation_unit_decl): Remove function.
27463 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
27465 2017-01-30 Martin Liska <mliska@suse.cz>
27467 PR gcov-profile/79259
27468 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
27469 -fprofile-generate.
27471 2017-01-30 Martin Liska <mliska@suse.cz>
27474 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
27475 Initialize variables with NULL value.
27477 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
27480 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
27482 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
27484 2017-01-30 Richard Biener <rguenther@suse.de>
27486 PR tree-optimization/79276
27487 * tree-vrp.c (process_assert_insertions): Properly adjust common
27488 when removing a duplicate.
27490 * gcc.dg/torture/pr79276.c: New testcase.
27492 2017-01-30 Richard Biener <rguenther@suse.de>
27494 PR tree-optimization/79256
27495 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27496 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27498 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
27500 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
27503 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
27504 ("*r<noxa>sbg_<mode>_sll_bitmask")
27505 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
27506 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
27507 Use contiguous_bitmask_nowrap_operand.
27509 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27512 * config/rs6000/altivec.h (vec_xl): Revise #define.
27513 (vec_xst): Likewise.
27515 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
27517 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
27519 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
27521 PR rtl-optimization/79194
27522 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
27523 traps before call to bypass_conditional_jumps.
27525 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
27527 PR tree-optimization/71374
27528 * lra-constraints.c (check_conflict_input_operands): New.
27529 (match_reload): Use it.
27531 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
27534 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
27535 account to calculate conflict_set.
27537 2017-01-27 Bin Cheng <bin.cheng@arm.com>
27539 PR rtl-optimization/78559
27540 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
27541 other_insn in combine.
27543 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27545 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
27546 uint16_type_node for BT_UINT16.
27548 2017-01-27 David Malcolm <dmalcolm@redhat.com>
27550 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
27551 "RTL Tests" to menu.
27552 (GIMPLE Tests): New node.
27553 (RTL Tests): New node.
27555 2017-01-27 Richard Biener <rguenther@suse.de>
27557 PR tree-optimization/79245
27558 * tree-loop-distribution.c (distribute_loop): Apply cost
27559 modeling also to detected patterns.
27561 2017-01-27 Richard Biener <rguenther@suse.de>
27563 PR tree-optimization/71433
27564 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
27565 (compare_assert_loc): New function.
27566 (process_assert_insertions): Sort and optimize assert locations
27567 to remove duplicates and push down identical assertions on
27568 edges to their destination block.
27570 2017-01-27 Richard Biener <rguenther@suse.de>
27572 PR tree-optimization/79244
27573 * tree-vrp.c (remove_range_assertions): Forcefully propagate
27574 out SSA names even if abnormal.
27576 2017-01-27 Jakub Jelinek <jakub@redhat.com>
27578 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
27579 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
27580 instead of MPFR_RNDN.
27582 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
27585 * arm.c (arm_option_override): Don't call build_target_option_node
27586 until after doing all option overrides.
27587 (arm_valid_target_attribute_tree): Likewise.
27589 2017-01-27 Martin Liska <mliska@suse.cz>
27591 * doc/invoke.texi (-fprofile-arcs): Document profiling support
27592 for {cd}tors and C++ {cd}tors.
27594 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
27596 * config/s390/s390.md ("*setmem_long_and")
27597 ("*setmem_long_and_31z"): Use zero_extend instead of and.
27599 2017-01-26 Martin Sebor <msebor@redhat.com>
27601 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
27604 2017-01-26 Martin Sebor <msebor@redhat.com>
27606 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
27607 HAVE_DFmode before using XFmode or DFmode.
27608 (parse_directive): Avoid using the z length modifier to avoid
27609 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
27611 PR middle-end/78703
27612 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
27613 to accept adjustment as an array.
27614 (get_int_range): New function.
27615 (struct directive): Make width and prec arrays.
27616 (directive::set_width, directive::set_precision): Call get_int_range.
27617 (format_integer, format_floating): Handle width and precision ranges.
27618 (format_string, parse_directive): Same.
27620 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27623 * dwarf2out.c (generate_skeleton_bottom_up): For children with
27624 comdat_type_p set, just clone them, but keep the children in the
27628 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
27629 which have direct callers with -fvar-tracking-assignments enabled
27631 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
27632 inside of type units.
27634 2017-01-26 Martin Sebor <msebor@redhat.com>
27636 PR middle-end/78703
27637 * gimple-ssa-sprintf.c (struct result_range): Add likely and
27639 (struct format_result): Replace number_chars, number_chars_min,
27640 and number_chars_max with a single member of struct result_range.
27642 (format_result::operator+=): Adjust.
27643 (struct fmtresult): Remove bounded. Handle likely and unlikely
27645 (fmtresult::adjust_for_width_or_precision): New function.
27646 (fmtresult:type_max_digits): New function.
27647 (bytes_remaining): Handle likely and unlikely counters.
27648 (min_bytes_remaining): Remove.
27649 (format_percent): Simplify.
27650 (format_integer, format_floating): Set likely and unlikely counters.
27651 (get_string_length, format_character, format_string): Same.
27652 (format_plain, should_warn_p): New function.
27653 (maybe_warn): Call should_warn_p. Update diagnostic messages
27654 and handle those for all directives, including plain strings.
27655 (format_directive): Handle likely and unlikely counters.
27656 Remove unnecessary quoting from diagnostics. Add an informational
27658 (add_bytes): Remove.
27659 (pass_sprintf_length::compute_format_length): Simplify.
27660 (try_substitute_return_value): Handle likely and unlikely counters.
27662 2017-01-26 Carl Love <cel@us.ibm.com>
27664 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
27665 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
27667 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
27670 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
27671 endianess for subregs into account.
27672 * lra-constraints.c (lra_constraints): Do risky transformations
27673 always on the first iteration.
27674 * lra-lives.c (check_pseudos_live_through_calls): Add arg
27675 last_call_used_reg_set.
27676 (process_bb_lives): Define and use last_call_used_reg_set.
27677 * lra.c (lra): Always continue after lra_constraints on the first
27680 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
27682 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
27684 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
27686 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27688 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
27689 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
27690 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
27691 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
27692 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
27693 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
27694 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
27695 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
27696 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
27698 2017-01-26 Marek Polacek <polacek@redhat.com>
27701 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
27702 for the third operand.
27704 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27706 PR middle-end/79236
27707 * omp-low.c (struct omp_context): Add simt_stmt field.
27708 (scan_omp_for): Return omp_context *.
27709 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
27710 context to the _simt_ SIMD stmt.
27711 (lower_omp_for): For combined SIMD with sibling _simt_
27712 SIMD, make sure to use the same decls in _looptemp_
27713 clauses as in the sibling.
27715 2017-01-26 David Sherwood <david.sherwood@arm.com>
27717 PR middle-end/79212
27718 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
27721 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27724 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
27725 emit fld b; fld a; if possible.
27727 * brig-builtins.def: Update copyright years.
27728 * config/arm/arm_acle_builtins.def: Update copyright years.
27730 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
27733 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
27734 constraint instead of o for the stxsd instruction.
27736 2017-01-25 Carl Love <cel@us.ibm.com>
27738 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
27739 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
27741 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
27743 * doc/invoke.texi (C++ Dialect Options): Fix typo.
27745 2017-01-25 Richard Biener <rguenther@suse.de>
27747 PR tree-optimization/69264
27748 * target.def (vector_alignment_reachable): Improve documentation.
27749 * doc/tm.texi: Regenerate.
27750 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
27752 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
27753 earlier changes with respect to TYPE_USER_ALIGN.
27754 (vector_alignment_reachable_p): Likewise. Improve dumping.
27756 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27759 * config/arm/arm.md (xordi3): Force constant operand into a register
27762 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27764 * doc/invoke.texi (-fstore-merging): Correct default optimization
27765 levels at which it is enabled.
27766 (-O): Move -fstore-merging from list to...
27769 2017-01-25 Richard Biener <rguenther@suse.de>
27772 * omp-expand.c: Include debug.h.
27773 (expand_omp_taskreg): Make sure to generate early debug before
27774 outlining anything from a function.
27775 (expand_omp_target): Likewise.
27776 (grid_expand_target_grid_body): Likewise.
27778 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
27781 * asan.c (get_translation_unit_decl): New function.
27782 (asan_add_global): Extract modules file name from globals
27783 TRANSLATION_UNIT_DECL name.
27785 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
27788 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
27789 for long calls with APCS frame and VFP.
27791 2017-01-24 David Malcolm <dmalcolm@redhat.com>
27793 * cfg.c (original_copy_tables_initialized_p): New function.
27794 * cfg.h (original_copy_tables_initialized_p): New decl.
27795 * cfgrtl.c (relink_block_chain): Guard the call to
27796 free_original_copy_tables with a call to
27797 original_copy_tables_initialized_p.
27798 * cgraph.h (symtab_node::native_rtl_p): New decl.
27799 * cgraphunit.c (symtab_node::native_rtl_p): New function.
27800 (symtab_node::needed_p): Don't assert for early assembly output
27801 for __RTL functions.
27802 (cgraph_node::finalize_function): Set "force_output" for __RTL
27804 (cgraph_node::analyze): Bail out early for __RTL functions.
27805 (analyze_functions): Update assertion to support __RTL functions.
27806 (cgraph_node::expand): Bail out early for __RTL functions.
27807 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
27809 * function.h (struct function): Update comment for field
27811 * gimple-expr.c: Include "tree-pass.h".
27812 (gimple_has_body_p): Return false for __RTL functions.
27813 * Makefile.in (OBJS): Add run-rtl-passes.o.
27814 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
27816 (gcc::pass_manager::get_clean_slate): New accessor.
27817 * passes.c: Include "insn-addr.h".
27818 (should_skip_pass_p): Add logging. Update logic for running
27819 "expand" to be compatible with both __GIMPLE and __RTL. Guard
27820 property-provider override so it is only done for gimple passes.
27822 (skip_pass): New function.
27823 (execute_one_pass): Call skip_pass when skipping passes.
27824 * read-md.c (md_reader::read_char): Support filtering
27825 the input to a subset of line numbers.
27826 (md_reader::md_reader): Initialize fields
27827 m_first_line and m_last_line.
27828 (md_reader::read_file_fragment): New function.
27829 * read-md.h (md_reader::read_file_fragment): New decl.
27830 (md_reader::m_first_line): New field.
27831 (md_reader::m_last_line): New field.
27832 * read-rtl-function.c (function_reader::create_function): Only
27833 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
27834 curr_properties. Set DECL_INITIAL to a dummy block.
27835 (read_rtl_function_body_from_file_range): New function.
27836 * read-rtl-function.h (read_rtl_function_body_from_file_range):
27838 * run-rtl-passes.c: New file.
27839 * run-rtl-passes.h: New file.
27841 2017-01-24 Jeff Law <law@redhat.com>
27843 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
27846 2017-01-24 Bin Cheng <bin.cheng@arm.com>
27848 PR tree-optimization/79159
27849 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
27850 (record_nonwrapping_iv): Improve boundary using above function if no
27851 value range information.
27853 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
27854 Martin Jambor <mjambor@suse.cz>
27856 * brig-builtins.def: New file.
27857 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
27858 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
27859 (DEF_HSAIL_SAT_BUILTIN): Likewise.
27860 (DEF_HSAIL_INTR_BUILTIN): Likewise.
