1 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4 * tree-cfg.c (verify_gimple_assign_binary): Don't require
5 types_compatible_p, just that TYPE_MODE is the same.
7 2017-12-07 Martin Sebor <msebor@redhat.com>
10 * attribs.c (empty_attribute_table): Initialize new member of
11 struct attribute_spec.
12 (decl_attributes): Add argument. Handle mutually exclusive
13 combinations of attributes.
14 (selftests::test_attribute_exclusions): New function.
15 (selftests::attribute_c_tests): Ditto.
16 * attribs.h (decl_attributes): Add default argument.
17 * selftest.h (attribute_c_tests): Declare.
18 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
19 * tree-core.h (attribute_spec::exclusions, exclude): New type and
21 * doc/extend.texi (Common Function Attributes): Update const and pure.
22 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
23 of struct attribute_spec.
24 * config/arc/arc.c (arc_attribute_table): Same.
25 * config/arm/arm.c (arm_attribute_table): Same.
26 * config/avr/avr.c ( avr_attribute_table): Same.
27 * config/bfin/bfin.c (bfin_attribute_table): Same.
28 * config/cr16/cr16.c (cr16_attribute_table): Same.
29 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
30 * config/h8300/h8300.c (h8300_attribute_table): Same.
31 * config/i386/i386.c (ix86_attribute_table): Same.
32 * config/ia64/ia64.c (ia64_attribute_table): Same.
33 * config/m32c/m32c.c (m32c_attribute_table): Same.
34 * config/m32r/m32r.c (m32r_attribute_table): Same.
35 * config/m68k/m68k.c (m68k_attribute_table): Same.
36 * config/mcore/mcore.c (mcore_attribute_table): Same.
37 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
38 * config/mips/mips.c (mips_attribute_table): Same.
39 * config/msp430/msp430.c (msp430_attribute_table): Same.
40 * config/nds32/nds32.c (nds32_attribute_table): Same.
41 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
42 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
43 * config/rl78/rl78.c (rl78__attribute_table): Same.
44 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
45 * onfig/rx/rx.c (rx_attribute_table): Same.
46 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
47 * config/sh/sh.c (sh_attribute_table): Same.
48 * config/sparc/sparc.c (sparc_attribute_table): Same.
49 * config/spu/spu.c (spu_attribute_table): Same.
50 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
51 * config/v850/v850.c (v850_attribute_table): Same.
52 * config/visium/visium.c (visium_attribute_table): Same.
54 2017-12-07 Tamar Christina <tamar.christina@arm.com>
57 * config/arm/arm.c (INCLUDE_STRING): Define.
58 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
59 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
62 2017-12-07 Michael Matz <matz@suse.de>
64 Add unroll and jam pass
66 * gimple-loop-jam.c: New file.
67 * Makefile.in (OBJS): Add gimple-loop-jam.o.
68 * common.opt (funroll-and-jam): New option.
69 * opts.c (default_options_table): Add unroll-and-jam at -O3.
70 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
71 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
72 * passes.def: Add pass_loop_jam.
73 * timevar.def (TV_LOOP_JAM): Add.
74 * tree-pass.h (make_pass_loop_jam): Declare.
75 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
76 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
77 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
78 to flow_loop_tree_node_add.
79 (duplicate_subloops, copy_loops_to): Append to sibling list.
80 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
81 * doc/invoke.texi (-funroll-and-jam): Document new option.
82 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
84 2017-12-07 Richard Biener <rguenther@suse.de>
86 PR tree-optimization/83296
87 PR tree-optimization/67769
88 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
89 flow sensitive info in an unrelated BB.
90 (value_replacement): Use reset_flow_sensitive_info.
91 (minmax_replacement): Reset flow sensitive info on the def
92 we move. Do not reset flow sensitive info in the whole BB
94 (abs_replacement): Likewise.
96 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
99 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
100 rs6000_cpu to the given -mcpu=, or to the default processor.
102 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
104 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
105 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
106 instead of rs6000_cpu_attr.
107 (rs6000_adjust_cost): Ditto.
108 (is_microcoded_insn): Ditto.
109 (rs6000_adjust_priority): Ditto.
110 (rs6000_issue_rate): Ditto.
111 (rs6000_use_sched_lookahead): Ditto.
112 (rs6000_use_sched_lookahead_guard): Ditto.
113 (rs6000_sched_reorder): Ditto.
114 (force_new_group): Ditto.
115 * config/rs6000/rs6000.md (cpu attribute): Ditto.
116 (group_ending_nop): Ditto.
118 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
120 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
121 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
122 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
123 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
125 (rs6000_reassociation_width): Ditto.
126 (rs6000_emit_epilogue): Ditto.
127 (rs6000_adjust_cost): Ditto.
128 (is_microcoded_insn): Ditto.
129 (is_cracked_insn): Ditto.
130 (rs6000_adjust_priority): Ditto.
131 (rs6000_sched_reorder): Ditto.
132 (rs6000_sched_reorder2): Ditto.
133 (insn_must_be_first_in_group): Ditto.
134 (insn_must_be_last_in_group): Ditto.
135 (rs6000_register_move_cost): Ditto.
136 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
139 2017-12-07 Julia Koval <julia.koval@intel.com>
141 * config.gcc: Add vaesintrin.h.
142 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI): New type.
143 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
144 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi): New builtins.
145 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
146 * config/i386/immintrin.h: Include vaesintrin.h.
147 * config/i386/sse.md (vaesdec_<mode>): New pattern.
148 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
149 _mm_aesdec_epi128): New intrinsics.
151 2017-12-06 David Malcolm <dmalcolm@redhat.com>
153 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
154 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
155 c-family/c-spellcheck.cc.
156 (best_macro_match::best_macro_match): Likewise.
158 (struct edit_distance_traits<cpp_hashnode *>): Move to
159 c-family/c-spellcheck.h.
160 (class best_macro_match): Likewise.
162 2017-12-06 Jakub Jelinek <jakub@redhat.com>
164 PR tree-optimization/83293
165 * gimple-ssa-strength-reduction.c (insert_initializers): Use
166 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
167 might insert into empty bb.
170 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
171 simplify for plus with :c added, and pointer_plus without that.
172 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
173 with undefined overflow and the conversion is not widening,
174 perform negation in utype and only convert to type afterwards.
175 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
176 simplify for plus with :c added, and pointer_plus without that.
177 If type is integral with undefined overflow and the conversion is
178 not widening, perform minus in utype and only convert to type
179 afterwards. Move the last pointer_diff_expr simplify into the
182 2017-12-06 Martin Sebor <msebor@redhat.com>
184 PR tree-optimization/82646
185 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
188 2017-12-06 Martin Sebor <msebor@redhat.com>
190 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
193 PR tree-optimization/83075
194 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
195 strncat/strncpy don't change length of source string.
197 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
200 2017-11-29 Martin Aberg <maberg@gaisler.com>
202 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
203 to prevent b2bst errata sequence.
204 (sqrtdf2_fix): Likewise.
206 2017-12-06 Jakub Jelinek <jakub@redhat.com>
208 PR tree-optimization/81945
209 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
210 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
211 to dest_cfun has orig_loop_num set, either remap it to the new
212 loop number if the loop got moved too, or clear it.
214 2017-12-05 Steve Ellcey <sellcey@cavium.com>
216 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
218 (thunderx2t99_nothing): New insn reservation.
219 (thunderx2t99_mrs): New insn reservation.
220 (thunderx2t99_multiple): New insn reservation.
221 (thunderx2t99_alu_basi): Add bfx to reservation.
222 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
224 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
227 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
229 2017-12-05 Bin Cheng <bin.cheng@arm.com>
231 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
232 tree-ssa-pre.c::remove_dead_inserted_code.
233 * tree-ssa-dce.h: New file.
234 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
235 (remove_dead_inserted_code): Move and rename to function
236 tree-ssa-dce.c::simple_dce_from_worklist.
237 (pass_pre::execute): Update use.
239 2017-12-05 Richard Biener <rguenther@suse.de>
241 PR tree-optimization/83277
242 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
243 to code-gen liveout vars.
245 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
247 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
248 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
249 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
252 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
254 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
256 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
257 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
260 2017-12-05 Richard Biener <rguenther@suse.de>
262 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL):
264 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
265 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
266 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
268 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
270 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
272 2017-12-05 Martin Liska <mliska@suse.cz>
273 Jakub Jelinek <jakub@redhat.com>
275 * doc/invoke.texi: Document the options.
276 * flag-types.h (enum sanitize_code): Add
277 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
278 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
279 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
280 * opts.c: Define new sanitizer options.
281 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
282 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
284 2017-12-05 Julia Koval <julia.koval@intel.com>
286 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
287 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
288 (ix86_handle_option): Handle -mavx512vnni.
289 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
290 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
291 * config/i386/i386-c (__AVX512VNNI__): New.
292 * config/i386/i386.c (ix86_target_string): Handle new option.
293 (ix86_valid_target_attribute_inner_p): Handle new option.
294 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
295 * config/i386/i386.opt (mavx512vnni): New option.
297 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
300 * athlon.md: Disable for generic.
301 * haswell.md: Enable for generic.
302 * i386.c (ix86_sched_init_global): Add core hooks for generic.
303 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
305 (ix86_adjust_cost): Move generic to haswell path.
307 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
309 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
310 instead of MEM_P in a couple more places. Fix formatting issues.
312 2017-12-04 Jim Wilson <jimw@sifive.com>
314 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
315 instead of GP_REG_LAST-1.
316 (riscv_adjust_libcall_cfi_prologue): Likewise.
317 (riscv_adjust_libcall_cri_epilogue): Likewise.
318 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
321 2017-12-04 Luis Machado <luis.machado@linaro.org>
323 * ipa-pure-const.c (check_decl): Add missing newline.
324 (state_from_flags): Likewise.
326 2017-12-04 Jeff Law <law@redhat.com>
328 PR tree-optimizatin/78496
329 * gimple-ssa-evrp-analyze.h
330 (evrp_range_analyzer::get_vr_values): Simplify.
331 * gimple-ssa-evrp-analyze.c: Corresponding changes.
332 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
333 and gimple-ssa-evrp-analyze.h.
334 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
335 (simplify_stmt_for_jump_threading): Copy a blob of code from
336 tree-vrp.c to use ranges to simplify statements.