27861 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
27862 * builtin-types.def (BT_INT8): New.
27863 (BT_INT16): Likewise.
27864 (BT_UINT8): Likewise.
27865 (BT_UINT16): Likewise.
27866 (BT_FN_ULONG): Likewise.
27867 (BT_FN_UINT_INT): Likewise.
27868 (BT_FN_UINT_ULONG): Likewise.
27869 (BT_FN_UINT_LONG): Likewise.
27870 (BT_FN_UINT_PTR): Likewise.
27871 (BT_FN_ULONG_PTR): Likewise.
27872 (BT_FN_INT8_FLOAT): Likewise.
27873 (BT_FN_INT16_FLOAT): Likewise.
27874 (BT_FN_UINT32_FLOAT): Likewise.
27875 (BT_FN_UINT16_FLOAT): Likewise.
27876 (BT_FN_UINT8_FLOAT): Likewise.
27877 (BT_FN_UINT64_FLOAT): Likewise.
27878 (BT_FN_UINT16_UINT32): Likewise.
27879 (BT_FN_UINT32_UINT16): Likewise.
27880 (BT_FN_UINT16_UINT16_UINT16): Likewise.
27881 (BT_FN_INT_PTR_INT): Likewise.
27882 (BT_FN_UINT_PTR_UINT): Likewise.
27883 (BT_FN_LONG_PTR_LONG): Likewise.
27884 (BT_FN_ULONG_PTR_ULONG): Likewise.
27885 (BT_FN_VOID_UINT64_UINT64): Likewise.
27886 (BT_FN_UINT8_UINT8_UINT8): Likewise.
27887 (BT_FN_INT8_INT8_INT8): Likewise.
27888 (BT_FN_INT16_INT16_INT16): Likewise.
27889 (BT_FN_INT_INT_INT): Likewise.
27890 (BT_FN_UINT_FLOAT_UINT): Likewise.
27891 (BT_FN_FLOAT_UINT_UINT): Likewise.
27892 (BT_FN_ULONG_UINT_UINT): Likewise.
27893 (BT_FN_ULONG_UINT_PTR): Likewise.
27894 (BT_FN_ULONG_ULONG_ULONG): Likewise.
27895 (BT_FN_UINT_UINT_UINT): Likewise.
27896 (BT_FN_VOID_UINT_PTR): Likewise.
27897 (BT_FN_UINT_UINT_PTR: Likewise.
27898 (BT_FN_UINT32_UINT64_PTR): Likewise.
27899 (BT_FN_INT_INT_UINT_UINT): Likewise.
27900 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
27901 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
27902 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
27903 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
27904 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
27905 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
27906 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
27907 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
27908 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
27909 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
27910 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
27911 * doc/frontends.texi: List BRIG FE.
27912 * doc/install.texi (Testing): Add BRIG tesring requirements.
27913 * doc/invoke.texi (Overall Options): Mention BRIG.
27914 * doc/standards.texi (Standards): Doucment BRIG HSA version.
27916 2017-01-24 Richard Biener <rguenther@suse.de>
27918 PR translation/79208
27919 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
27921 2017-01-24 Martin Jambor <mjambor@suse.cz>
27924 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
27925 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
27926 and known_contexts.
27928 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
27930 PR middle-end/79123
27931 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
27932 casts from signed to unsigned really don't have a range.
27934 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
27936 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
27937 GMP_RNDx for compatiblity.
27939 2017-01-24 Martin Liska <mliska@suse.cz>
27942 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
27943 that would prevent us to call alloca with -1 as argument.
27945 2017-01-24 Jakub Jelinek <jakub@redhat.com>
27947 * dwarf2out.c (output_compilation_unit_header, output_file_names):
27948 Avoid -Wformat-security warning.
27950 2017-01-23 Andrew Pinski <apinski@cavium.com>
27952 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
27955 2017-01-23 Martin Sebor <msebor@redhat.com>
27957 PR middle-end/78703
27958 * gimple-ssa-sprintf.c (warn_level): New global.
27959 (format_integer): Use it here and throughout the rest of the file.
27960 Use the same switch to compute sign as base.
27961 (maybe_warn): New function.
27962 (format_directive): Factor out warnings into maybe_warn.
27963 Add debugging output. Use warn_level.
27964 (add_bytes): Use warn_level.
27965 (pass_sprintf_length::compute_format_length): Add debugging output.
27966 (try_substitute_return_value): Same.
27967 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
27969 PR middle-end/78703
27970 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
27971 (struct fmtresult, format_integer, format_floating): Adjust.
27972 (fmtresult::fmtresult): Set max correctly in two argument ctor.
27973 (get_string_length, format_string,format_directive): Same.
27974 (pass_sprintf_length::compute_format_length): Same.
27975 (try_substitute_return_value): Simplify slightly.
27977 PR middle-end/78703
27978 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
27979 (fmtresult::operator+=): Outlined.
27980 (struct fmtresult): Add ctors.
27981 (struct conversion_spec): Rename...
27982 (struct directive): ...to this. Add and remove data members.
27983 (directive::set_width, directive::set_precision): New functions.
27984 (format_percent): Use fmtresult ctor.
27985 (get_width_and_precision): Remove.
27986 (format_integer): Make naming changes. Avoid computing width and
27988 (format_floating): Same. Adjust indentation.
27989 (format_character, format_none): New functions.
27990 (format_string): Moved character handling to format_character.
27991 (format_directive): Remove arguments, change return type.
27992 (parse_directive): New function.
27993 (pass_sprintf_length::compute_format_length): Move directive
27994 parsing to parse_directive.
27996 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27998 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
27999 (assign_assembler_name_if_needed): ... this.
28000 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
28001 (assign_assembler_name_if_needed): ... this.
28002 (free_lang_data_in_cgraph): Adjust callers.
28003 * cgraphunit.c (cgraph_node::analyze): Likewise.
28004 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
28007 2017-01-23 Richard Biener <rguenther@suse.de>
28009 PR tree-optimization/79088
28010 PR tree-optimization/79188
28011 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
28012 resetting loop bounds after last path deletion. Reset loop
28013 bounds of the target loop, make code match the comments.
28014 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
28015 Make sure loops need no fixups.
28017 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28019 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
28020 exponent support with double type for first argument.
28021 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
28022 type returned by __builtin_vec_extract_sig,
28023 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
28024 functions from "vector int" to "vector unsigned int" or from
28025 "vector long long int" to "vector unsigned long long int".
28026 Changed type returned by __builtin_vec_extract_exp,
28027 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
28028 functions from "vector int" to "vector unsigned int" or from
28029 "vector long long int" to "vector unsigned long long int".
28030 Changed return type of __builtin_vec_test_data_class,
28031 __builtin_vec_test_data_class_sp, and
28032 __builtin_vec_test_data_class_dp from "vector int" to
28033 "vector bool int" or from "vector long long int" to "vector bool
28034 long long int" and changed second argument type from "unsigned
28035 int" to "int". Added new overloaded function forms "vector float
28036 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
28037 "vector float __builtin_vec_insert_exp_sp (vector float, vector
28038 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
28039 double, vector unsigned long long int)" and "vector double
28040 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
28041 long int)". Changed return type of
28042 __builtin_scalar_test_data_class and
28043 __builtin_scalar_test_data_class_sp and
28044 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
28045 int" and changed second argument from "unsigned int" to "int".
28046 Changed type returned by __builtin_scalar_test_neg,
28047 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
28048 from "int" to "bool int". Added new overloaded function form
28049 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
28050 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
28051 exponent double-precision with floating point first argument.
28052 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
28053 documentation of scalar_test_data_class, scalar_test_neg,
28054 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
28055 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
28056 vec_test_data_class built-in functions to reflect refinements in
28057 their type signatures.
28059 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
28061 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
28063 (aarch64_elf_asm_destructor): Likewise.
28065 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
28067 PR rtl-optimization/78634
28068 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
28069 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
28070 * ifcvt.c (noce_try_cmove): Add missing cost check.
28072 PR rtl-optimization/71724
28073 * combine.c (if_then_else_cond): Look for situations where it is
28074 beneficial to undo the work of one of the recursive calls.
28076 2017-01-23 Bin Cheng <bin.cheng@arm.com>
28078 PR tree-optimization/70754
28079 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
28080 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
28081 combined stmt before it if not NULL.
28082 (combine_chains): Process refs reversely and compute dominance point
28085 2017-01-23 Martin Liska <mliska@suse.cz>
28087 PR tree-optimization/79196
28088 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
28089 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
28091 (strlen_optimize_stmt): Call the renamed function.
28093 2017-01-23 Michael Matz <matz@suse.de>
28095 PR tree-optimization/78384
28096 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
28098 2017-01-23 Richard Biener <rguenther@suse.de>
28100 PR tree-optimization/79186
28101 * tree-vrp.c (register_new_assert_for): Make sure we've seen
28102 both incoming edges before moving an assert.
28104 2017-01-23 Martin Jambor <mjambor@suse.cz>
28106 * ipa-prop.c (load_from_param_1): Removed.
28107 (load_from_unmodified_param): Bits from load_from_param_1 put back
28109 (load_from_param): Removed.
28110 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
28111 with stmt. Reverted back to use of load_from_unmodified_param.
28113 2017-01-23 Martin Jambor <mjambor@suse.cz>
28116 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
28117 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
28118 field a pointer to garbage collected vector, mark lattices and
28119 ipcp_orig_node with GTY((skip)).
28120 (ipa_get_param_count): Adjust to descriptors being a pointer.
28121 (ipa_get_param): Likewise.
28122 (ipa_get_type): Likewise.
28123 (ipa_get_param_move_cost): Likewise.
28124 (ipa_set_param_used): Likewise.
28125 (ipa_get_controlled_uses): Likewise.
28126 (ipa_set_controlled_uses): Likewise.
28127 (ipa_is_param_used): Likewise.
28128 (ipa_node_params_t): Move into garbage collector. New methods insert
28130 (ipa_node_params_sum): Annotate wth GTY(()).
28131 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
28133 (ipa_load_from_parm_agg): Adjust declaration.
28134 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
28135 * ipa-profile.c (ipa_profile): Likewise.
28136 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
28137 (ipa_populate_param_decls): Make descriptors parameter garbage
28139 (ipa_dump_param): Adjust to descriptors being a pointer.
28140 (ipa_alloc_node_params): Likewise.
28141 (ipa_initialize_node_params): Likewise.
28142 (load_from_param_1): Make descriptors parameter garbage collected.
28143 (load_from_unmodified_param): Likewise.
28144 (load_from_param): Likewise.
28145 (ipa_load_from_parm_agg): Likewise.
28146 (ipa_node_params::~ipa_node_params): Removed.
28147 (ipa_free_all_node_params): Remove call to delete operator.
28148 (ipa_node_params_t::insert): New.
28149 (ipa_node_params_t::remove): Likewise.
28150 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
28151 copy known_csts and known_contexts vectors.
28152 (ipa_read_node_info): Adjust to descriptors being a pointer.
28153 (ipcp_modif_dom_walker): Make m_descriptors field garbage
28155 (ipcp_transform_function): Make descriptors variable garbage
28158 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
28160 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
28161 * config/i386/avx512dqintrin.h: Ditto.
28162 * config/i386/avx512fintrin.h: Ditto.
28163 * config/i386/i386.c: Handle new builtins.
28164 * config/i386/i386-builtin.def: Add new builtins.
28165 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
28166 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
28168 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28169 Martin Liska <mliska@suse.cz>
28171 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
28172 * asan.c (asan_expand_poison_ifn): Support stores and use
28173 appropriate ASAN report function.
28174 * internal-fn.c (expand_ASAN_POISON_USE): New function.
28175 * internal-fn.def (ASAN_POISON_USE): Declare.
28176 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
28177 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
28178 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
28179 ASAN_POISON calls w/o LHS.
28180 * tree-ssa.c (execute_update_addresses_taken): Create clobber
28181 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
28182 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
28183 * gimplify.c (asan_poison_variables): Add attribute
28184 use_after_scope_memory to variables that really needs to live
28186 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
28187 having the attribute.
28189 2017-01-23 Martin Liska <mliska@suse.cz>
28191 * asan.c (create_asan_shadow_var): New function.
28192 (asan_expand_poison_ifn): Likewise.
28193 * asan.h (asan_expand_poison_ifn): New declaration.
28194 * internal-fn.c (expand_ASAN_POISON): Likewise.
28195 * internal-fn.def (ASAN_POISON): New builtin.
28196 * sanopt.c (pass_sanopt::execute): Expand
28197 asan_expand_poison_ifn.
28198 * tree-inline.c (copy_decl_for_dup_finish): Make function
28200 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
28201 * tree-ssa.c (is_asan_mark_p): New function.
28202 (execute_update_addresses_taken): Rewrite local variables
28203 (identified just by use-after-scope as addressable) into SSA.
28205 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28207 * doc/install.texi (Specific): opensource.apple.com uses https
28208 now. Remove trailing slash.
28210 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28212 * README.Portability: Remove note on an Irix compatibility issue.
28214 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
28216 * gcov.c (INCLUDE_ALGORITHM): Define.
28217 (INCLUDE_VECTOR): Define.
28218 No longer include <vector> and <algorithm> directly.
28220 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
28222 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
28224 * doc/invoke.texi (Code Gen Options): Ditto.
28226 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
28229 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
28231 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
28233 rtl-optimization/79125
28234 * cprop.c (local_cprop_pass): Handle cases where we make an
28235 unconditional trap.
28237 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
28241 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
28242 read from, for big endian.
28244 2017-01-20 Jiong Wang <jiong.wang@arm.com>
28246 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
28247 register pauth builtins for LP64 only.
28249 2017-01-20 Marek Polacek <polacek@redhat.com>
28252 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
28255 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28257 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
28259 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
28260 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
28261 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
28263 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28266 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
28267 in big-endian mode when they are not a single duplicated value.
28269 2017-01-20 Richard Biener <rguenther@suse.de>
28271 * BASE-VER: Bump to 7.0.1.
28273 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28275 * omp-low.c (omplow_simd_context): New struct. Use it...
28276 (lower_rec_simd_input_clauses): ...here and...
28277 (lower_rec_input_clauses): ...here to hold common data. Adjust all
28278 references to idx, lane, max_vf, is_simt.
28280 2017-01-20 Graham Markall <graham.markall@embecosm.com>
28282 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
28285 2017-01-20 Martin Jambor <mjambor@suse.cz>
28287 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
28288 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
28290 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
28291 (GTFILES): Rename hsa.c to hsa-common.c.
28292 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
28293 * hsa-dump.c: Likewise.
28294 * hsa-gen.c: Likewise.
28295 * hsa-regalloc.c: Likewise.
28296 * ipa-hsa.c: Likewise.
28297 * omp-expand.c: Likewise.
28298 * omp-low.c: Likewise.
28299 * toplev.c: Likewise.
28301 2017-01-20 Marek Polacek <polacek@redhat.com>
28304 * doc/invoke.texi: Document -Wduplicated-branches.
28305 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
28306 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
28307 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
28308 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
28309 return 0 only when not OEP_LEXICOGRAPHIC.
28310 (fold_build_cleanup_point_expr): Use the expression
28311 location when building CLEANUP_POINT_EXPR.
28312 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
28313 * tree.c (add_expr): Handle error_mark_node.
28315 2017-01-20 Martin Liska <mliska@suse.cz>
28318 * tree-profile.c (init_ic_make_global_vars): Do not call
28320 (gimple_init_gcov_profiler): Likewise.
28322 2017-01-20 Martin Liska <mliska@suse.cz>
28325 * cgraph.h (maybe_create_reference): Remove argument and
28327 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
28329 * ipa-cp.c (create_specialized_node): Likewise.
28330 * symtab.c (symtab_node::maybe_create_reference): Handle
28331 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
28333 2017-01-20 Martin Liska <mliska@suse.cz>
28335 * read-rtl-function.c (function_reader::create_function): Use
28336 build_decl instread of build_decl_stat.
28338 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
28340 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
28341 * config/i386/avx512dqintrin.h: Ditto.
28342 * config/i386/avx512fintrin.h: Ditto.
28343 * config/i386/i386-builtin-types.def: Add new types.
28344 * config/i386/i386.c: Handle new types.
28345 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
28346 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
28347 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
28348 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
28349 (__builtin_ia32_kshiftridi): New.
28350 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
28352 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
28356 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
28357 define to rs6000_init_stack_protect_guard.
28358 (rs6000_init_stack_protect_guard): New function.
28360 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28361 Yunqiang Su <yunqiang.su@imgtec.com>
28363 * config.gcc (supported_defaults): Add madd4.
28364 (with_madd4): Add validation.
28365 (all_defaults): Add madd4.
28366 * config/mips/mips.opt (mmadd4): New option.
28367 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28369 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28371 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
28372 (ISA_HAS_FUSED_MADD4): Likewise.
28373 * doc/invoke.texi (-mmadd4): Document the new option.
28374 * doc/install.texi (--with-madd4): Document the new option.
28376 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28378 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
28379 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
28380 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
28381 (aarch64_init_pauth_hint_builtins): New.
28382 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
28383 (aarch64_expand_builtin): Expand new builtins.
28385 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28387 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
28388 * combine-stack-adj.c (no_unhandled_cfa): Handle
28389 REG_CFA_TOGGLE_RA_MANGLE.
28390 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
28391 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
28392 info for return address signing.
28393 (aarch64_expand_epilogue): Likewise.
28395 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28397 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
28398 * config/aarch64/aarch64-protos.h
28399 (aarch64_return_address_signing_enabled): New declaration.
28400 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
28402 (aarch64_expand_prologue): Sign return address before it's pushed onto
28404 (aarch64_expand_epilogue): Authenticate return address fetched from
28406 (aarch64_override_options): Sanity check for ILP32 and ISA level.
28407 (aarch64_attributes): New function attributes for "sign-return-address".
28408 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
28409 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
28410 ("*do_return"): Generate combined instructions according to key index.
28411 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
28412 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
28414 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
28415 * config/aarch64/aarch64.opt (msign-return-address=): New.
28416 * doc/extend.texi (AArch64 Function Attributes): Documents
28417 "sign-return-address=".
28418 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
28420 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28422 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
28423 overall option summary.
28425 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28427 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
28428 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
28429 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
28430 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
28432 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
28434 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
28435 -mpower9-minmax by default for -mcpu=power9.
28436 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
28437 128-bit floating point.
28439 2017-01-20 Alan Modra <amodra@gmail.com>
28441 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
28442 optimizing for size.
28444 2017-01-20 Alan Modra <amodra@gmail.com>
28447 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
28448 for strcmp and strncmp from corresponding builtin decl.
28450 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28452 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
28453 instead of i386/rtems-64.h.
28454 * config/i386/rtems-64.h: Remove.
28456 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28460 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
28462 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
28464 2017-01-19 Tamar Christina <tamar.christina@arm.com>
28466 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
28467 Change int to HOST_WIDE_INT.
28468 * config/aarch64/aarch64-protos.h
28469 (aarch64_simd_gen_const_vector_dup): Likewise.
28470 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
28472 2017-01-19 David Malcolm <dmalcolm@redhat.com>
28474 * langhooks-def.h (lhd_type_for_size): New decl.
28475 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
28476 * langhooks.c (lhd_type_for_size): New function, taken from
28479 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
28481 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
28482 define_bypass for CR latency.
28483 (power9-cracked-alu): Update bypass latency and remove power9-branch.
28484 (power9-alu2): Add define_bypass for CR latency.
28486 (power9-mul): Update insn latency.
28487 (power9-mul-compare): Update insn latency, bypass latency and remove
28490 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28492 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
28494 * config/aarch64/aarch64.md
28495 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
28496 aarch64_nopcrelative_literal_loads.
28497 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
28499 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
28501 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
28502 TARGET_LOONGSON_3A.
28503 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
28505 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
28508 * config.gcc (supported_defaults): Add lxc1-sxc1.
28509 (with_lxc1_sxc1): Add validation.
28510 (all_defaults): Add lxc1-sxc1.
28511 * config/mips/mips.opt (mlxc1-sxc1): New option.
28512 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28514 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28515 __mips_no_lxc1_sxc1.
28516 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
28517 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
28518 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
28520 2017-01-19 Richard Biener <rguenther@suse.de>
28522 PR tree-optimization/72488
28523 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
28524 sure to restore SSA info.
28525 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
28527 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
28529 PR rtl-optimization/79121
28530 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
28531 of the inner type when shifting an extended value.
28533 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28536 * symtab.c (symtab_node::equal_address_to): Fix comparing of
28537 interposable aliases.
28539 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
28542 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
28543 Use the evmergelohi instruction.
28544 (mov_si<mode>_e500_subreg4_2_le): Likewise.
28545 (mov_sitf_e500_subreg8_2_be): Likewise.
28546 (mov_sitf_e500_subreg12_2_le): Likewise.
28547 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
28548 (mov_si<mode>_e500_subreg4_2_be): Likewise.
28549 (mov_sitf_e500_subreg8_2_le): Likewise.
28550 (mov_sitf_e500_subreg12_2_be): Likewise.
28552 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28554 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
28555 attribute from vecsimple to vecperm.
28556 (altivec_vbpermq2): Likewise.
28558 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28561 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
28563 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28564 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
28565 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
28566 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
28567 case where N arg is SIZE_MAX.
28568 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
28569 (cmpstrsi): Add pattern.
28571 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28573 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28574 __builtin_vec_revb builtins.
28575 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
28576 built-in functions to support generation of the ISA 3.0 XXBR<x>
28577 vector byte reverse instructions.
28578 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
28579 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
28580 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
28581 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
28582 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
28583 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
28584 (P9V_BUILTIN_VEC_REVB): Likewise.
28585 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
28586 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
28587 (p9_xxbrq_v16qi): Likewise.
28588 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
28589 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
28590 (p9_xxbrh_v8hi): Likewise.
28591 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
28592 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
28593 vec_revb built-in functions.
28595 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
28597 PR rtl-optimization/78952
28598 * config/i386/i386.md (any_extract): New code iterator.
28599 (*insvqi_2): Use any_extract for source operand.
28600 (*insvqi_3): Use any_shiftrt for source operand.
28602 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
28604 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
28606 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
28608 2017-01-18 Matthias Klose <doko@ubuntu.com>
28610 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
28612 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28614 * config/rs6000/altivec.h (vec_bperm): Change #define.
28615 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
28616 (altivec_vbpermq2): New define_insn.
28617 (altivec_vbpermd): Likewise.
28618 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
28619 function interface.
28620 (VBPERMD): Likewise.