337 (dom_opt_dom_walker::before_dom_children): Call
338 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
339 (dom_opt_dom_walker::after_dom_children): Similarly for
340 evrp_range_analyzer::leave.
341 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
344 * gimple-ssa-evrp-analyze.c
345 (evrp_range_analyzer::extract_range_from_stmt): Always use
346 vr_values::update_value_range so preexisting range info is
347 medged with new range info, even if the new range is VR_VARYING.
349 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
351 * combine.c: Adjust comment.
352 (use_crosses_set_p): Delete.
353 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
354 (try_combine): Ditto.
356 2017-12-04 Richard Biener <rguenther@suse.de>
358 PR tree-optimization/83255
359 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
360 Re-add zero-iteration check.
362 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
364 PR rtl-optimization/83245
365 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
366 hard registers as earlyclobber, also if not in an asm.
368 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
372 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
375 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
376 rs6000_cpu based on cpu_index, not tune_index.
378 2017-12-04 Richard Biener <rguenther@suse.de>
380 PR tree-optimization/83238
381 * graphite-scop-detection.c (scop_detection::merge_sese): Make
382 code match comment, rejecting invalid SESE regions.
384 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
386 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
387 require base operand is a REG_POINTER prior to reload on targets
388 with non-equivalent space registers.
390 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
392 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
393 (update_specialized_profile): Fix updating of counts.
394 (perhaps_add_new_callers): Likewise.
396 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
399 * x86-tune.def: Remove obsolette FIXMEs.
400 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
401 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
402 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
404 (X86_TUNE_PAD_RETURNS): Disable for generic.
406 2017-12-02 Jakub Jelinek <jakub@redhat.com>
408 PR tree-optimization/83170
409 PR tree-optimization/83241
410 * gimple-ssa-store-merging.c
411 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
412 gimple_vuse (ins_stmt) in case it has changed.
413 (imm_store_chain_info::output_merged_store): Likewise.
415 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
419 * tree-cfg.c (pass_warn_function_return::execute): Handle
420 __builtin_ubsan_handle_missing_return like __builtin_unreachable
421 with BUILTINS_LOCATION.
425 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
426 is BLKmode for vector field with vector raw mode, use TYPE_MODE
427 instead of DECL_MODE.
429 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
430 last argument to rtx pointer.
431 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
432 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
433 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
434 depending on the chosen ISAs.
435 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
436 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
437 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
439 * config/i386/sse.md (mov<mode>_internal): Likewise.
440 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
442 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
444 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
446 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
448 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
449 parameters from prototype.
450 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
451 parameters. Don't print a comment.
452 (emit_fusion_gpr_load): Adjust.
453 (emit_fusion_load_store): Adjust.
454 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
455 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
456 comment on the second line.
458 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
461 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
462 rs6000_cpu based on cpu_index, not tune_index.
464 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
466 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
467 which_alternative instead of which_alternative + 1.
468 (output_asm_insn): Print an extra tab if the template is short.
470 2017-12-01 Jim Wilson <jimw@sifive.com>
472 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
474 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
475 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
477 * doc/tm.texi: Regenerate.
479 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
482 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
483 whether we can allocate pseudos before trying to fix an address.
484 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
485 memory address is indexed or indirect.
486 (floatuns_<mode>si2_hw2): Likewise.
488 2017-12-01 Jason Merrill <jason@redhat.com>
490 * Makefile.in (TAGS): Add c-family/*.cc.
492 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
494 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
495 (call_value_insn): Likewise.
496 (sibcall_insn): Likewise.
497 (sibcall_value_insn): Likewise.
498 (movsi_aarch64): Likewise.
499 (movdi_aarch64): Likewise.
500 (add_losym_): Likewise.
501 (ldr_got_small_): Likewise.
502 (ldr_got_small_sidi): Likewise.
503 (ldr_got_small_28k_): Likewise.
504 (ldr_got_small_28k_sidi): Likewise.
505 * config/aarch64/aarch64.c (aarch64_print_address_internal):
506 Move output_addr_const to symbolic case. Add error check.
508 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
510 * config/s390/predicates.md (plus16_Q_operand): New predicate.
511 * config/s390/s390.md: Disable MVC merging peephole if it would
512 disable operand forwarding.
513 (new peephole2): Split MVCs if it would turn them into up to 2
516 2017-12-01 Richard Biener <rguenther@suse.de>
518 PR tree-optimization/83232
519 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
520 detection of same access. Instead of breaking the group here
521 do not consider the duplicate. Add comment explaining real fix.
523 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
525 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
527 2017-12-01 Sudakshina Das <sudi.das@arm.com>
529 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
531 2017-12-01 Jakub Jelinek <jakub@redhat.com>
533 * function.h (struct function): Remove cilk_frame_decl,
534 is_cilk_function and calls_cilk_spawn fields.
535 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
537 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
538 cilk_elemental field.
539 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
540 * target.def: Adjust comment.
541 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
542 Don't test cilk_elemental.
544 PR tree-optimization/83233
545 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
546 bswap_stat name for the struct.
549 * tree.h (SWITCH_BREAK_LABEL_P): Define.
550 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
551 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
552 SWITCH_BREAK_LABEL_P set on the label.
553 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
554 added for default case if it was missing and not all cases covered.
555 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
556 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
558 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
559 to avoid -Wimplicit-fallthrough warning.
560 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
561 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
563 PR tree-optimization/83221
564 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
566 (init_reassoc): Formatting fix.
569 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
570 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
574 * tree-cfg.c: Include asan.h.
575 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
577 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
579 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
581 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
583 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
584 vpcmpeqd instruction.
586 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
588 * config/i386/i386.c (standard_sse_constant_opcode): Fix
589 registers type for 128bit mode.
591 2017-11-30 Jakub Jelinek <jakub@redhat.com>
593 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
595 * gengtype-state.c (read_state_pair): Likewise.
596 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
597 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
598 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
599 * ifcvt.c (noce_try_store_flag_constants): Likewise.
600 * tree-ssa-ccp.c (ccp_finalize): Likewise.
601 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
602 * builtins.c (fold_builtin_3): Likewise.
603 * graphite-scop-detection.c
604 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
605 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
608 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
611 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
612 parameter. Return true if ignore_decl_rtl_set_p is true and other
613 conditions are satisfied.
614 * asan.h (asan_protect_global): Add new parameter.
615 * varasm.c (categorize_decl_for_section): Pass true as second parameter
616 to asan_protect_global calls.
618 2017-11-30 Jim Wilson <jimw@sifive.com>
620 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
621 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
622 -mstrict-align, add info on default value. Delete redundant lines for
623 -mabi. Add missing -mexplicit-relocs docs.
625 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
627 * config/arc/arc.md (trap): New pattern.
629 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
631 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
632 instruction to end into a delay slot.
633 * config/arc/arc.md (cond_delay_insn): Check if the instruction
634 can be placed into a delay slot against reg_note.
636 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
638 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
639 labels number of usages.
641 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
643 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
645 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
647 2017-11-30 Jakub Jelinek <jakub@redhat.com>
650 * internal-fn.c (expand_mul_overflow): Optimize unsigned
651 multiplication by power of 2 constant into two shifts + comparison.
653 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
656 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
658 2017-11-30 Richard Biener <rguenther@suse.de>
660 PR tree-optimization/83202
661 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
662 allow_peel argument and guard peeling.
663 (canonicalize_loop_induction_variables): Likewise.
664 (canonicalize_induction_variables): Pass false.
665 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
666 peeling from cunrolli.
668 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
670 * combine.c (try_combine): Print a message to dump file whenever
671 I0, I1, or I2 cannot be combined into I3.
673 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
675 PR rtl-optimization/83156
676 PR rtl-optimization/82621
677 * combine.c (try_combine): Don't split an I2 if one of the dests is
678 set again before I3. Allow unused dests.
680 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
682 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
684 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
686 PR rtl-optimization/80818
687 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
688 recursively. Use insn code for clobber.
689 (lra_set_insn_recog_data): Pass the new arg to
690 collect_non_operand_hard_regs.
691 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
693 (lra_update_insn_regno_info): Pass insn to
694 add_regs_to_insn_regno_info.
696 2017-11-29 Jim Wilson <jimw@sifive.com>
697 Andrew Waterman <andrew@sifive.com>
699 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
700 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
701 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
702 use SINGLE_SHIFT_COST.
703 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
704 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
706 2017-11-29 Julia Koval <julia.koval@intel.com>
708 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
709 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
710 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
711 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
712 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
713 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
714 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
715 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
716 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
717 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
718 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
719 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
720 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
721 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
722 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
723 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
724 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
725 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
726 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
727 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
728 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
729 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
730 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
731 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
732 __builtin_ia32_vpshldv_v2di_mask,
733 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
734 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
735 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
737 2017-11-29 Julia Koval <julia.koval@intel.com>
739 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
740 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
741 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
742 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
743 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
744 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
745 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
746 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
747 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
748 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
749 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
750 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
751 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
752 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
753 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
754 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
755 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
756 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
757 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
758 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
759 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
761 2017-11-29 Daniel Cederman <cederman@gaisler.com>
763 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
764 movsi_pic_gotdata_op instruction as a load for the UT699 errata
767 2017-11-29 Martin Aberg <maberg@gaisler.com>
769 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
770 to prevent b2bst errata sequence.
771 (sqrtdf2_fix): Likewise.
773 2017-11-29 Daniel Cederman <cederman@gaisler.com>
775 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
776 (div_sqrt_insn_p): New function.
777 (sparc_do_work_around_errata): Insert NOP instructions to
778 prevent sequences that could trigger the TN-0013 errata for
779 certain LEON3 processors.
780 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
781 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
782 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
783 (in_branch_delay): Prevent div and sqrt in delay slot if
785 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
787 2017-11-29 Daniel Cederman <cederman@gaisler.com>
789 * config/sparc/sparc.c (atomic_insn_p): New function.
790 (sparc_do_work_around_errata): Insert NOP instructions to
791 prevent sequences that could trigger the TN-0010 errata for
793 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
794 instruction referable in atomic_insns_p.
796 2017-11-29 Daniel Cederman <cederman@gaisler.com>
798 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
799 (atomic_compare_and_swap_leon3_1): Likewise.
802 2017-11-29 Daniel Cederman <cederman@gaisler.com>
804 * config/sparc/sparc.c (fpop_insn_p): New function.
805 (sparc_do_work_around_errata): Insert NOP instructions to
806 prevent sequences that could trigger the TN-0012 errata for
808 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
809 * config/sparc/sparc.md (fix_gr712rc): New attribute.
810 (in_branch_annul_delay): Prevent floating-point instructions
811 in delay slot of annulled integer branch.
813 2017-11-29 Richard Biener <rguenther@suse.de>
815 PR tree-optimization/83202
816 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
818 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
819 nodes vectorized to the same stmts multiple times.
820 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
821 (vect_analyze_slp_instance): Adjust.
822 (scalar_stmts_to_slp_tree_map_t): New typedef.
823 (vect_schedule_slp_instance): Add a map recording the SLP node
824 representing the vectorized stmts for a set of scalar stmts.
825 Avoid code-generating redundancies.
826 (vect_schedule_slp): Allocate map and pass it down.
828 2017-11-29 Nathan Sidwell <nathan@acm.org>
831 * tree.c (build_complex_type): Fix canonicalization. Only fill in
834 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
836 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
837 cases for printing LDP/STP memory addresses.
838 (aarch64_print_address_internal): Renamed from
839 aarch64_print_operand_address, added parameter, add Pmode check.
840 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
841 (aarch64_print_operand_address): Indirect to
842 aarch64_print_address_internal.
843 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
844 'y' operand output specifier.
846 2017-11-29 Jakub Jelinek <jakub@redhat.com>
849 * tree.c (build_simple_mem_ref_loc): Handle
850 get_addr_base_and_unit_offset returning a MEM_REF.
853 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
856 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
859 2017-11-28 David Malcolm <dmalcolm@redhat.com>
861 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
863 (layout::move_to_column): Likewise.
865 2017-11-28 Jakub Jelinek <jakub@redhat.com>
868 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
869 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
871 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
872 Martin Jambor <mjambor@suse.cz>
875 * tree.h (expr_type_first_operand_type_p): Declare
876 * tree.c (expr_type_first_operand_type_p): New function.
877 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
878 (ipa_value_from_jfunc): Adjust declaration.
879 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
880 Use it as result type for arithmetics, unless it is NULL in which case
881 be more conservative.
882 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
883 ipa_get_jf_pass_through_result.
884 (propagate_vals_across_pass_through): Likewise.
885 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
886 is to propagate_vals_across_pass_through.
887 (propagate_constants_across_call): Pass PARM_TYPE to
888 propagate_scalar_across_jump_function.
889 (find_more_scalar_values_for_callers_subset): Pass parameter type to
890 ipa_value_from_jfunc.
891 (cgraph_edge_brings_all_scalars_for_node): Likewise.
892 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
893 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
894 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
895 target_type, pass it to ipa_value_from_jfunc.
896 (update_indirect_edges_after_inlining): Pass parameter type to
897 try_make_edge_direct_simple_call.
899 2017-11-28 Jeff Law <law@redhat.com>
901 * gimple-ssa-evrp-analyze.c
902 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
903 refine ranges if scev_initialized_p returns true.
904 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
906 2017-11-28 Julia Koval <julia.koval@intel.com>
908 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
909 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
910 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
911 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
912 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
913 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
914 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
915 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
916 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
917 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
918 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
919 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
920 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
921 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
922 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
923 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
924 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
925 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
926 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
927 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
928 __builtin_ia32_vpshrd_v2di_mask): New builtins.
929 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
931 2017-11-28 Julia Koval <julia.koval@intel.com>
933 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
934 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
935 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
936 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
937 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
938 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
939 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
940 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
941 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
942 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
943 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
944 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
945 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
946 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
947 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
948 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
949 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
950 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
951 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
952 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
953 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
954 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
955 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
956 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
957 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
958 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
959 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
960 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
961 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
962 __builtin_ia32_vpshld_v2di_mask): New builtins.
963 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
964 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
966 2017-11-28 Richard Biener <rguenther@suse.de>
968 PR tree-optimization/80776
969 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
971 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
973 (evrp_range_analyzer::record_ranges_from_incoming_edges):
974 If the incoming edge is an effective fallthru because the other
975 edge only reaches a __builtin_unreachable () then record ranges
976 derived from the controlling condition in SSA info.
977 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
978 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
980 2017-11-28 Olivier Hainque <hainque@adacore.com>
982 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
983 on mingw build hosts.
985 2017-11-28 Jakub Jelinek <jakub@redhat.com>
988 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
990 2017-11-28 Richard Biener <rguenther@suse.de>
993 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
994 copies generated from memcpy use a character array as reference
997 2017-11-28 Julia Koval <julia.koval@intel.com>
998 Sebastian Peryt <sebastian.peryt@intel.com>
1000 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1001 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1002 cilk-common.o, cilk.h, cilk-common.c): Remove.
1004 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1005 * builtins.c (is_builtin_name): Remove cilkplus condition.
1006 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1007 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1008 cilk-builtins.def, cilkplus.def): Remove.
1009 * cif-code.def (CILK_SPAWN): Remove.
1010 * cilk-builtins.def: Delete.
1011 * cilk-common.c: Ditto.
1013 * cilkplus.def: Ditto.
1014 * config/darwin.h (fcilkplus): Delete.
1015 * cppbuiltin.c: Ditto.
1016 * doc/extend.texi: Remove cilkplus doc.
1017 * doc/generic.texi: Ditto.
1018 * doc/invoke.texi: Ditto.
1019 * doc/passes.texi: Ditto.
1020 * gcc.c (fcilkplus): Remove.
1021 * gengtype.c (cilk.h): Remove.
1022 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
1024 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
1026 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
1028 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
1029 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
1030 cilkplus conditions.
1031 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
1032 inline_read_section): Ditto.
1033 * ipa-inline-analysis.c (cilk.h): Remove.
1034 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
1035 * lto-wrapper.c (merge_and_complain, append_compiler_options,
1036 append_linker_options): Remove condition for fcilkplus.
1037 * lto/lto-lang.c (cilk.h): Remove.
1038 (lto_init): Remove condition for fcilkplus.
1039 * omp-expand.c (expand_cilk_for_call): Delete.
1040 (expand_omp_taskreg, expand_omp_for_static_chunk,
1041 expand_omp_for): Remove cilkplus
1043 (expand_cilk_for): Delete.
1044 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
1045 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
1046 execute_lower_omp, diagnose_sb_0): Ditto.
1047 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
1048 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1049 * tree-nested.c: Ditto.
1050 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
1051 (dump_generic_node): Ditto.
1052 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1053 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
1055 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
1057 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1059 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
1060 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
1061 for vector mode and !TARGET_SIMD.
1063 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1065 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
1067 * tree.h (SWITCH_LABELS): Remove.
1068 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
1069 assert SWITCH_BODY is non-NULL.
1070 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
1072 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
1074 PR tree-optimization/80788
1075 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
1076 has TREE_OVERFLOW set, call drop_tree_overflow.
1078 2017-11-28 Richard Biener <rguenther@suse.de>
1080 PR tree-optimization/83158
1081 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
1084 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
1087 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
1088 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
1090 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
1092 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
1093 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
1095 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1098 * varasm.c (compare_constant): Take the mode of the constants into
1099 account when comparing floating point constants.
1101 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
1103 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
1104 from explicit instantiation of debug_helper.
1105 * vec.h (DEFINE_DEBUG_VEC): Ditto.
1107 2017-11-27 Richard Biener <rguenther@suse.de>
1109 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
1112 2017-11-27 Richard Biener <rguenther@suse.de>
1114 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
1115 (type_hash_canon): Also clear int_cst_hash_table entry for
1117 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
1119 2017-11-27 Tamar Christina <tamar.christina@arm.com>
1121 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1122 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1123 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1125 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1127 * hash-map.h (gt_cleare_cache): Avoid UB.
1129 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
1131 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
1132 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
1133 (decide_unroll_runtime_iterations): Likewise.
1134 (decide_unroll_stupid): Likewise.
1136 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1139 * config/i386/i386.md: Add a loop with incssp.
1141 2017-11-27 Martin Jambor <mjambor@suse.cz>
1143 PR tree-optimization/81248
1144 * tree-sra.c (splice_param_accesses): Remove size check.
1145 (decide_one_param_reduction): Fix size check.
1146 * gimple-pretty-print.c (dump_profile): Silence warning.
1147 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
1149 2017-11-27 Jakub Jelinek <jakub@redhat.com>
1152 * dbxout.c (lastlineno): New variable.
1153 (dbx_debug_hooks): Use dbxout_switch_text_section as
1154 switch_text_section debug hook.
1155 (dbxout_function_end): Switch to current_function_section
1156 rather than function_section. If crtl->has_bb_partition,
1157 output just one N_FUN, depending on in_cold_section_p.
1158 (dbxout_source_line): Remember last lineno in lastlineno.
1159 (dbxout_switch_text_section): New function.
1160 (dbxout_function_decl): Adjust dbxout_block caller.
1161 (dbx_block_with_cold_children): New function.
1162 (dbxout_block): Return true if any LBRAC/RBRAC have been
1163 emitted. Use dbx_block_with_cold_children at depth == 0
1164 in second partition. Add PARENT_BLOCKNUM argument, pass
1165 it optionally adjusted to children. Output LBRAC/RBRAC
1166 around recursive call only if the block is in the current
1167 partition, if not and anything was output, emit empty
1169 * final.c (final_scan_insn): Compute cold_function_name
1170 before calling switch_text_section debug hook. Call
1171 that hook even if dwarf2out_do_frame if not emitting
1175 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
1176 TREE_READONLY decls.
1178 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1180 PR rtl-optimization/82488
1181 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
1183 2017-11-26 Julia Koval <julia.koval@intel.com>
1185 * config/i386/i386.c (processor_target_table): Add skylake_cost for
1187 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
1190 2017-11-26 Julia Koval <julia.koval@intel.com>
1192 * config/i386/driver-i386.c (host_detect_local_cpu):
1193 Detect skylake-avx512.
1195 2017-11-26 Julia Koval <julia.koval@intel.com>
1197 * config.gcc: Add -march=cannonlake.
1198 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
1199 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
1200 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
1201 (PTA_CANNONLAKE): New.
1202 (processor_target_table): Add cannonlake.
1203 (ix86_option_override_internal): Ditto.
1204 (fold_builtin_cpu): Ditto.
1205 (get_builtin_code_for_version): Handle cannonlake.
1206 (M_INTEL_COREI7_CANNONLAKE): New.
1207 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
1208 * doc/invoke.texi: Add -march=cannonlake.