28621 (VBPERM): New polymorphic function interface.
28622 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
28623 Add entries for P9V_BUILTIN_VEC_VBPERM.
28624 * doc/extend.texi: Add interfaces for vec_bperm.
28626 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28628 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
28629 first letter of error messages.
28630 (s390_resolve_overloaded_builtin): Likewise.
28631 * config/s390/s390.c (s390_expand_builtin): Likewise.
28632 (s390_invalid_arg_for_unprototyped_fn): Likewise.
28633 (s390_valid_target_attribute_inner_p): Likewise.
28634 * config/s390/s390.md ("tabort"): Likewise.
28636 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
28638 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
28639 (ISA_AVOID_DIV_HILO): New macro.
28640 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
28641 (ISA_HAS_DDIV): Likewise.
28643 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28645 * doc/invoke.texi (fabi-version): Correct number of occurrences.
28647 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28649 * doc/invoke.texi (fabi-version): Spelling fix.
28651 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28654 * doc/invoke.texi (fabi-version): Mention mangling fix for
28657 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28660 * doc/invoke.texi (fabi-version): Document discriminator mangling.
28662 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
28665 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
28666 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
28668 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
28670 (stack_protect_test): Ditto.
28671 * config/rs6000/rs6000.opt (mstack-protector-guard=,
28672 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
28674 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
28675 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
28676 -mstack-protector-guard-offset=.
28677 (RS/6000 and PowerPC Options): Ditto.
28679 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28681 * config/i386/i386.h (MASK_CLASS_P): New define.
28682 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
28683 there are no registers from different register sets also when
28684 mask registers are used. Update function comment.
28685 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
28686 to (*k/*r) and (*k/*km) alternatives.
28688 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
28690 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
28691 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
28692 (EH_RETURN_HANDLER_RTX): New define.
28693 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
28694 Force frame pointer in EH return functions.
28695 (aarch64_expand_epilogue): Add barrier for eh_return.
28696 (aarch64_final_eh_return_addr): Remove.
28697 (aarch64_eh_return_handler_rtx): New function.
28698 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
28700 (aarch64_eh_return_handler_rtx): New prototype.
28702 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28704 * config/rs6000/altivec.h (vec_rlmi): New #define.
28705 (vec_vrlnm): Likewise.
28706 (vec_rlnm): Likewise.
28707 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
28708 (UNSPEC_VRLNM): Likewise.
28709 (VIlong): New mode iterator.
28710 (altivec_vrl<VI_char>mi): New define_insn.
28711 (altivec_vrl<VI_char>nm): Likewise.
28712 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
28714 (VRLDNM): Likewise.
28715 (RLNM): New polymorphic function entry.
28716 (VRLWMI): New monomorphic function entry.
28717 (VRLDMI): Likewise.
28718 (RLMI): New polymorphic function entry.
28719 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
28720 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
28721 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
28724 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28727 * dwarf2out.c (field_byte_offset): Restore the
28728 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
28729 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
28730 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
28733 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28736 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
28738 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28741 * dwarf2out.c (add_data_member_location_attribute): For constant
28742 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
28743 instead of DW_AT_data_member_location, DW_AT_bit_offset and
28744 DW_AT_byte_size attributes.
28746 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28748 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
28749 after forcing to constant memory when the code model is medium.
28751 2017-01-17 Julia Koval <julia.koval@intel.com>
28754 * config/i386/avx512fintrin.h
28755 (_mm512_i32gather_ps): Change __addr type to void const*.
28756 (_mm512_mask_i32gather_ps): Ditto.
28757 (_mm512_i32gather_pd): Ditto.
28758 (_mm512_mask_i32gather_pd): Ditto.
28759 (_mm512_i64gather_ps): Ditto.
28760 (_mm512_mask_i64gather_ps): Ditto.
28761 (_mm512_i64gather_pd): Ditto.
28762 (_mm512_mask_i64gather_pd): Ditto.
28763 (_mm512_i32gather_epi32): Ditto.
28764 (_mm512_mask_i32gather_epi32): Ditto.
28765 (_mm512_i32gather_epi64): Ditto.
28766 (_mm512_mask_i32gather_epi64): Ditto.
28767 (_mm512_i64gather_epi32): Ditto.
28768 (_mm512_mask_i64gather_epi32): Ditto.
28769 (_mm512_i64gather_epi64): Ditto.
28770 (_mm512_mask_i64gather_epi64): Ditto.
28771 (_mm512_i32scatter_ps): Change __addr type to void*.
28772 (_mm512_mask_i32scatter_ps): Ditto.
28773 (_mm512_i32scatter_pd): Ditto.
28774 (_mm512_mask_i32scatter_pd): Ditto.
28775 (_mm512_i64scatter_ps): Ditto.
28776 (_mm512_mask_i64scatter_ps): Ditto.
28777 (_mm512_i64scatter_pd): Ditto.
28778 (_mm512_mask_i64scatter_pd): Ditto.
28779 (_mm512_i32scatter_epi32): Ditto.
28780 (_mm512_mask_i32scatter_epi32): Ditto.
28781 (_mm512_i32scatter_epi64): Ditto.
28782 (_mm512_mask_i32scatter_epi64): Ditto.
28783 (_mm512_i64scatter_epi32): Ditto.
28784 (_mm512_mask_i64scatter_epi32): Ditto.
28785 (_mm512_i64scatter_epi64): Ditto.
28786 (_mm512_mask_i64scatter_epi64): Ditto.
28787 * config/i386/avx512pfintrin.h
28788 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
28789 (_mm512_mask_prefetch_i32gather_ps): Ditto.
28790 (_mm512_mask_prefetch_i64gather_pd): Ditto.
28791 (_mm512_mask_prefetch_i64gather_ps): Ditto.
28792 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
28793 (_mm512_prefetch_i32scatter_ps): Ditto.
28794 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
28795 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
28796 (_mm512_prefetch_i64scatter_pd): Ditto.
28797 (_mm512_prefetch_i64scatter_ps): Ditto.
28798 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
28799 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
28800 * config/i386/avx512vlintrin.h
28801 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
28802 (_mm_mmask_i32gather_ps): Ditto.
28803 (_mm256_mmask_i32gather_pd): Ditto.
28804 (_mm_mmask_i32gather_pd): Ditto.
28805 (_mm256_mmask_i64gather_ps): Ditto.
28806 (_mm_mmask_i64gather_ps): Ditto.
28807 (_mm256_mmask_i64gather_pd): Ditto.
28808 (_mm_mmask_i64gather_pd): Ditto.
28809 (_mm256_mmask_i32gather_epi32): Ditto.
28810 (_mm_mmask_i32gather_epi32): Ditto.
28811 (_mm256_mmask_i32gather_epi64): Ditto.
28812 (_mm_mmask_i32gather_epi64): Ditto.
28813 (_mm256_mmask_i64gather_epi32): Ditto.
28814 (_mm_mmask_i64gather_epi32): Ditto.
28815 (_mm256_mmask_i64gather_epi64): Ditto.
28816 (_mm_mmask_i64gather_epi64): Ditto.
28817 (_mm256_i32scatter_ps): Change __addr type to void*.
28818 (_mm256_mask_i32scatter_ps): Ditto.
28819 (_mm_i32scatter_ps): Ditto.
28820 (_mm_mask_i32scatter_ps): Ditto.
28821 (_mm256_i32scatter_pd): Ditto.
28822 (_mm256_mask_i32scatter_pd): Ditto.
28823 (_mm_i32scatter_pd): Ditto.
28824 (_mm_mask_i32scatter_pd): Ditto.
28825 (_mm256_i64scatter_ps): Ditto.
28826 (_mm256_mask_i64scatter_ps): Ditto.
28827 (_mm_i64scatter_ps): Ditto.
28828 (_mm_mask_i64scatter_ps): Ditto.
28829 (_mm256_i64scatter_pd): Ditto.
28830 (_mm256_mask_i64scatter_pd): Ditto.
28831 (_mm_i64scatter_pd): Ditto.
28832 (_mm_mask_i64scatter_pd): Ditto.
28833 (_mm256_i32scatter_epi32): Ditto.
28834 (_mm256_mask_i32scatter_epi32): Ditto.
28835 (_mm_i32scatter_epi32): Ditto.
28836 (_mm_mask_i32scatter_epi32): Ditto.
28837 (_mm256_i32scatter_epi64): Ditto.
28838 (_mm256_mask_i32scatter_epi64): Ditto.
28839 (_mm_i32scatter_epi64): Ditto.
28840 (_mm_mask_i32scatter_epi64): Ditto.
28841 (_mm256_i64scatter_epi32): Ditto.
28842 (_mm256_mask_i64scatter_epi32): Ditto.
28843 (_mm_i64scatter_epi32): Ditto.
28844 (_mm_mask_i64scatter_epi32): Ditto.
28845 (_mm256_i64scatter_epi64): Ditto.
28846 (_mm256_mask_i64scatter_epi64): Ditto.
28847 (_mm_i64scatter_epi64): Ditto.
28848 (_mm_mask_i64scatter_epi64): Ditto.
28849 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
28850 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
28851 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
28852 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
28853 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
28854 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
28855 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
28856 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
28857 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
28858 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
28859 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
28860 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
28861 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
28862 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
28863 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
28864 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
28865 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
28866 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
28867 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
28868 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
28869 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
28870 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
28871 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
28872 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
28873 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
28874 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
28875 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
28876 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
28877 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
28878 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
28879 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
28880 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
28881 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
28882 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
28883 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
28884 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
28885 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
28886 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
28887 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
28888 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
28889 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
28890 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
28891 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
28892 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
28893 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
28894 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
28895 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
28896 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
28897 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
28898 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
28899 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
28900 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
28901 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
28902 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
28903 definitions accordingly.
28905 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
28906 Kuan-Lin Chen <kuanlinchentw@gmail.com>
28909 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
28912 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
28915 * ira-conflicts.c (ira_build_conflicts): Update total conflict
28916 hard regs for inner regno.
28918 2017-01-17 Martin Liska <mliska@suse.cz>
28921 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
28922 assumption and add comment.
28924 2017-01-17 Nathan Sidwell <nathan@acm.org>
28926 * ipa-visibility.c (localize_node): New function, broken out of ...
28927 (function_and_variable_visibility): ... here. Call it.
28929 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28931 PR middle-end/77445
28932 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
28933 correctly set frequency of oudgoing edge.
28934 (duplicate_thread_path): Fix profile updating.
28936 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28939 * configure.ac: Add GCC_BASE_VER.
28940 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
28941 version from BASE-VER file.
28942 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
28943 (gcc.o): Depend on $(BASEVER).
28944 * common.opt (dumpfullversion): New option.
28945 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
28946 * doc/invoke.texi: Document -dumpfullversion.
28947 * doc/install.texi: Document --with-gcc-major-version-only.
28948 * configure: Regenerated.
28950 2017-01-17 Richard Biener <rguenther@suse.de>
28952 PR tree-optimization/71433
28953 * tree-vrp.c (register_new_assert_for): Merge same asserts
28954 on all incoming edges.
28955 (process_assert_insertions_for): Handle insertions at the
28958 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
28960 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
28961 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
28963 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
28966 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
28969 2017-01-17 Alan Modra <amodra@gmail.com>
28972 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
28973 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
28974 symbolic stack limit when pic.
28976 2017-01-16 Martin Sebor <msebor@redhat.com>
28978 PR tree-optimization/78608
28979 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
28981 2017-01-16 Jeff Law <law@redhat.com>
28984 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
28985 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
28986 for several include directories that may be relative to sysroot.
28987 * config/i386/x-mingw32 (gplus_includedir): Define.
28988 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
28989 (native_system_includedir): Likewise.
28990 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
28991 override if TARGET_SYSTEM_ROOT is defined.
28992 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
28994 PR tree-optimization/79090
28995 PR tree-optimization/33562
28996 PR tree-optimization/61912
28997 PR tree-optimization/77485
28998 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
28999 and computed trims into the dump file.
29001 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29003 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
29005 2017-01-16 Jakub Jelinek <jakub@redhat.com>
29008 * gimplify.c (gimplify_init_constructor): If want_value and
29009 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
29013 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
29014 sequence. Formatting fixes.
29015 (doloop_optimize): Formatting fixes.
29018 * gcc.c (debug_level_greater_than_spec_func): New function.
29019 (static_spec_functions): Add debug-level-gt spec function.
29020 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
29022 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29023 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29024 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
29025 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
29026 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
29027 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
29029 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
29031 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
29032 QImode fixups to general and mask registers only.
29034 2017-01-16 Carl Love <cel@us.ibm.com>
29036 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
29037 for built-in functions
29038 vector signed char vec_nabs (vector signed char)
29039 vector signed short vec_nabs (vector signed short)
29040 vector signed int vec_nabs (vector signed int)
29041 vector signed long long vec_nabs (vector signed long long)
29042 vector float vec_nabs (vector float)
29043 vector double vec_nabs (vector double)
29044 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
29046 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
29047 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
29048 * doc/extend.texi: Update the documentation file for the new built-in
29051 2017-01-16 Martin Sebor <msebor@redhat.com>
29053 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
29056 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29058 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
29059 UNSPEC_VSX__XXSPLTD to require special splat handling.
29061 2017-01-16 David Malcolm <dmalcolm@redhat.com>
29064 * system.h: Poison strndup.
29066 2017-01-16 Alan Modra <amodra@gmail.com>
29069 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
29072 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
29074 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
29076 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
29078 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
29079 call recog here. Assert that INSN_CODE (insn) is non-negative.
29081 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29084 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
29086 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
29087 in the currently scheduled RTL fragment.
29089 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29091 PR rtl-optimization/78751
29092 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
29095 2017-01-14 Jeff Law <law@redhat.com>
29097 PR tree-optimization/79090
29098 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
29099 variable length stores.
29100 (compute_trims): Delete dead assignment to *trim_tail.
29101 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
29104 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
29106 PR rtl-optimization/78626
29107 PR rtl-optimization/78727
29108 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
29109 of a block, and split such blocks after everything else is finished.
29111 2017-01-14 Alan Modra <amodra@gmail.com>
29114 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
29115 target legitimate_combined_insn.
29116 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
29117 (rs6000_legitimate_combined_insn): New function.
29118 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
29120 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
29121 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
29122 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
29124 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
29126 * doc/frontends.texi (G++ and GCC): Remove references to Java.
29128 2017-01-13 Jeff Law <law@redhat.com>
29130 PR tree-optimization/33562
29131 PR tree-optimization/61912
29132 PR tree-optimization/77485
29133 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
29135 (delete_dead_assignment): Likewise.
29136 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
29137 statement to delete_dead_call and delete_dead_assignment.
29139 2017-01-13 David Malcolm <dmalcolm@redhat.com>
29142 * substring-locations.c (format_warning_va): Strengthen case 1 so
29143 that both endpoints of the substring must be within the format
29144 range for just the substring to be printed.
29146 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
29148 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
29149 * config/i386/i386.c (ix86_target_string): Add missing options
29150 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
29151 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
29152 flags_other and ix86_target_other to flags2_other. Display unknown
29154 (ix86_valid_target_attribute_inner_p): Add missing options and
29155 reorder options by implied ISAs, as in ix86_target_string.
29157 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29159 * hash-table.h (hash_table::too_empty_p): New function.
29160 (hash_table::expand): Use it.
29161 (hash_table::traverse): Likewise.
29162 (hash_table::empty_slot): Use sizeof (value_type) instead of
29163 sizeof (PTR) to convert bytes to elements. Shrink the table
29164 if the current size is excessive for the current number of
29167 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29169 * ira-costs.c (record_reg_classes): Break from the inner loop
29170 early once alt_fail is known to be true. Update outer loop
29171 handling accordingly.
29173 2017-01-13 Jeff Law <law@redhat.com>
29175 * tree-ssa-dse.c (decrement_count): New function.
29176 (increment_start_addr, maybe_trim_memstar_call): Likewise.
29177 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
29178 when we know the partially dead statement is a mem* function.
29180 PR tree-optimization/61912
29181 PR tree-optimization/77485
29182 * tree-ssa-dse.c: Include expr.h.
29183 (maybe_trim_constructor_store): New function.
29184 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
29186 PR tree-optimization/33562
29187 PR tree-optimization/61912
29188 PR tree-optimization/77485
29189 * doc/invoke.texi: Document new dse-max-object-size param.
29190 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
29191 * tree-ssa-dse.c: Include params.h.
29192 (dse_store_status): New enum.
29193 (initialize_ao_ref_for_dse): New, partially extracted from
29195 (valid_ao_ref_for_dse, normalize_ref): New.
29196 (setup_live_bytes_from_ref, compute_trims): Likewise.
29197 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
29198 (maybe_trim_partially_dead_store): Likewise.
29199 (maybe_trim_complex_store): Likewise.
29200 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
29201 Track what bytes live from the original store. Return tri-state
29202 for dead, partially dead or live.
29203 (dse_dom_walker): Add constructor, destructor and new private members.
29204 (delete_dead_call, delete_dead_assignment): New extracted from
29206 (dse_optimize_stmt): Make a member of dse_dom_walker.
29207 Use initialize_ao_ref_for_dse.
29209 PR tree-optimization/33562
29210 PR tree-optimization/61912
29211 PR tree-optimization/77485
29212 * sbitmap.h (bitmap_count_bits): Prototype.
29213 (bitmap_clear_range, bitmap_set_range): Likewise.
29214 * sbitmap.c (bitmap_clear_range): New function.
29215 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
29217 2017-01-13 Martin Liska <mliska@suse.cz>
29220 * function.c (set_cfun): Add new argument force.
29221 * function.h (set_cfun): Likewise.
29222 * ipa-inline-transform.c (inline_call): Use the function when
29223 strict alising from is dropped for function we inline to.
29225 2017-01-13 Richard Biener <rguenther@suse.de>
29227 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
29228 for dumping GIMPLE INTEGER_CSTs.
29230 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29232 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
29233 to 201112L since C++17.
29235 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
29238 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
29239 if -fsanitize=kernel-address is present.
29241 2017-01-13 Richard Biener <rguenther@suse.de>
29243 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
29244 as _Literal ( type ) number in case usual suffixes do not
29245 preserve all information.
29247 2017-01-13 Richard Biener <rguenther@suse.de>
29249 PR tree-optimization/77283
29250 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
29251 and ssa-iterators.h.
29252 (is_feasible_trace): Implement a cost model based on joiner
29255 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
29258 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
29259 char or short to __float128/_Float128 directly.
29261 2017-01-12 Martin Sebor <msebor@redhat.com>
29263 to -Wformat-overflow.
29264 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
29265 (min_bytes_remaining): Same.
29266 (get_string_length): Same.
29267 (format_string): Same.
29268 (format_directive): Same.
29270 (pass_sprintf_length::handle_gimple_call): Same.
29272 2017-01-12 Jakub Jelinek <jakub@redhat.com>
29274 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
29275 info.nowrite calls with no lhs that can't throw. Return bool
29276 whether gsi_remove has been called or not.
29277 (pass_sprintf_length::handle_gimple_call): Return bool whether
29278 try_substitute_return_value called gsi_remove. Formatting fix.
29279 (pass_sprintf_length::execute): Don't use gsi_remove if
29280 handle_gimple_call returned true.
29283 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
29284 be removed due to side-effects, don't remove following barrier nor
29285 turn the successor edge into fallthru edge.
29287 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29290 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
29291 element-reversing loads and stores as not swappable.
29293 2017-01-12 Nathan Sidwell <nathan@acm.org>
29294 Nicolai Stange <nicstange@gmail.com>
29296 * combine.c (try_combine): Don't ignore result of overlap checking
29297 loop. Combine overlap & asm check into single loop.
29299 2017-01-12 Richard Biener <rguenther@suse.de>
29301 * tree-pretty-print.c (dump_generic_node): Provide -gimple
29302 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
29304 2017-01-12 Richard Biener <rguenther@suse.de>
29306 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
29307 and TS_TARGET_OPTION directly derive from TS_BASE.
29308 * tree-core.h (tree_optimization_option): Derive from tree_base.
29309 (tree_target_option): Likewise.
29311 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
29313 * config/i386/i386.c (memory_address_length): Increase len
29314 only when rip_relative_addr_p returns false.
29316 2017-01-11 Julia Koval <julia.koval@intel.com>
29318 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
29319 (OPTION_MASK_ISA_SGX_SET): New.
29320 (ix86_handle_option): Handle OPT_msgx.
29321 * config.gcc: Added sgxintrin.h.
29322 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
29323 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
29324 * config/i386/i386.c (ix86_target_string): Add -msgx.
29326 (ix86_option_override_internal): Handle new options.
29327 (ix86_valid_target_attribute_inner_p): Add sgx.
29328 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
29329 * config/i386/i386.opt: Add msgx.
29330 * config/i386/sgxintrin.h: New file.
29331 * config/i386/x86intrin.h: Add sgxintrin.h.
29333 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29336 * fold-const.c (maybe_nonzero_address): Return 1 for function
29338 (tree_single_nonzero_warnv_p): Don't handle function local objects
29342 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
29345 2017-01-11 David Malcolm <dmalcolm@redhat.com>
29348 * opts.c: Include "spellcheck.h"
29349 (struct string_fragment): New struct.
29350 (struct edit_distance_traits<const string_fragment &>): New
29352 (get_closest_sanitizer_option): New function.
29353 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
29355 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29357 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
29359 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
29360 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
29361 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
29362 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
29363 for initial die_offset if dwarf_split_debug_info.
29364 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
29365 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
29367 (output_skeleton_debug_sections): Formatting fix. Use
29368 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
29369 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
29371 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
29373 * config/arm/cortex-a53.md: Add bypasses for
29374 cortex_a53_r2f_cvt.
29375 (cortex_a53_r2f): Only use for transfers.
29376 (cortex_a53_f2r): Likewise.
29377 (cortex_a53_r2f_cvt): Add reservation for conversions.
29378 (cortex_a53_f2r_cvt): Likewise.
29380 2017-01-11 Tamar Christina <tamar.christina@arm.com>
29382 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
29383 to all inlined functions, change static to extern.
29385 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
29388 * config/arm/arm.c (legitimize_pic_address): Handle reference to
29390 (arm_assemble_integer): Likewise.
29392 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
29394 * config.gcc: Use new awk script to check CPU, FPU and architecture
29395 parameters for --with-... options.
29396 * config/arm/parsecpu.awk: New file
29397 * config/arm/arm-cpus.in: New file.
29398 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
29400 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
29402 * config/arm/t-arm: Update dependency rules.