1210 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
1212 * plugin.c (add_new_plugin): Use platform-specific library extensions.
1213 (try_init_one_plugin): Alternative implementation for MinGW.
1214 * Makefile.in (plugin_implib): New.
1215 (gengtype-lex.c): Fix broken AIX workaround.
1216 * configure: Regenerate.
1217 * doc/plugins.texi: Document support for MinGW.
1219 2017-11-25 Jakub Jelinek <jakub@redhat.com>
1221 PR rtl-optimization/81553
1222 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
1223 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
1224 is a shift where C1 has different mode than the whole shift, use C1's
1225 mode for MULT rather than the shift's mode.
1228 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
1229 builtins not enabled in the currently selected ISA.
1231 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
1233 PR tree-optimization/71026
1234 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
1235 (insert_reciprocals): Change to insert reciprocals before a division
1236 by a square and to insert the square of a reciprocal.
1237 (execute_cse_reciprocals_1): Change to consider division by a square.
1238 (register_division_in): Add importance parameter.
1240 2017-11-24 Richard Biener <rguenther@suse.de>
1242 PR tree-optimization/82402
1243 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
1244 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
1246 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
1248 * match.pd (0-ptr): New transformation.
1250 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
1253 * ipa-inline.c (inline_small_functions): Set current badnes correctly
1254 when skipping checking.
1256 2017-11-24 Richard Biener <rguenther@suse.de>
1258 PR tree-optimization/83128
1259 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
1260 (vn_reference_lookup_3): Likewise.
1262 2017-11-24 Jakub Jelinek <jakub@redhat.com>
1265 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
1266 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
1269 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
1270 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
1271 check for the latter with gimple_call_builtin_p. Do not handle
1272 BUILT_IN_STPNCPY_CHK which is not a pass through call.
1274 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
1276 * config/arm/arm_neon.h: Fix pragma GCC push_options before
1279 2017-11-23 Julia Koval <julia.koval@intel.com>
1281 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
1282 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
1283 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
1284 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
1285 _mm512_maskz_expandloadu_epi16): New intrinsics.
1286 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
1287 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
1288 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
1289 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
1290 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
1291 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
1292 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
1293 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
1294 _mm256_maskz_expandloadu_epi8): New intrinsics.
1295 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
1296 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
1297 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
1298 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
1299 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
1300 * config/i386/sse.md (VI248_VLBW): New iterator.
1301 (expand<mode>_mask, expand<mode>_maskz): New patterns.
1303 2017-11-23 Julia Koval <julia.koval@intel.com>
1305 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
1306 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
1307 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
1308 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
1309 _mm512_mask_compressstoreu_epi16): New.
1310 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
1311 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
1312 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
1313 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
1314 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
1315 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
1316 _mm256_mask_compressstoreu_epi8): New.
1317 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
1318 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
1319 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
1320 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
1321 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
1322 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
1323 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
1324 __builtin_ia32_compresshi128_mask,
1325 __builtin_ia32_compressstoreuqi512_mask,
1326 __builtin_ia32_compressstoreuhi512_mask,
1327 __builtin_ia32_compressstoreuqi256_mask,
1328 __builtin_ia32_compressstoreuqi128_mask,
1329 __builtin_ia32_compressstoreuhi256_mask,
1330 __builtin_ia32_compressstoreuhi128_mask): New builtins.
1331 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
1333 (ix86_expand_special_args_builtin): Handle new types.
1334 (s4fma_expand): Handle new builtin array.
1335 * config/i386/immintrin.h: Include new headers.
1336 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
1337 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
1339 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1341 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
1342 fixes. Declare temp and g variables at the top in order to avoid
1343 {} in most of the cases.
1345 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
1347 * match.pd (ptr-0): New transformation.
1349 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1351 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
1352 (struct addr_mode_cost_table): New.
1353 (struct tune_params): Add field addr_mode_costs.
1354 * config/arm/arm.c (generic_addr_mode_costs): New.
1355 (arm_slowmul_tune): Initialise addr_mode_costs field.
1356 (arm_fastmul_tune): Likewise.
1357 (arm_strongarm_tune): Likewise.
1358 (arm_xscale_tune): Likewise.
1359 (arm_9e_tune): Likewise.
1360 (arm_marvell_pj4_tune): Likewise.
1361 (arm_v6t2_tune): Likewise.
1362 (arm_cortex_tune): Likewise.
1363 (arm_cortex_a8_tune): Likewise.
1364 (arm_cortex_a7_tune): Likewise.
1365 (arm_cortex_a15_tune): Likewise.
1366 (arm_cortex_a35_tune): Likewise.
1367 (arm_cortex_a53_tune): Likewise.
1368 (arm_cortex_a57_tune): Likewise.
1369 (arm_exynosm1_tune): Likewise.
1370 (arm_xgene1_tune): Likewise.
1371 (arm_cortex_a5_tune): Likewise.
1372 (arm_cortex_a9_tune): Likewise.
1373 (arm_cortex_a12_tune): Likewise.
1374 (arm_cortex_a73_tune): Likewise.
1375 (arm_v7m_tune): Likewise.
1376 (arm_cortex_m7_tune): Likewise.
1377 (arm_v6m_tune): Likewise.
1378 (arm_fa726te_tune): Likewise.
1379 (arm_mem_costs): Use table lookup to calculate cost of addressing
1382 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1384 * config/arm/arm.c (arm_mem_costs): New function.
1385 (arm_rtx_costs_internal): Use arm_mem_costs.
1387 2017-11-23 Mark Wielaard <mark@klomp.org>
1389 * dwarf2out.c (init_sections_and_labels): Use generation to create
1390 unique ranges_section_label and ranges_base_label. Return generation.
1391 (output_rnglists): Add generation argument. Use generation to create
1392 unique ranges labels.
1393 (dwarf2out_finish): Get generation from init_sections_and_labels
1394 and pass generation to output_rnglists.
1396 2017-11-23 Mike Stump <mikestump@comcast.net>
1397 Eric Botcazou <ebotcazou@adacore.com>
1399 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
1400 * cfgloop.h (struct loop): Add unroll field.
1401 * function.h (struct function): Add has_unroll bitfield.
1402 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
1403 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
1404 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
1405 (pass_rtl_unroll_loops::gate): Likewise.
1406 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
1407 for which loop->unroll==1.
1408 (decide_unroll_constant_iterations): Use note for consistency and
1409 take loop->unroll into account. Return early if loop->unroll is set.
1410 Fix thinko in existing test.
1411 (decide_unroll_runtime_iterations): Use note for consistency and
1412 take loop->unroll into account.
1413 (decide_unroll_stupid): Likewise.
1414 * lto-streamer-in.c (input_cfg): Read loop->unroll.
1415 * lto-streamer-out.c (output_cfg): Write loop->unroll.
1416 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
1418 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
1419 (print_loop): Print loop->unroll if set.
1420 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
1421 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
1422 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
1424 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
1425 loop->unroll is set and smaller than the trip count. Otherwise bypass
1426 entirely the heuristics if loop->unroll is set. Remove dead note.
1427 Fix off-by-one bug in other note.
1428 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
1429 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
1431 (tree_unroll_loops_completely): Make static.
1432 (pass_complete_unroll::execute): Use correct type for variable.
1433 (pass_complete_unrolli::execute): Fix formatting.
1434 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
1436 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
1438 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
1439 enable when TARGET_PREFER_AVX128 is set.
1441 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1443 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
1445 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
1446 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
1448 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1450 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
1451 frequencies. Fix estimation of aggregate parameters.
1453 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1455 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
1458 2017-11-23 Tom de Vries <tom@codesourcery.com>
1460 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
1462 2017-11-23 Tom de Vries <tom@codesourcery.com>
1464 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
1467 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
1470 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
1471 sibcall_value_pcrel_fdpic): Use local variable instead of
1473 (calli_tbr_rel): Add missing operand 2.
1474 (call_valuei_tbr_rel): Add missing operand 3.
1476 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1479 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
1480 bitpos/bitsize covering the whole destination, use store_expr only if
1481 the complex mode is the same. Otherwise, use expand_normal and if
1482 it returns CONCAT, subreg each part separately instead of trying to
1483 subreg the whole result.
1485 2017-11-23 Richard Biener <rguenther@suse.de>
1487 PR tree-optimization/23094
1488 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
1489 come here from walking over backedges in the first iteration.
1490 (vn_reference_lookup_3): Skip clobbers that store the same value.
1492 2017-11-23 Richard Biener <rguenther@suse.de>
1494 PR tree-optimization/81403
1495 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
1496 a block we need a leader relative to.
1497 (phi_translate_1): For nary processing require a leader from
1498 get_representative_for given we run expression simplification
1499 using match-and-simplify. Remove previous fix.
1501 2017-11-22 Jeff Law <law@redhat.com>
1503 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
1504 Use new method allocate_value_range rather than accessing the
1505 vrp_value_range_pool data member directly.
1506 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
1507 to use extract_range_from_stmt method to avoid need for
1508 extract_range_from_assignment method.
1509 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
1510 method rather than setting values_propgated data member directly.
1511 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
1512 and values propagated data members and extract_range_from_assignment
1513 method. Reorder private data members to conform to standards.
1514 Add new methods set_lattice_propagation_complete and
1515 allocate_value_range.
1517 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
1519 PR rtl-optimization/83030
1520 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
1521 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
1522 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
1523 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
1524 CROSSING_JUMP_P flag is already set before setting it.
1525 * cfgrtl.c (fixup_partition_crossing): Likewise.
1526 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
1529 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1531 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
1532 Handle the case where both arguments are using gen_const_vec_series.
1534 2017-11-22 David Malcolm <dmalcolm@redhat.com>
1537 * pretty-print.c (pp_format): Move quoting implementation to
1538 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
1539 to pass address of "quote" local.
1540 (pp_begin_quote): New function.
1541 (pp_end_quote): New function.
1542 * pretty-print.h (printer_fn): Convert penultimate param from bool
1544 (pp_begin_quote): New decl.
1545 (pp_end_quote): New decl.
1546 * tree-diagnostic.c (default_tree_printer): Convert penultimate
1547 param from bool to bool *.
1548 * tree-diagnostic.h (default_tree_printer): Likewise.
1550 2017-11-22 Jeff Law <law@redhat.com>
1552 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
1553 Initialize vr_values.
1554 (evrp_range_analyzer::try_find_new_range): Call methods attached to
1555 vr_values via vr_values class instance rather than delegators.