29403 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
29404 of processing .def files.
29405 * config/arm/genopt.sh: Deleted.
29406 * config/arm/gentune.sh: Deleted.
29407 * config/arm/arm-cores.def: Deleted.
29408 * config/arm/arm-arches.def: Deleted.
29409 * config/arm/arm-fpus.def: Deleted.
29410 * config/arm/arm-tune.md: Regenerated.
29411 * config/arm/arm-tables.opt: Regenerated.
29412 * config/arm/arm-cpu.h: New generated file.
29413 * config/arm/arm-cpu-data.h: New generated file.
29414 * config/arm/arm-cpu-cdata.h: New generated file.
29416 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
29419 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
29421 (input_varpool_node): Unpack dynamically_initialized bit.
29423 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
29425 PR rtl-optimization/79032
29426 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
29427 the alignment of the adjusted memory reference against that of MODE,
29428 instead of the alignment of the original memory reference.
29430 2017-01-11 Martin Jambor <mjambor@suse.cz>
29432 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
29434 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
29435 decorated functions.
29437 2017-01-11 Richard Biener <rguenther@suse.de>
29439 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
29440 set range/nonnull info for PHI results. Do not set it on
29441 stmts marked for removal.
29443 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
29445 * expr.c (store_field): In the bitfield case, fetch the return value
29446 from the registers before applying a single big-endian adjustment.
29447 Always do a final load for a BLKmode value not larger than a word.
29449 2017-01-10 David Malcolm <dmalcolm@redhat.com>
29452 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29453 that we correctly handle column numbers greater than
29454 LINE_MAP_MAX_COLUMN_NUMBER.
29456 2017-01-10 Martin Sebor <msebor@redhat.com>
29458 PR middle-end/78245
29459 * gimple-ssa-sprintf.c (get_destination_size): Call
29460 {init,fini}object_sizes.
29461 * tree-object-size.c (addr_object_size): Adjust.
29462 (pass_through_call): Adjust.
29463 (pass_object_sizes::execute): Adjust.
29464 * tree-object-size.h (fini_object_sizes): Declare.
29466 2017-01-10 Martin Sebor <msebor@redhat.com>
29468 PR tree-optimization/78775
29469 * builtins.c (get_size_range): Move...
29470 * calls.c: ...to here.
29471 (alloc_max_size): Accept zero argument.
29472 (operand_signed_p): Remove.
29473 (maybe_warn_alloc_args_overflow): Call get_size_range.
29474 * calls.h (get_size_range): Declare.
29476 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
29478 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
29479 from TI's devices.csv file as of September 2016.
29480 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
29482 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
29484 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
29485 * doc/invoke.texi: Likewise.
29486 * doc/md.texi: Likewise.
29487 * doc/objc.texi: Likewise.
29489 2017-01-10 Joshua Conner <joshconner@google.com>
29491 * config/arm/fuchsia-elf.h: New file.
29492 * config/fuchsia.h: New file.
29493 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
29494 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
29496 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
29498 2016-01-10 Richard Biener <rguenther@suse.de>
29500 PR tree-optimization/79034
29501 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29502 Propagate out degenerate PHIs in the joiner.
29504 2017-01-10 Martin Liska <mliska@suse.cz>
29506 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
29507 (sort_congruence_classes_by_decl_uid): Likewise.
29508 (sort_congruence_class_groups_by_decl_uid): Likewise.
29509 (sem_item_optimizer::merge_classes): Sort class, groups in these
29510 classes and members in the groups by DECL_UID of declarations.
29511 This would make merge operations stable.
29513 2017-01-10 Martin Liska <mliska@suse.cz>
29515 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
29516 usage of m_classes_vec.
29517 (sem_item_optimizer::~sem_item_optimizer): Likewise.
29518 (sem_item_optimizer::get_group_by_hash): Likewise.
29519 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
29520 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
29521 (sem_item_optimizer::verify_classes): Likewise.
29522 (sem_item_optimizer::process_cong_reduction): Likewise.
29523 (sem_item_optimizer::dump_cong_classes): Likewise.
29524 (sem_item_optimizer::merge_classes): Likewise.
29525 * ipa-icf.h (congruence_class_hash): Rename from
29526 congruence_class_group_hash. Remove declaration of m_classes_vec.
29528 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
29530 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
29531 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
29532 * config.gcc: Add avx512vpopcntdqintrin.h.
29533 * config/i386/avx512vpopcntdqintrin.h: New.
29534 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
29535 * config/i386/i386-builtin-types.def: Add new types.
29536 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
29537 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
29538 __builtin_ia32_vpopcountq_v8di_mask): New.
29539 * config/i386/i386-c.c (ix86_target_macros_internal): Define
29540 __AVX512VPOPCNTDQ__.
29541 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
29542 (PTA_AVX512VPOPCNTDQ): Define.
29543 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
29544 TARGET_AVX512VPOPCNTDQ_P): Define.
29545 * config/i386/i386.opt: Add mavx512vpopcntdq.
29546 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
29547 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
29549 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29551 PR middle-end/77484
29552 * predict.def (PRED_CALL): Set to 67.
29554 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
29556 * expr.c (store_field): In the bitfield case, if the value comes from
29557 a function call and is of an aggregate type returned in registers, do
29558 not modify the field mode; extract the value in all cases if the mode
29559 is BLKmode and the size is not larger than a word.
29561 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
29564 * config/i386/cpuid.h: Fix undefined behavior.
29566 2017-01-04 Jeff Law <law@redhat.com>
29568 PR tree-optimization/79007
29569 PR tree-optimization/67955
29570 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
29571 conservative for pt.null when flag_non_call_exceptions is on.
29573 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29575 PR translation/79019
29576 PR translation/79020
29577 * params.def (PARAM_INLINE_MIN_SPEEDUP,
29578 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
29579 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
29581 * config/avr/avr.opt (maccumulate-args): Likewise.
29582 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
29583 * common.opt (freport-bug): Likewise.
29584 * cif-code.def (CIF_FINAL_ERROR): Likewise.
29585 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
29586 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
29587 translatable string.
29588 * config/i386/i386.c (function_value_32): Likewise.
29589 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
29590 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
29592 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
29593 * common/config/msp430/msp430-common.c (msp430_handle_option):
29595 * symtab.c (symtab_node::verify_base): Likewise.
29596 * opts.c (set_debug_level): Likewise.
29597 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
29598 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
29599 missing whitespace to translatable strings.
29600 * config/avr/avr.md (bswapsi2): Fix typo in comment.
29601 * config/sh/superh.h: Likewise.
29602 * config/i386/xopintrin.h: Likewise.
29603 * config/i386/znver1.md: Likewise.
29604 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
29605 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
29606 * double-int.h (struct double_int): Likewise.
29607 * double-int.c (div_and_round_double): Likewise.
29608 * wide-int.cc: Likewise.
29609 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
29610 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
29611 * cfgcleanup.c (crossjumps_occured): Renamed to ...
29612 (crossjumps_occurred): ... this.
29613 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
29616 PR tree-optimization/78899
29617 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
29618 returning bool return struct loop *, NULL for failure and the new
29620 (versionable_outer_loop_p): Don't version outer loop if it has
29621 dont_vectorized bit set.
29622 (tree_if_conversion): When versioning outer loop, ensure
29623 tree_if_conversion is performed also on the inner loop of the
29624 non-vectorizable outer loop copy.
29625 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
29626 LOOP_VECTORIZED in inner loop of the scalar outer loop and
29627 prevent vectorization of it.
29628 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
29629 the outer loop vectorization of the non-scalar version is attempted
29630 before vectorization of the inner loop in scalar version. If
29631 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
29632 vectorization of its inner loop.
29633 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
29634 has 2 inner loops, rename also on edges from bb whose single pred
29635 is outer_loop->header. Fix typo in function comment.
29637 2017-01-09 Martin Sebor <msebor@redhat.com>
29640 * asan.c (asan_emit_stack_protection): Increase local buffer size
29641 to avoid snprintf truncation warning.
29643 2017-01-09 Andrew Pinski <apinski@cavium.com>
29645 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
29646 to reference thunderx2t99 for the tuning structure
29647 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
29649 (thunderx2t99_extra_costs): This.
29650 * config/aarch64/aarch64-tune.md: Regenerate.
29651 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
29652 (vulcan_addrcost_table): This.
29653 (vulcan_regmove_cost): Rename to ...
29654 (thunderx2t99_regmove_cost): This.
29655 (vulcan_vector_cost): Rename to ...
29656 (thunderx2t99_vector_cost): this.
29657 (vulcan_branch_cost): Rename to ...
29658 (thunderx2t99_branch_cost): This.
29659 (vulcan_tunings): Rename to ...
29660 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
29661 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
29663 2017-01-09 Martin Jambor <mjambor@suse.cz>
29667 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
29668 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
29669 (propagate_vr_accross_jump_function): Use the above function for all
29670 value range computations for pass-through jump functions and type
29671 converasion from explicit value range values.
29672 (ipcp_propagate_stage): Do not attempt to deduce types of formal
29673 parameters from TYPE_ARG_TYPES.
29674 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
29675 (ipa_write_node_info): Stream type of the actual argument.
29676 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
29678 2017-01-09 Martin Liska <mliska@suse.cz>
29681 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
29682 (lookup_compiler): Do not show error message with have_E.
29684 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29686 PR tree-optimization/78938
29687 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
29688 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
29689 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
29690 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
29693 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29695 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
29698 2017-01-09 Richard Biener <rguenther@suse.de>
29700 PR tree-optimization/78997
29701 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
29702 name condition properly.
29704 2017-01-09 Richard Biener <rguenther@suse.de>
29707 * dwarf2out.c (is_cxx): New overload with context.
29708 (is_naming_typedef_decl): Use it.
29710 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29712 * invoke.texi (Option Summary): Correct spacing in option lists
29713 and add line breaks to fix over-long lines.
29715 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29717 PR middle-end/17660
29719 * extend.texi (Common Variable Attributes): Add xref to GCC
29720 Internals manual to explain mode attribute keywords.
29722 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29725 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
29726 and Preprocessor Options.
29727 (Options for Linking): Document -pthread here....
29728 (RS/6000 and PowerPC Options): ...not here.
29729 (Solaris 2 Options): ...or here.
29730 * doc/cppopts.texi: Document -pthread.
29732 2017-01-08 Martin Sebor <msebor@redhat.com>
29734 PR middle-end/77708
29735 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
29736 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
29737 New member functions.
29738 (format_directive): Used them.
29740 (pass_sprintf_length::handle_gimple_call): Same.
29741 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
29742 to avoid truncation for any argument.
29743 (extract_affine_mul): Same.
29744 * tree.c (get_file_function_name): Same.
29746 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29748 PR middle-end/77484
29749 * predict.def (PRED_INDIR_CALL): Set to 86.
29751 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29753 PR preprocessor/54124
29754 * doc/cppopts.texi: Reformat -d subtable to list the full name
29755 of the options. Add cross-reference to the docs for the general
29756 compiler -d options.
29757 * doc/invoke.texi (Developer Options): Add cross-reference to the
29758 preprocessor-specific -d option documentation.
29760 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29762 PR preprocessor/13498
29763 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
29764 redudant material, and reflect new command-line options.
29765 (System Headers): Likewise.
29767 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29769 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
29770 -isystem, and -idirafter. Copy-edit.