1556 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
1557 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1558 (evrp_range_analyzer::push_value_range): Likewise.
1559 (evrp_range_analyzer::pop_value_range): Likewise.
1560 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
1561 most delegators. Those remaining are exposed as public interfaces.
1562 Make vr_values a pointer and private.
1563 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
1565 (evrp_range_analyzer::get_vr_value): New method.
1566 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
1567 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
1568 (class evrp_dom_walker): Attach evrp_folder class, initialize
1569 it in the ctor. Remove temporary delegators.
1570 (evrp_dom_walker::before_dom_children): Call methods in attached
1571 evrp_range_analyzer class via class instance pointer. Use
1572 free value_range_constant_singleton to remove need for
1573 op_with_constant_singleton_value delegator method. Do not
1574 create a vrp_prop class instance for every call! Narrow
1575 scope of a couple variables.
1576 (evrp_dom_walker::cleanup): Call methods in attached
1577 evrp_range_analyzer class via class instance pointer.
1578 * vr-values.h (class vr_values): Privatize many methods and
1581 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1583 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
1584 * cfgexpand.c (expand_debug_expr): Remove handling for them.
1585 * expr.c (expand_expr_real_2): Likewise.
1586 * fold-const.c (const_unop): Likewise.
1587 * optabs-tree.c (optab_for_tree_code): Likewise.
1588 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1589 * tree-inline.c (estimate_operator_cost): Likewise.
1590 * tree-pretty-print.c (dump_generic_node): Likewise.
1591 (op_code_prio): Likewise.
1592 (op_symbol_code): Likewise.
1593 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
1594 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
1595 * internal-fn.c (direct_internal_fn_optab): New function.
1596 (direct_internal_fn_array, direct_internal_fn_supported_p
1597 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
1598 * fold-const-call.c (fold_const_reduction): New function.
1599 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
1601 * tree-vect-loop.c: Include internal-fn.h.
1602 (reduction_code_for_scalar_code): Rename to...
1603 (reduction_fn_for_scalar_code): ...this and return an internal
1605 (vect_model_reduction_cost): Take an internal_fn rather than
1607 (vect_create_epilog_for_reduction): Likewise. Build calls rather
1609 (vectorizable_reduction): Use internal functions rather than tree
1610 codes for the reduction operation. Update calls to the functions
1612 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
1613 Use calls to internal functions rather than REDUC tree codes.
1614 * config/aarch64/aarch64-simd.md: Update comment accordingly.
1616 2017-11-22 Olivier Hainque <hainque@adacore.com>
1618 * config/vxworks.c (vxworks_override_options): Pick default
1619 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
1620 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
1621 DWARF_GNAT_ENCODINGS_DEFAULT.
1622 * config/vxworksae.h: Likewise.
1624 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
1626 PR tree-optimization/83104
1627 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
1630 2017-11-22 Marek Polacek <polacek@redhat.com>
1631 H.J. Lu <hongjiu.lu@intel.com>
1632 Jason Merrill <jason@redhat.com>
1637 * calls.c (initialize_argument_information): Call
1638 warn_parameter_passing_abi target hook.
1639 (store_one_arg): Use 0 for empty record size. Don't push 0 size
1640 argument onto stack.
1641 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
1642 * common.opt: Update -fabi-version description.
1643 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
1644 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
1646 (ix86_is_empty_record): New function.
1647 (ix86_warn_parameter_passing_abi): New function.
1648 (TARGET_EMPTY_RECORD_P): Redefine.
1649 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
1650 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
1651 * doc/tm.texi: Regenerated.
1652 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
1653 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
1654 * dwarf2out.c (get_ultimate_context): Move to tree.c.
1655 * explow.c (hard_function_value): Call arg_int_size_in_bytes
1656 instead of int_size_in_bytes.
1657 * expr.c (copy_blkmode_to_reg): Likewise.
1658 * function.c (aggregate_value_p): Return 0 for empty types.
1659 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
1660 (locate_and_pad_parm): Call arg size_in_bytes instead
1662 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
1663 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
1664 * target.def (empty_record_p, warn_parameter_passing_abi): New target
1666 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
1667 (std_gimplify_va_arg_expr): Skip empty records. Call
1668 arg_size_in_bytes instead size_in_bytes.
1669 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
1670 * tree-core.h (tree_type_common): Add empty_flag.
1671 (tree_decl_common): Update comments.
1672 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
1674 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1675 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
1677 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1678 * tree.c (default_is_empty_type): New function.
1679 (default_is_empty_record): New function.
1680 (arg_int_size_in_bytes): New function.
1681 (arg_size_in_bytes): New function.
1682 (get_ultimate_context): New function.
1683 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
1684 TRANSLATION_UNIT_WARN_EMPTY_P.
1685 (default_is_empty_record, arg_int_size_in_bytes,
1686 arg_size_in_bytes, get_ultimate_context): Declare.
1688 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1690 * config/arm/arm.c (cmse_clear_registers): New function.
1691 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
1692 code by call to cmse_clear_registers.
1693 (cmse_nonsecure_entry_clear_before_return): Likewise.
1695 2017-11-22 Tamar Christina <tamar.christina@arm.com>
1697 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
1698 (vdot_s32, vdotq_s32): New.
1699 (vdot_lane_u32, vdotq_lane_u32): New.
1700 (vdot_lane_s32, vdotq_lane_s32): New.
1703 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1706 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
1707 for unsigned values with fewer HWIs than the precision.
1708 (test_overflow): New function.
1709 (wide_int_cc_tests): Call it.
1711 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1713 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
1714 has the same mode class as Pmode.
1716 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1718 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
1719 padding_bits_to_clear_ptr.
1720 (cmse_nonsecure_entry_clear_before_return): Likewise.
1722 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1724 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
1725 auto_sbitap instead of integer bitfield to control register needing
1728 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1730 PR tree-optimization/83044
1731 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
1732 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
1733 that need the upper bound. Subtract offset from
1734 get_addr_base_and_unit_offset only if positive and subtract it
1735 before division by eltsize rather than after it.
1738 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
1739 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
1740 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
1744 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
1745 expand_binop with *mul_widen_optab, make sure at least one of the
1746 operands doesn't have VOIDmode.
1749 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
1751 PR rtl-optimization/82044
1752 PR tree-optimization/82042
1753 * dse.c (record_store): Check for overflow.
1754 (check_mem_read_rtx): Properly check for overflow if width == -1, call
1755 add_wild_read instead of clear_rhs_from_active_local_stores on
1756 overflow and log it into dump_file.
1758 2017-11-22 Richard Biener <rguenther@suse.de>
1760 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
1761 fake edges to exit when looking for a place to insert.
1762 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
1764 (insert_into_preds_of_block): Commit edge insertion immediately,
1765 assert that doesn't require new BBs.
1766 (fini_pre): Release expressions.
1767 (pass_pre::execute): Shuffle things around a bit, if the fn
1768 is too large do not compute AVAIL either as this is really the
1771 2017-11-22 Richard Biener <rguenther@suse.de>
1773 PR tree-optimization/83089
1774 * tree-if-conv.c (pass_if_conversion::execute): If anything
1775 changed reset SCEV and free the number of iteration estimates.
1777 2017-11-21 Martin Sebor <msebor@redhat.com>
1779 PR tree-optimization/82945
1780 * calls.h (warn_nonstring_bound): Remove unused function.
1782 2017-11-21 Martin Sebor <msebor@redhat.com>
1784 PR tree-optimization/82945
1785 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
1786 * calls.h (maybe_warn_nonstring_arg): Declare new function.
1787 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
1789 (initialize_argument_information): Call maybe_warn_nonstring_arg.
1790 * calls.h (get_attr_nonstring_decl): Declare new function.
1791 * doc/extend.texi (attribute nonstring): Update.
1792 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
1793 get_attr_nonstring_decl and handle it.
1794 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
1795 detection of nul-termination.
1796 (strlen_to_stridx): Change to a pointer.
1797 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
1798 (pass_strlen::execute): Same.
1800 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
1802 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
1803 for the new option -mprefer-vector-width=[none|128|256|512].
1804 * config/i386/i386.c (ix86_target_string): remove old style options
1805 -mprefer-avx256 and make -mprefer-avx128 as alias.
1806 (ix86_option_override_internal): Apply defaults for the
1807 -mprefer-vector-width=[128|256] option.
1808 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
1809 Implement macros to work with -mprefer-vector-width=.
1810 * config/i386/i386.opt: Implemented option
1811 -mprefer-vector-width=[none|128|256|512].
1812 * doc/invoke.texi: Documentation for
1813 -mprefer-vector-width=[none|128|256|512].
1815 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
1817 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
1819 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
1821 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
1823 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
1825 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
1827 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
1828 * expr.c (expand_expr_real_2): Likewise.
1829 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
1830 fold_binary_loc): Likewise.
1831 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
1832 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
1833 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
1834 MINUS_EXPR transformations.
1835 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
1836 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
1837 * tree-inline.c (estimate_operator_cost): Likewise.
1838 * tree-pretty-print.c (dump_generic_node, op_code_prio,
1839 op_symbol_code): Likewise.
1840 * tree-vect-stmts.c (vectorizable_operation): Likewise.
1841 * vr-values.c (extract_range_from_binary_expr): Likewise.
1842 * varasm.c (initializer_constant_valid_p_1): Likewise.
1843 * tree.def: New tree code POINTER_DIFF_EXPR.
1845 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
1847 * config/i386/i386.md (*bswap<mode>2_movbe): Add
1848 integer suffix to movbe mnemonic.
1849 (*bswaphi2_movbe): Ditto.
1850 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
1852 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1855 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
1856 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
1857 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
1859 2017-11-21 Martin Liska <mliska@suse.cz>
1861 * tree-inline.c (expand_call_inline): Remove not needed
1864 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
1865 Jakub Jelinek <jakub@redhat.com>
1868 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
1869 Remove static keyword from f variable.
1871 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1873 PR tree-optimization/83086
1874 * gimple-ssa-store-merging.c
1875 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
1876 rather than n.base_addr.
1878 2017-11-21 Martin Liska <mliska@suse.cz>
1880 PR rtl-optimization/82044
1881 PR tree-optimization/82042
1882 * dse.c (check_mem_read_rtx): Check for overflow.
1884 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1886 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
1889 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1891 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
1892 and its corresponding call arg location note.
1894 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1895 Andrew Burgess <andrew.burgess@embecosm.com>
1897 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
1899 (arc_return_slot_offset): Likewise.
1900 (arc_eh_return_address_location): New declaration.
1901 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
1902 (MUST_SAVE_REGISTER): Add exception handler case.
1903 (MUST_SAVE_RETURN_ADDR): Likewise.
1904 (arc_frame_pointer_required): Likewise.
1905 (arc_frame_pointer_needed): New function.
1906 (arc_compute_frame_size): Changed.
1907 (arc_expand_prologue): Likewise.
1908 (arc_expand_epilogue): Likewise.
1909 (arc_initial_elimination_offset): Likewise.
1910 (arc_return_slot_offset): Delete.
1911 (arc_eh_return_address_location): New function.
1912 (arc_builtin_setjmp_frame_value): Likewise.
1913 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
1914 (EH_RETURN_STACKADJ_RTX): Define.
1915 (EH_RETURN_HANDLER_RTX): Likewise.
1916 * config/arc/arc.md (eh_return): Delete.
1918 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1920 * print-tree.h (debug_vec_tree): Remove prototype.
1921 * gdbinit.in (pvt): Remove macro.
1923 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1925 PR tree-optimization/83047
1926 * gimple-ssa-store-merging.c
1927 (imm_store_chain_info::output_merged_store): If the loads with the
1928 same vuse are in different basic blocks, for load_gsi pick a load
1929 location that is dominated by the other loads.
1932 * config/i386/i386.c (ix86_memmodel_check): Start
1933 -Winvalid-memory-model diagnostics with lowercase letter.
1936 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
1937 set in_cold_section_p to the partition containing loc_list->first.
1938 When seeing loc_list->last_before_switch node, update secname and
1939 perform range_across_switch second partition handling only after that.
1942 * run-rtl-passes.c: Include debug.h.
1943 (run_rtl_passes): Call debug_hooks->assembly_start.
1944 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
1948 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
1949 OPTAB_DIRECT in calls to expand_simple_binop.
1951 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1954 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
1956 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1959 * spellcheck.h (best_match::blithely_get_best_candidate): New
1962 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1964 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
1965 with lower case letter.
1967 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
1969 * config/i386/i386.md (bswaphi2): New expander.
1970 (*bswaphi2_movbe): New insn pattern.
1971 (bswaphi -> rorhi pepehole2): New peephole pattern.
1973 2017-11-20 Jeff Law <law@redhat.com>
1975 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
1976 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
1977 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
1978 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
1979 gimple-ssa-evrp-analyze.h.
1981 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
1982 set BB_VISITED here.
1983 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
1985 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1987 * tree-predcom.c: Add general comment on Store-Store chains.
1988 (split_data_refs_to_components): Postpone clearing eliminate_store_p
1990 (get_chain_last_ref_at): Rename into...
1991 (get_chain_last_write_at): ...this.
1992 (get_chain_last_write_before_load): New function.
1993 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
1994 CT_STORE_STORE when write reference is added.
1995 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
1996 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
1997 (initialize_root_vars_store_elim_1): Ditto.
1998 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
1999 definition is created.
2000 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2001 chain by replacing it with dominant stored value.
2003 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2005 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2007 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2009 * vr-values.c (extract_range_from_binary_expr): Use a full range
2012 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2014 * config/arm/arm.md (R4_REGNUM): Define constant.
2015 (nonsecure_call_internal): Remove r4 clobber.
2016 (nonsecure_call_value_internal): Likewise.
2017 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2018 clobber and resequence match_operands.
2019 (nonsecure_call_value_reg_thumb1_v5): Likewise.
2020 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
2021 (nonsecure_call_value_reg_thumb2): Likewise.
2023 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2025 PR tree-optimization/78821
2026 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
2027 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
2028 to the address of the base rather than the base itself.
2029 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
2030 (find_bswap_or_nop_finalize): New function.
2031 (find_bswap_or_nop): Use it.
2032 (bswap_replace): Return a tree rather than bool, change first
2033 argument from gimple * to gimple_stmt_iterator, allow inserting
2034 into an empty sequence, allow ins_stmt to be NULL - then emit
2035 all stmts into gsi. Fix up MEM_REF address gimplification.
2036 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
2037 (struct store_immediate_info): Add N and INS_STMT non-static
2039 (store_immediate_info::store_immediate_info): Initialize them
2040 from newly added ctor args.
2041 (merged_store_group::apply_stores): Formatting fixes. Sort by
2043 (stmts_may_clobber_ref_p): For stores call also
2044 refs_anti_dependent_p.
2045 (gather_bswap_load_refs): New function.
2046 (imm_store_chain_info::try_coalesce_bswap): New method.
2047 (imm_store_chain_info::coalesce_immediate_stores): Use it.
2048 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
2049 (imm_store_chain_info::output_merged_store): Fail if number of
2050 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
2051 and NOP_EXPR rhs_code.
2052 (pass_store_merging::process_store): Compute n and ins_stmt, if
2053 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
2054 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
2056 (pass_store_merging::execute): Calculate dominators.
2058 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
2059 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
2060 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
2061 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2062 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
2063 class pass_optimize_bswap, bswap_replace,
2064 pass_optimize_bswap::execute): Moved to ...
2065 * gimple-ssa-store-merging.c: ... this file.
2066 Include optabs-tree.h.
2067 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
2068 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2069 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
2070 anonymous namespace, remove static keywords.
2071 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
2072 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
2074 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
2077 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
2079 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
2081 * vec.h (debug_helper): New function.
2082 (DEFINE_DEBUG_VEC): New macro.
2083 * hash-set.h (debug_helper): New function.
2084 (DEFINE_DEBUG_HASH_SET): New macro.
2085 * cfg.c (debug_slim (edge)): New function.
2086 Call DEFINE_DEBUG_VEC for edges.
2087 Call DEFINE_DEBUG_HASH_SET for edges.
2088 * cfghooks.c (debug_slim (basic_block)): New function.
2089 Call DEFINE_DEBUG_VEC for basic blocks.
2090 Call DEFINE_DEBUG_HASH_SET for basic blocks.
2091 * print-tree.c (debug_slim): New function to handle trees.
2092 Call DEFINE_DEBUG_VEC for trees.
2093 Call DEFINE_DEBUG_HASH_SET for trees.
2094 (debug (vec<tree, va_gc>) &): Remove.
2095 (debug (<vec<tree, va_gc>) *): Remove.
2096 * print-rtl.c (debug_slim): New function to handle const_rtx.
2097 Call DEFINE_DEBUG_VEC for rtx_def.
2098 Call DEFINE_DEBUG_VEC for rtx_insn.
2099 Call DEFINE_DEBUG_HASH_SET for rtx_def.
2100 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
2101 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
2102 (debug (vec<rtx_insn *> *ptr): Remove.
2103 (debug_insn_vector): Remove.
2104 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
2106 2017-11-20 Tom de Vries <tom@codesourcery.com>
2108 PR rtl-optimization/82020
2109 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
2110 IF_THEN_ELSE condition.
2112 2017-11-19 Jeff Law <law@redhat.com>
2114 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2115 of degenerates resulting from ignoring an edge.
2117 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2120 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
2122 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2125 * profile-count.c (profile_count::to_sreal_scale): Fix return value
2126 for uninitialied counts.
2128 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2131 * tree-inline.c (estimate_num_insns): Set to 1 at least.
2133 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2136 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
2138 2017-11-19 Tom de Vries <tom@codesourcery.com>
2140 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
2143 2017-11-19 Tom de Vries <tom@codesourcery.com>
2145 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
2148 2017-11-19 Tom de Vries <tom@codesourcery.com>
2150 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2153 2017-11-19 Tom de Vries <tom@codesourcery.com>
2155 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
2157 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
2159 2017-11-19 Tom de Vries <tom@codesourcery.com>
2161 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
2163 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
2164 semicolon after MCORE_EXPORT_NAME call.
2166 2017-11-19 Tom de Vries <tom@codesourcery.com>
2169 * vmsdbgout.c (vmsdbgout_early_finish): New function.
2170 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
2172 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
2174 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
2175 (cgraph_edge::clone): Cleanup updating of profile.
2176 * ipa-cp.c (update_profiling_info): Likewise.
2177 * ipa-inline-transform.c (inline_transform): Likewise.
2178 * ipa-inline.c (inline_small_functions): Add missing space to dump.
2179 * ipa-split.c (execute_split_functions): Do not split when function
2181 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
2182 * profile-count.c (profile_count::dump): Add global0.
2183 (profile_count::to_cgraph_frequency): Do not ICE when entry is
2185 (profile_count::to_sreal_scale): Likewise.
2186 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
2187 (profile_count::combine_with_ipa_count): New function.
2188 * profile-count.h (profile_guessed_global0adjusted): New.
2189 (profile_count::adjusted_zero): New.
2190 (profile_count::global0adjusted): New.
2191 (profile_count::combine_with_ipa_count): New.
2192 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
2193 correct profile of return block of split functions.
2194 (copy_cfg_body): Remove unused profile_count.
2195 (copy_body): Likewise.
2196 (expand_call_inline): Update.
2197 (tree_function_versioning): Update.
2199 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
2201 * hash-set.h (hash_set::empty): New.
2202 * tree-ssa-threadbackward.h: Delete.
2203 * tree-ssa-threadbackward.c (class thread_jumps): New.
2204 Move max_threaded_paths into class.
2205 (fsm_find_thread_path): Remove arguments that are now in class.
2206 (profitable_jump_thread_path): Rename to...
2207 (thread_jumps::profitable_jump_thread_path): ...this.
2208 (convert_and_register_jump_thread_path): Rename to...
2209 (thread_jumps::convert_and_register_current_path): ...this.
2210 (check_subpath_and_update_thread_path): Rename to...
2211 (thread_jumps::check_subpath_and_update_thread_path): ...this.
2212 (register_jump_thread_path_if_profitable): Rename to...
2213 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
2214 (handle_phi): Rename to...
2215 (thread_jumps::handle_phi): ...this.
2216 (handle_assignment): Rename to...
2217 (thread_jumps::handle_assignment): ...this.
2218 (fsm_find_control_statement_thread_paths): Rename to...
2219 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
2220 (find_jump_threads_backwards): Rename to...
2221 (thread_jumps::find_jump_threads_backwards): ...this.