29771 * doc/cppopts.texi: Copy-edit. Remove contradiction about
29772 default for -ftrack-macro-expansion. Delete obsolete and
29773 badly-formatted implementation details about -fdebug-cpp output.
29774 * doc/cppwarnopts.texi: Copy-edit.
29776 2017-01-07 David Malcolm <dmalcolm@redhat.com>
29779 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29780 that the transition from a max line width >= 1<<10 to narrower
29781 lines works correctly.
29783 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
29785 * doc/options.texi (PerFunction): New.
29786 * opt-functions.awk (switch_flags): Map both Optimization and
29787 PerFunction to CL_OPTIMIZATION.
29788 * opth-gen.awk: Test for PerFunction flag along with
29790 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
29791 it only when the latter is present. Skip those that don't in
29792 the hash function generator.
29793 * common.opt (fvar-tracking): Mark as PerFunction instead of
29795 (fvar-tracking-assignments): Likewise.
29796 (fvar-tracking-assignments-toggle): Likewise.
29797 (fvar-tracking-uninit): Likewise.
29799 2017-01-07 Jakub Jelinek <jakub@redhat.com>
29801 PR translation/79018
29802 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
29805 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
29808 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
29809 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
29810 TARGET_LONG_JUMP_TABLE_OFFSETS.
29811 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
29812 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
29813 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
29814 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
29815 * config/m68k/m68k.md (tablejump expander): Likewise.
29816 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
29817 TARGET_LONG_JUMP_TABLE_OFFSETS.
29818 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
29819 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
29821 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
29822 David Holsgrove <david.holsgrove@xilinx.com>
29824 * common/config/microblaze/microblaze-common.c
29825 (TARGET_EXCEPT_UNWIND_INFO): Remove.
29826 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
29828 * config/microblaze/microblaze.c (microblaze_must_save_register)
29829 (microblaze_expand_epilogue, microblaze_return_addr): Handle
29831 (microblaze_eh_return): New function.
29832 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
29833 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
29834 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
29835 * config/microblaze/microblaze.md (eh_return): New pattern.
29837 2017-01-06 Jakub Jelinek <jakub@redhat.com>
29839 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
29840 GCC_DIAGNOSTIC_STRINGIFY): Define.
29842 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
29844 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29846 * config/arm/arm.md (<mcrr>): New.
29848 * config/arm/arm.c (arm_arch5te): New.
29849 (arm_option_override): Set arm_arch5te.
29850 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
29852 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
29853 (arm_mcrr_qualifiers): ... this. New.
29854 (MRRC_QUALIFIERS): Define to...
29855 (arm_mrrc_qualifiers): ... this. New.
29856 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
29858 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
29859 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
29860 (MRRCI, mrrc, MRRC): New.
29861 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
29862 VUNSPEC_MRRC2): New.
29864 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29866 * config/arm/arm.md (<mcr>): New.
29868 * config/arm/arm.c (arm_coproc_builtin_available): Add
29869 support for mcr, mrc, mcr2 and mrc2.
29870 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
29871 (arm_mcr_qualifiers): ... this. New.
29872 (MRC_QUALIFIERS): Define to ...
29873 (arm_mrc_qualifiers): ... this. New.
29874 (MCR_QUALIFIERS): Define to ...
29875 (arm_mcr_qualifiers): ... this. New.
29876 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
29878 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
29879 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
29880 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
29881 VUNSPEC_MRC2): New.
29883 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29885 * config/arm/arm.md (*ldc): New.
29889 * config/arm/arm.c (arm_coproc_builtin_available): Add
29890 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
29891 (arm_coproc_ldc_stc_legitimate_address): New.
29892 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
29893 'qualifier_const_pointer'.
29894 (LDC_QUALIFIERS): Define to...
29895 (arm_ldc_qualifiers): ... this. New.
29896 (STC_QUALIFIERS): Define to...
29897 (arm_stc_qualifiers): ... this. New.
29898 * config/arm/arm-protos.h
29899 (arm_coproc_ldc_stc_legitimate_address): New.
29900 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
29901 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
29902 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
29903 stc2, stcl, stc2l): New.
29904 * config/arm/constraints.md (Uz): New.
29905 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
29906 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
29907 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
29908 VUNSPEC_STC2L): New.
29910 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29912 * config/arm/arm.md (<cdp>): New.
29913 * config/arm/arm.c (neon_const_bounds): Rename this ...
29914 (arm_const_bounds): ... this.
29915 (arm_coproc_builtin_available): New.
29916 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
29917 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
29918 (CDP_QUALIFIERS): Define to...
29919 (arm_cdp_qualifiers): ... this. New.
29921 (arm_expand_builtin_args): Add case for 6 arguments.
29922 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
29923 (arm_const_bounds): ... this.
29924 (arm_coproc_builtin_available): New.
29925 * config/arm/arm_acle.h (__arm_cdp): New.
29927 * config/arm/arm_acle_builtins.def (cdp): New.
29929 * config/arm/iterators.md (CDPI,CDP,cdp): New.
29930 * config/arm/neon.md: Rename all 'neon_const_bounds' to
29931 'arm_const_bounds'.
29932 * config/arm/types.md (coproc): New.
29933 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
29934 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
29935 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
29936 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
29938 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29940 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
29941 (UBINOP_QUALIFIERS): New.
29943 (acle_builtin_data): New. Change comment.
29944 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
29945 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
29946 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
29947 arm_acle_builtins.def.
29948 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
29949 (arm_init_acle_builtins): New.
29950 (CRC32_BUILTIN): Remove.
29951 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
29952 crc32cb, crc32ch and crc32cw.
29953 (arm_init_crc32_builtins): Remove.
29954 (arm_init_builtins): Use arm_init_acle_builtins rather
29955 than arm_init_crc32_builtins.
29956 (arm_expand_acle_builtin): New.
29957 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
29958 * config/arm/arm_acle_builtins.def: New.
29960 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29962 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
29963 (arm_builtin_datum): ... this.
29964 (arm_init_neon_builtin): Rename to ...
29965 (arm_init_builtin): ... this. Add a new parameters PREFIX
29966 and USE_SIG_IN_NAME.
29967 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
29968 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
29969 'arm_builtin_datum'.
29970 (arm_init_vfp_builtins): Likewise.
29971 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
29972 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
29973 (arm_expand_neon_args): Rename to ...
29974 (arm_expand_builtin_args): ... this. Rename builtin_arg
29975 enum values and differentiate between ARG_BUILTIN_MEMORY
29976 and ARG_BUILTIN_NEON_MEMORY.
29977 (arm_expand_neon_builtin_1): Rename to ...
29978 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
29979 values, arm_expand_builtin_args and add bool parameter NEON.
29980 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
29981 (arm_expand_vfp_builtin): Likewise.
29982 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
29984 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29986 PR middle-end/77484
29987 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
29988 * predict.c (tree_estimate_probability_bb): Reverse direction of
29989 polymorphic call predictor.
29991 2017-01-06 David Malcolm <dmalcolm@redhat.com>
29993 * passes.c (execute_one_pass): Split out pass-skipping logic into...
29994 (determine_pass_name_match): ...this new function and...
29995 (should_skip_pass_p): ...this new function.
29997 2017-01-06 Nathan Sidwell <nathan@acm.org>
29999 * ipa-visibility.c (function_and_variable_visibility): Reformat
30000 comments and long lines. Remove extrneous if.
30001 * symtab.c (symtab_node::make_decl_local): Fix code format.
30002 (symtab_node::set_section_for_node): Fix comment typo.
30004 2017-01-06 Martin Liska <mliska@suse.cz>
30007 * lra-constraints.c: Rename invariant to lra_invariant.
30008 * predict.c (set_even_probabilities): Initialize e to NULL.
30010 2017-01-05 Martin Sebor <msebor@redhat.com>
30012 PR tree-optimization/78910
30013 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
30014 (format_integer): Correct off-by-one error in the handling
30015 of precision with negative numbers in signed conversions..
30017 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
30019 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
30021 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30023 PR tree-optimization/71016
30024 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
30025 factor_out_conditional_conversion. Formatting fix.
30026 (factor_out_conditional_conversion): Add cond_stmt argument.
30027 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
30028 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
30031 2017-01-05 David Malcolm <dmalcolm@redhat.com>
30033 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
30034 read-rtl-function.o, and selftest-rtl.o.
30035 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
30036 (selftest::aarch64_test_loading_full_dump): New function.
30037 (selftest::aarch64_run_selftests): New function.
30038 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
30039 selftest::aarch64_run_selftests.
30040 * config/i386/i386.c
30041 (selftest::ix86_test_loading_dump_fragment_1): New function.
30042 (selftest::ix86_test_loading_call_insn): New function.
30043 (selftest::ix86_test_loading_full_dump): New function.
30044 (selftest::ix86_test_loading_unspec): New function.
30045 (selftest::ix86_run_selftests): Call the new functions.
30046 * emit-rtl.c (maybe_set_max_label_num): New function.
30047 * emit-rtl.h (maybe_set_max_label_num): New decl.
30048 * function.c (instantiate_decls): Guard call to
30049 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
30050 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
30052 * gensupport.c (gen_reader::gen_reader): Pass "false"
30053 for new "compact" param of rtx_reader.
30054 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
30055 rather than an empty string for NULL strings.
30056 * read-md.c: Potentially include config.h rather than bconfig.h.
30057 Wrap include of errors.h with #ifdef GENERATOR_FILE.
30058 (have_error): New global, copied from errors.c.
30059 (md_reader::read_name): Rename to...
30060 (md_reader::read_name_1): ...this, adding "out_loc" param,
30061 and converting "missing name or number" to returning false, rather
30063 (md_reader::read_name): Reimplement in terms of read_name_1.
30064 (md_reader::read_name_or_nil): New function.
30065 (md_reader::read_string): Handle "(nil)" by returning NULL.
30066 (md_reader::md_reader): Add new param "compact".
30067 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
30068 (md_reader::read_file): New method.
30069 * read-md.h (md_reader::md_reader): Add new param "compact".
30070 (md_reader::read_file): New method.
30071 (md_reader::is_compact): New accessor.
30072 (md_reader::read_name): Convert return type from void to file_location.
30073 (md_reader::read_name_or_nil): New decl.
30074 (md_reader::read_name_1): New decl.
30075 (md_reader::m_compact): New field.
30076 (noop_reader::noop_reader): Pass "false" for new "compact" param
30078 (rtx_reader::rtx_reader): Add new "compact" param.
30079 (rtx_reader::read_rtx_operand): Make virtual and convert return
30080 type from void to rtx.
30081 (rtx_reader::read_until): New decl.
30082 (rtx_reader::handle_any_trailing_information): New virtual function.
30083 (rtx_reader::postprocess): New virtual function.
30084 (rtx_reader::finalize_string): New virtual function.
30085 (rtx_reader::m_in_call_function_usage): New field.
30086 (rtx_reader::m_reuse_rtx_by_id): New field.
30087 * read-rtl-function.c: New file.
30088 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
30089 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
30090 (selftest::verify_three_block_rtl_cfg): New decl.
30091 * read-rtl-function.h: New file.
30092 * read-rtl.c: Potentially include config.h rather than bconfig.h.
30093 For host, include function.h, memmodel.h, and emit-rtl.h.
30094 (one_time_initialization): New function.
30095 (struct compact_insn_name): New struct.
30096 (compact_insn_names): New array.