2222 Initialize path local data.
2223 (pass_thread_jumps::execute): Call find_jump_threads_backwards
2224 from within thread_jumps class.
2225 (pass_early_thread_jumps::execute): Same.
2227 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2229 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
2231 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2233 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
2234 to canonical location.
2236 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2239 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
2241 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
2243 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
2245 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
2246 rcrt1.o%s/grcrt1.o%s for -static-pie.
2248 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2250 * i386.c (ix86_multiplication_cost, ix86_division_cost,
2251 ix86_shift_rotate_cost): Break out from ...
2252 (ix86_rtx_costs): ... here.
2253 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
2256 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2258 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
2259 when entry block was promoted unlikely.
2260 (estimate_bb_frequencies): Increase frequency scale.
2261 * profile-count.h (profile_count): Export precision info.
2263 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2265 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
2266 disturbin profile of entry block.
2268 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2270 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
2273 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2275 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
2276 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
2278 2017-11-17 Jeff Law <law@redhat.com>
2280 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
2281 from evrp_dom_walker class. Various methods moved into new class.
2282 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
2283 (evrp_range_analyzer::enter): New method.
2284 (evrp_range_analyzer::leave): New method.
2285 (evrp_dom_walker): Remove delegators no longer needed by this class.
2286 Replace vr_values data member with evrp_range_analyzer
2288 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
2289 method extracted from evrp_dom_walker::before_dom_children.
2290 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
2291 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
2293 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
2294 Add private copy constructor and move assignment operators.
2295 Privatize methods and class data where trivially possible.
2296 (evrp_dom_walker::cleanup): New function, extracted from
2297 execute_early_vrp. Simplify access to class data.
2299 * vr-values.h (get_output_for_vrp): Prototype.
2300 * vr-values.c (get_output_for_vrp): New function extracted from
2301 vrp_visit_assignment_or_call and extract_range_from_stmt.
2302 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
2304 2017-11-17 Luis Machado <luis.machado@linaro.org>
2306 * config/aarch64/aarch64.c
2307 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
2308 (qdf24xx_tunings) <autoprefetcher_model>: Set to
2309 tune_params::AUTOPREFETCHER_WEAK.
2311 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2314 * config/arm/arm.c (arm_valid_target_attribute_rec):
2315 Parse "arch=" and "+<ext>".
2316 (arm_valid_target_attribute_tree): Re-init global options.
2317 (arm_option_override): Make non-static.
2318 (arm_options_perform_arch_sanity_checks): Make errors fatal.
2319 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
2320 (__ARM_FEATURE_CRC32): Support undef.
2321 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
2322 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
2324 2017-11-17 David Malcolm <dmalcolm@redhat.com>
2326 * gdbinit.in (break-on-diagnostic): New command.
2328 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2330 * config/i386/i386.c (ix86_expand_epilogue): Change simple
2331 return to indirect jump for EH return if control-flow
2332 protection is enabled. Change explicit 'false' argument in
2333 pro_epilogue_adjust_stack with a value of
2335 * config/i386/i386.md (simple_return_indirect_internal):
2336 Remove SImode restriction to support 64-bit.
2338 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2340 * combine.c (added_notes_insn): New.
2341 (try_combine): Handle added_notes_insn like added_links_insn.
2342 Rewrite return value code.
2343 (distribute_notes): Set added_notes_insn to the earliest insn we added
2346 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2348 PR rtl-optimization/82621
2349 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
2350 dest of one of those SETs is unused.
2352 2017-11-17 Richard Biener <rguenther@suse.de>
2355 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
2356 * tree-pretty-print.c (dump_generic_node): Handle
2357 annot_expr_parallel_kind.
2358 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
2359 * gimplify.c (gimple_boolify): Likewise.
2361 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2363 * config.gcc (extra_headers): Add cet.h for x86 targets.
2364 * config/i386/cet.h: New file.
2365 * doc/install.texi: Add --enable-cet/--disable-cet.
2367 2017-11-17 Richard Biener <rguenther@suse.de>
2369 PR tree-optimization/83017
2370 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
2371 (gen_parallel_loop): Properly count iterations.
2372 (parallelize_loops): Handle loop->can_be_parallel independent
2373 of flag_loop_parallelize_all. Make static profitability test match
2375 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
2376 * invoke.texi (parloops-min-per-thread): Document.
2378 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
2380 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
2381 upstreaming review comments.
2383 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2385 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
2386 with fast unaligned access.
2387 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
2389 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
2391 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
2392 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
2394 2017-11-17 Richard Biener <rguenther@suse.de>
2396 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
2397 folding of references.
2399 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
2402 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
2403 of replacing call to strncmp with corresponding call to strcmp when
2406 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
2408 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
2409 option prefer-avx256 for skylake-avx512 configuration.
2410 * config/i386/i386.c (ix86_option_override_internal): Ditto.
2411 (get_builtin_code_for_version): Ditto.
2413 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2414 Monk Chiang <sh.chiang04@gmail.com>
2416 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
2417 (FIXED_REGISTERS): Reserve more register numbers.
2418 (CALL_USED_REGISTERS): Likewise.
2419 (REG_ALLOC_ORDER): Likewise.
2420 (REG_CLASS_CONTENTS): Likewise.
2421 (REGISTER_NAMES): Likewise.
2423 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2424 Kito Cheng <kito.cheng@gmail.com>
2426 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
2428 * config/nds32/iterators.md: Add vector mode iterators and attributes.
2430 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
2432 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
2433 parameter list for vec_splats.
2435 2017-11-16 Joseph Myers <joseph@codesourcery.com>
2437 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
2439 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
2441 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
2444 * doc/install.texi: Document incompatibility of Perl >=5.6.26
2445 with the required version of automake 1.11.6.
2447 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
2449 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
2451 (DU_C2_3_power9): Correct reservation combinations.
2452 (FP_DIV_power9, VEC_DIV_power9): New.
2453 (power9-alu): Split out rotate/shift...
2454 (power9-rot): ...to here, correct dispatch resource.
2455 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
2457 (power9-fp): Correct latency.
2458 (power9-sdiv): Add div/sqrt resource.
2459 (power9-ddiv): Correct latency, add div/sqrt resource.
2460 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
2461 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
2463 (power9-qpdiv, power9-qpmul): Adjust resource usage.
2465 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2467 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
2468 switch statement mapping KF built-ins to TF built-ins if we don't
2469 have the proper ISA 3.0 assembler support.
2471 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2473 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
2474 (lower_emutls_function_body): Do not compute it.
2476 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2478 * ipa-split.c (split_bb_info): Turn time to sreal.
2479 (split_point): Likewise.
2480 (dump_split_point): Likewise.
2481 (fine_split_points): Likewise.
2482 (execute_split_functions): Only zero split_bbs; turn time to sreals.
2484 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2486 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
2489 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2491 * predict.c (combine_predictions_for_bb): Preserve zero predicted
2493 (expensive_function_p): Remove useless assert.
2494 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
2496 2017-11-16 Martin Sebor <msebor@redhat.com>
2498 PR tree-optimization/82588
2499 PR tree-optimization/82583
2500 * tree-vrp.c (check_array_ref): Handle flexible array members,
2501 string literals, and inner indices.
2502 (search_for_addr_array): Add detail to diagnostics.
2504 2017-11-16 Nathan Sidwell <nathan@acm.org>
2508 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
2509 (SET_DECL_ASSEMBLER_NAME): Forward to
2510 overwrite_decl_assembler_name.
2511 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
2512 (overwrite_decl_assembler_name): Declare.
2513 * tree.c (overwrite_decl_assembler_name): New.
2514 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
2515 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
2516 (LANG_HOOKS_INITIALIZER): Add it.
2517 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
2518 * langhooks.c (lhd_set_decl_assembler_name): Use
2519 SET_DECL_ASSEMBLER_NAME.
2520 (lhd_overwrite_decl_assembler_name): Default implementation.
2522 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2523 Jackson Woodruff <jackson.woodruff@arm.com>
2525 PR tree-optimization/71026
2526 * match.pd: Canonicalize constant multiplies in division.
2528 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2530 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
2531 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
2532 Remove OPT_fomit_frame_pointer entry.
2533 * common/config/arc/arc-common.c: Likewise.
2534 * common/config/arm/arm-common.c: Likewise.
2535 * common/config/avr/avr-common.c: Likewise.
2536 * common/config/c6x/c6x-common.c: Likewise.
2537 * common/config/cr16/cr16-common.c: Likewise.
2538 * common/config/cris/cris-common.c: Likewise.
2539 * common/config/epiphany/epiphany-common.c: Likewise.
2540 * common/config/fr30/fr30-common.c: Likewise.
2541 * common/config/frv/frv-common.c: Likewise.
2542 * common/config/ia64/ia64-common.c: Likewise.
2543 * common/config/iq2000/iq2000-common.c: Likewise.
2544 * common/config/lm32/lm32-common.c: Likewise.
2545 * common/config/m32r/m32r-common.c: Likewise.
2546 * common/config/mcore/mcore-common.c: Likewise.
2547 * common/config/microblaze/microblaze-common.c: Likewise.
2548 * common/config/mips/mips-common.c: Likewise.
2549 * common/config/mmix/mmix-common.c: Likewise.
2550 * common/config/mn10300/mn10300-common.c: Likewise.
2551 * common/config/nios2/nios2-common.c: Likewise.
2552 * common/config/pa/pa-common.c: Likewise.
2553 * common/config/pdp11/pdp11-common.c: Likewise.
2554 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
2555 * common/config/riscv/riscv-common.c: Likewise.
2556 * common/config/rs6000/rs6000-common.c: Likewise.
2557 * common/config/rx/rx-common.c: Likewise.
2558 * common/config/s390/s390-common.c: Likewise.
2559 * common/config/sh/sh-common.c: Likewise.
2560 * common/config/sparc/sparc-common.c: Likewise.
2561 * common/config/tilegx/tilegx-common.c: Likewise.
2562 * common/config/tilepro/tilepro-common.c: Likewise.
2563 * common/config/v850/v850-common.c: Likewise.
2564 * common/config/visium/visium-common.c: Likewise.
2565 * common/config/xstormy16/xstormy16-common.c: Likewise.
2566 * common/config/xtensa/xtensa-common.c: Likewise.
2567 * invoke.texi (-fomit-frame-pointer): Update documentation.
2569 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2571 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
2573 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2575 * cfg.c (scale_bbs_frequencies_int,
2576 cale_bbs_frequencies_gcov_type): Remove.