30097 (find_code): Handle insn codes in compact dumps.
30098 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
30099 (bind_subst_iter_and_attr): Likewise.
30100 (add_condition_to_string): Likewise.
30101 (add_condition_to_rtx): Likewise.
30102 (apply_attribute_uses): Likewise.
30103 (add_current_iterators): Likewise.
30104 (apply_iterators): Likewise.
30105 (initialize_iterators): Guard usage of apply_subst_iterator with
30106 #ifdef GENERATOR_FILE.
30107 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
30108 (md_reader::read_mapping): Likewise.
30109 (add_define_attr_for_define_subst): Likewise.
30110 (add_define_subst_attr): Likewise.
30111 (read_subst_mapping): Likewise.
30112 (check_code_iterator): Likewise.
30113 (rtx_reader::read_rtx): Likewise. Move one-time initialization
30115 (one_time_initialization): New function.
30116 (rtx_reader::read_until): New method.
30117 (read_flags): New function.
30118 (parse_reg_note_name): New function.
30119 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
30120 Handle reuse_rtx ids.
30121 Wrap iterator lookup within #ifdef GENERATOR_FILE.
30122 Add parsing support for RTL dumps, mirroring the special-cases in
30123 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
30124 values, and calling handle_any_trailing_information.
30125 (rtx_reader::read_rtx_operand): Convert return type from void
30126 to rtx, returning return_rtx. Handle case 'e'. Call
30127 finalize_string on XSTR and XTMPL fields.
30128 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
30129 "(nil)" values were omitted. Call the postprocess vfunc on the
30131 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
30132 class ctor. Initialize m_in_call_function_usage. Call
30133 one_time_initialization.
30134 * rtl-tests.c (selftest::test_uncond_jump): Call
30135 set_new_first_and_last_insn.
30136 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
30137 * selftest-rtl.c: New file.
30138 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
30139 (selftest::get_insn_by_uid): New decl.
30140 * selftest-run-tests.c (selftest::run_tests): Call
30141 read_rtl_function_c_tests.
30142 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
30143 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
30146 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
30148 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
30149 operands in a special way. Assert that pos+len <= mode precision.
30151 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30153 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
30154 3 argument Alias with unlimited for the negative form.
30155 (fno-vect-cost-model): Removed.
30157 2017-01-05 Martin Liska <mliska@suse.cz>
30159 * hsa-gen.c (gen_hsa_divmod): New function.
30160 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
30162 2017-01-05 Martin Liska <mliska@suse.cz>
30165 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
30168 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30170 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
30171 small constant length operands.
30173 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30175 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
30176 between loop iterations.
30178 2017-01-05 Martin Liska <mliska@suse.cz>
30181 * gimplify.c (gimplify_decl_expr): Compare to
30182 asan_poisoned_variables instread of checking flags.
30183 (gimplify_target_expr): Likewise.
30184 (gimplify_expr): Likewise.
30185 (gimplify_function_tree): Conditionally initialize
30186 asan_poisoned_variables.
30188 2017-01-04 Jeff Law <law@redhat.com>
30190 PR tree-optimizatin/78812
30191 * rtl.h (contains_mem_rtx_p): Prototype.
30192 * ifcvt.c (containts_mem_rtx_p): Move from here to...
30193 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
30194 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
30195 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
30196 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
30198 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30200 * input.c (assert_char_at_range): Default-initialize actual_range.
30202 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30204 * df-scan.c (df_ref_create_structure): Make regno unsigned,
30205 to match the caller.
30207 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30209 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
30210 insns after final jump in test to emit dummy move.
30212 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30214 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
30215 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
30217 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30219 * multiple_target.c (create_dispatcher_calls): Init e_next.
30220 * tree-ssa-loop-split.c (split_loop): Init border.
30221 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
30224 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
30229 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
30230 (altivec_register_operand): Do not return true if the operand
30231 contains a SUBREG mixing SImode and SFmode.
30232 (vsx_register_operand): Likewise.
30233 (vsx_reg_sfsubreg_ok): New predicate.
30234 (vfloat_operand): Do not return true if the operand contains a
30235 SUBREG mixing SImode and SFmode.
30236 (vint_operand): Likewise.
30237 (vlogical_operand): Likewise.
30238 (gpc_reg_operand): Likewise.
30239 (int_reg_operand): Likewise.
30240 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
30241 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
30242 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
30244 (rs6000_emit_move_si_sf_subreg): New helper function.
30245 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
30246 fixup SUBREGs involving SImode and SFmode.
30247 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
30248 numbers for the new peephole2 optimization.
30249 (peephole2 for SFmode unions): New peephole2 to optimize cases in
30250 the GLIBC math library that do AND/IOR/XOR operations on single
30251 precision floating point.
30252 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
30253 target macros to say whether we need to avoid SUBREGs mixing
30255 (TARGET_ALLOW_SF_SUBREG): Likewise.
30256 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
30257 (UNSPEC_SI_FROM_SF): Likewise.
30258 (iorxor): Change spacing.
30259 (and_ior_xor): New iterator for AND, IOR, and XOR.
30260 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
30261 (movdi_from_sf_zero_ext): Likewise.
30262 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
30263 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
30264 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
30265 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
30266 (fms<mode>4): Likewise.
30267 (fnma<mode>4): Likewise.
30268 (fnms<mode>4): Likewise.
30269 (nfma<mode>4): Likewise.
30270 (nfms<mode>4): Likewise.
30272 2017-01-04 Marek Polacek <polacek@redhat.com>
30275 * doc/invoke.texi: Document -Wpointer-compare.
30277 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30279 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
30282 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
30283 descriptions for -gdwarf-5 and emit them as uleb128 instead of
30286 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
30289 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
30290 documentation of the powerpc_popcntb_ok attribute.
30291 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30292 code to issue warning messages if a requested CPU configuration is
30293 not supported by the binary (assembler and loader) toolchain.
30294 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
30295 made to define a built-in function that has been disabled.
30296 (paired_init_builtins): Add assertion to prevent ICE if attempt is
30297 made to define a built-in function that has been disabled.
30298 (altivec_init_builtins): Add comment explaining why definition
30299 of the DST built-in functions is not preceded by an assertion
30300 check. Add assertions to prevent ICE if attempts are made to
30301 define an altivec predicate or an abs* built-in function that has
30303 (htm_init_builtins): Add comment explaining why definition of the
30304 htm built-in functions is not preceded by an assertion check.
30306 2017-01-04 Jeff Law <law@redhat.com>
30308 PR tree-optimizatin/67955
30309 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
30310 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
30311 the points-to solution does not include pt_null. Use DECL_PT_UID
30314 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
30316 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
30317 Use gen_int_mode instead of gen_lopwart for const_int operands.
30319 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30321 PR tree-optimization/71563
30322 * match.pd: Simplify X << Y into X if Y is known to be 0 or
30323 out of range value - has low bits known to be zero.
30325 2017-01-04 Alan Modra <amodra@gmail.com>
30327 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
30328 * configure: Regenerate.
30329 * config.in: Regenerate.
30331 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30334 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
30335 a substring of the message, but strcmp with the whole message. Ifdef
30336 ENABLE_NLS, translate the message first using dgettext.
30338 2017-01-03 Jeff Law <law@redhat.com>
30340 PR tree-optimizatin/78856
30341 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
30342 (mark_threaded_blocks): Remove code to truncate thread paths that
30343 cross multiple loop headers. Instead invalidate the cached loop
30344 iteration information and handle case of a thread path walking
30345 into an irreducible region.
30347 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
30350 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
30351 assertions. Add support for doing the signbit if the IEEE 128-bit
30352 floating point value is in a GPR.
30353 * config/rs6000/rs6000.md (Fsignbit): Delete.
30354 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
30355 Update the length attribute if the value is in a GPR.
30356 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
30357 the sign or zero extension instruction, since the value is always 0/1.
30358 (signbit<mode>2_dm2): Delete using <Fsignbit>.
30361 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
30362 extracting SImode to a GPR register so that we can generate a
30363 store, limit the vector to be in a traditional Altivec register
30364 for the vextuwrx instruction.
30366 2017-01-03 Ian Lance Taylor <iant@google.com>
30368 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
30370 2017-01-03 Martin Sebor <msebor@redhat.com>
30372 PR tree-optimization/78696
30373 * gimple-ssa-sprintf.c (format_floating): Correct handling of
30374 precision. Use MPFR for %f for greater fidelity. Correct handling
30376 (pass_sprintf_length::compute_format_length): Set width and precision
30377 specified by asrerisk to void_node for vararg functions.
30378 (try_substitute_return_value): Adjust dump output.
30380 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
30382 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
30384 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
30386 * doc/invoke.texi (SPARC options): Document -mlra as the default.
30387 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
30388 -mlra/-mno-lra was passed to the compiler.
30390 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
30392 PR rtl-optimization/65618
30393 * emit-rtl.c (try_split): Move initialization of "before" and
30394 "after" to just before the call to emit_insn_after_setloc.
30396 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
30398 * doc/md.texi (Standard Names): Remove reference to Java frontend.
30400 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
30402 * dwarf2out.c (gen_enumeration_type_die): When
30403 -gno-strict-dwarf, add a DW_AT_encoding attribute.
30405 2017-01-03 Jakub Jelinek <jakub@redhat.com>
30407 PR tree-optimization/78965
30408 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
30409 Change first argument from const call_info & to call_info &. For %n
30410 set info.nowrite to false.
30412 PR middle-end/78901
30413 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
30414 possibly throwing calls.
30416 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
30417 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
30418 and fns handling, rather than in a separate case SSA_NAME.
30420 2017-01-02 Jeff Law <law@redhat.com>
30422 * config/darwin-driver.c (darwin_driver_init): Const-correctness
30423 fixes for first_period and second_period variables.
30425 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
30428 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
30429 (*insvqi_1): New insn pattern.
30430 (*insvqi_1_mem_rex64): Ditto.
30431 (*insvqi_2): Ditto.
30432 (*insvqi_3): Rename from *insvqi.
30434 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
30436 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
30438 * doc/cfg.texi (Edges): Remove reference to Java.
30439 (Maintaining the CFG): Ditto.
30441 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30443 PR middle-end/77674
30444 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
30445 transparent aliases.
30447 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30449 PR middle-end/77484
30450 * predict.def (PRED_CALL): Update hitrate.
30451 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
30452 * predict.c (tree_estimate_probability_bb): Split CALL predictor
30453 into direct/indirect/polymorphic variants.
30455 2017-01-01 Jakub Jelinek <jakub@redhat.com>
30457 Update copyright years.
30459 * gcc.c (process_command): Update copyright notice dates.
30460 * gcov-dump.c (print_version): Ditto.
30461 * gcov.c (print_version): Ditto.
30462 * gcov-tool.c (print_version): Ditto.
30463 * gengtype.c (create_file): Ditto.
30464 * doc/cpp.texi: Bump @copying's copyright year.
30465 * doc/cppinternals.texi: Ditto.
30466 * doc/gcc.texi: Ditto.
30467 * doc/gccint.texi: Ditto.
30468 * doc/gcov.texi: Ditto.
30469 * doc/install.texi: Ditto.
30470 * doc/invoke.texi: Ditto.
30472 Copyright (C) 2017 Free Software Foundation, Inc.
30474 Copying and distribution of this file, with or without modification,
30475 are permitted in any medium without royalty provided the copyright
30476 notice and this notice are preserved.