2577 * cfg.h (scale_bbs_frequencies_int,
2578 cale_bbs_frequencies_gcov_type): Remove.
2580 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2582 * tree-ssa-loop-manip.c
2583 (scale_dominated_blocks_in_loop): Update to profile counts.
2584 (tree_transform_and_unroll_loop): Likewise.
2586 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2588 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
2589 scale_bbs_frequencies_int.
2591 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2593 * final.c (compute_alignments): Use counts rather than frequencies.
2595 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2597 * cfgloopanal.c: Include sreal.h
2598 (average_num_loop_insns): Use counts and sreal for accounting.
2600 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2602 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
2605 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2606 Kito Cheng <kito.cheng@gmail.com>
2608 * config/nds32/constraints.md: Provide more constraints.
2609 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
2610 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
2611 support constraints usage.
2613 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2615 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
2617 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2618 Kito Cheng <kito.cheng@gmail.com>
2620 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
2621 * config/nds32/nds32.opt: Refine the layout.
2622 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
2623 TARGET_EXT_STRING): Support new options.
2624 * config/nds32/nds32.h: Likewise.
2625 * config/nds32/nds32.md: Likewise.
2626 * config/nds32/nds32-predicates.c: Likewise.
2627 * config/nds32/constraints.md: Likewise.
2628 * common/config/nds32/nds32-common.c: Likewise.
2630 2017-11-16 Julia Koval <julia.koval@intel.com>
2633 * config/i386/gfniintrin.h: Add sse check.
2634 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
2636 2017-11-16 Julia Koval <julia.koval@intel.com>
2638 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
2639 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
2640 (ix86_handle_option): Handle -mavx512vbmi2.
2641 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
2642 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2643 * config/i386/i386-c.c (__AVX512VBMI2__): New.
2644 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
2645 (ix86_valid_target_attribute_inner_p): Ditto.
2646 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
2647 * config/i386/i386.opt (mavx512vbmi2): New option.
2648 * doc/invoke.texi: Add new option.
2650 2017-11-16 Julia Koval <julia.koval@intel.com>
2652 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
2653 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
2654 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
2655 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
2656 _mm512_gf2p8mul_epi8): New intrinsics.
2657 * config/i386/i386-builtin-types.def
2658 (V64QI_FTYPE_V64QI_V64QI): New type.
2659 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
2660 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
2661 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
2662 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
2663 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
2664 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2666 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
2668 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
2669 explicitly as a stream of bytes.
2671 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2673 * config/rs6000/altivec.h (vec_xst_be): New #define.
2674 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
2675 and externalize from *altivec_vperm_<mode>_internal.
2676 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
2678 (XL_BE_V8HI): Likewise.
2679 (XL_BE_V4SI): Likewise.
2680 (XL_BE_V4SI): Likewise.
2681 (XL_BE_V2DI): Likewise.
2682 (XL_BE_V4SF): Likewise.
2683 (XL_BE_V2DF): Likewise.
2684 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
2685 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
2686 all array entries with these keys: VSX_BUILTIN_VEC_XL,
2687 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
2688 VSX_BUILTIN_VEC_XST_BE.
2689 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
2690 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
2692 (altivec_init_builtins): Replace conditional calls to def_builtin
2693 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
2694 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
2695 with unconditional calls. Remove calls to def_builtin for
2696 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
2697 __builtin_vec_xst_be.
2698 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
2699 to define_expand, and add alternate RTL generation for P8.
2700 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
2701 vsx_ld_elemrev_v8hi.
2702 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
2703 add alternate RTL generation for P8.
2704 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
2705 vsx_ld_elemrev_v16qi.
2706 (vsx_st_elemrev_v8hi): Convert define_insn
2707 to define_expand, and add alternate RTL generation for P8.
2708 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
2709 vsx_st_elemrev_v8hi.
2710 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
2711 add alternate RTL generation for P8.
2712 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
2713 vsx_st_elemrev_v16qi.
2715 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
2718 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
2719 TARGET_AVX512ER check.
2720 (ix86_option_override_internal): Set MASK_VZEROUPPER if
2721 neither -mzeroupper nor -mno-zeroupper is used and
2722 TARGET_EMIT_VZEROUPPER is set.
2723 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
2724 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
2726 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2728 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2729 folding of vector compares.
2730 (fold_build_vec_cmp): New helper function.
2731 (fold_compare_helper): New helper function.
2732 (builtin_function_type): Add compare builtins to the list of functions
2733 having unsigned arguments. Cosmetic updates to comment indentation.
2734 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
2735 the not+eq combination.
2737 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2739 PR tree-optimization/82726
2740 PR tree-optimization/70754
2741 * tree-predcom.c (order_drefs_by_pos): New function.
2742 (combine_chains): Move code setting has_max_use_after to...
2743 (try_combine_chains): ...here. New parameter. Sort combined chains
2744 according to position information.
2745 (tree_predictive_commoning_loop): Update call to above function.
2746 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
2748 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2750 PR tree-optimization/82726
2752 2017-01-23 Bin Cheng <bin.cheng@arm.com>
2754 PR tree-optimization/70754
2755 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
2756 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
2757 combined stmt before it if not NULL.
2758 (combine_chains): Process refs reversely and compute dominance point
2762 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2764 PR tree-optimization/79663
2765 * tree-predcom.c (combine_chains): Process refs in reverse order
2766 only for ZERO length chains, and add explaining comment.
2768 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2770 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
2771 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
2772 * doc/invoke.texi (ARM Options): Add armv8.3-a.
2774 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2776 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
2778 2017-11-15 Martin Liska <mliska@suse.cz>
2780 * tree-cfg.c (pass_warn_function_return::execute):
2781 Compare warn_return_type for greater than zero.
2783 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
2787 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
2788 to return true on Xeon and not on Xeon Phi.
2789 (ix86_check_avx256_register): Changed to ...
2790 (ix86_check_avx_upper_register): ... this. Add extra check for
2791 VALID_AVX512F_REG_OR_XI_MODE.
2792 (ix86_avx_u128_mode_needed): Changed
2793 ix86_check_avx256_register to ix86_check_avx_upper_register.
2794 (ix86_check_avx256_stores): Changed to ...
2795 (ix86_check_avx_upper_stores): ... this. Changed
2796 ix86_check_avx256_register to ix86_check_avx_upper_register.
2797 (ix86_avx_u128_mode_after): Changed
2798 avx_reg256_found to avx_upper_reg_found. Changed
2799 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
2800 (ix86_avx_u128_mode_entry): Changed
2801 ix86_check_avx256_register to ix86_check_avx_upper_register.
2802 (ix86_avx_u128_mode_exit): Ditto.
2803 * config/i386/i386.h: (host_detect_local_cpu): New define.
2805 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2807 * config/arm/xgene1.md (xgene1): Split into automatons
2808 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
2809 (xgene1_f_load): Adjust reservations and/or types.
2810 (xgene1_f_store): Likewise.
2811 (xgene1_load_pair): Likewise.
2812 (xgene1_store_pair): Likewise.
2813 (xgene1_fp_load1): Likewise.
2814 (xgene1_load1): Likewise.
2815 (xgene1_store1): Likewise.
2816 (xgene1_move): Likewise.
2817 (xgene1_alu): Likewise.
2818 (xgene1_simd): Likewise.
2819 (xgene1_bfm): Likewise.
2820 (xgene1_neon_load1): Likewise.
2821 (xgene1_neon_store1): Likewise.
2822 (xgene1_neon_logic): Likewise.
2823 (xgene1_neon_st1): Likewise.
2824 (xgene1_neon_ld1r): Likewise.
2825 (xgene1_alu_cond): Added.
2826 (xgene1_shift_reg): Likwise.
2827 (xgene1_bfx): Likewise.
2828 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
2830 2017-11-15 Jakub Jelinek <jakub@redhat.com>
2833 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
2835 (can_widen_mult_without_libcall): New function.
2836 (expand_mul_overflow): If only checking unsigned mul overflow,
2837 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
2838 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
2839 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
2840 (expand_DIVMOD): Formatting fix.
2841 * expmed.h (expand_mult): Add NO_LIBCALL argument.
2842 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
2843 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
2845 PR tree-optimization/82977
2846 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
2847 constructed temporary to strlen_to_stridx.put.
2849 2017-11-15 Martin Liska <mliska@suse.cz>
2851 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
2852 * configure: Regenerate.
2854 2017-11-15 Martin Liska <mliska@suse.cz>
2857 * config/sh/sh-mem.cc: Use proper probability for
2860 2017-11-14 Jeff Law <law@redhat.com>
2862 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
2863 the red zone for stack_clash_protection_final_dynamic_probe targets
2864 when the total dynamic stack size is zero bytes.
2866 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
2867 blocks is post order.
2869 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
2871 * dumpfile.h (TDF_COMPARE_DEBUG): New.
2872 * final.c (rest_of_clean_state): Set it for the
2873 -fcompare-debug dump.
2874 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
2875 class when TDF_COMPARE_DEBUG is set.
2877 * dwarf2out.c (gen_producer_string): Discard
2880 2017-11-15 Joseph Myers <joseph@codesourcery.com>
2883 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
2884 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
2885 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
2886 (__TGMATH_REAL_2_3): Remove macros.
2887 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
2888 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
2890 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
2892 (remquo): Define using __TGMATH_REAL_3.
2894 2017-11-14 Jeff Law <law@redhat.com>
2896 * vr-values.c: New file with contents extracted from tree-vrp.c.
2897 * Makefile.in (OBJS): Add vr-values.o
2898 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
2899 (set_value_range, set_and_canonicalize_value_range): Likewise.
2900 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
2901 (value_range_constant_singleton, symbolic_range_p): Likewise.
2902 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
2903 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
2904 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
2905 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
2906 (find_case_label_range, find_case_label_index): Likewise.
2907 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
2908 (range_int_cst_singleton_p, value_inside_range): Likewise.
2909 (get_single_symbol): Likewise.
2910 (switch_update): Move structure definition here.
2911 (to_remove_edges, to_update_switch_stmts): Provide externs.
2912 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
2913 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
2914 (vrp_val_is_min, set_value_range): Likewise.
2915 (set_and_canonicalize_value_range, copy_value_range): Likewise.
2916 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
2917 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
2918 (range_is_nonnull, range_int_cst_p): Likewwise.
2919 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
2920 (get_single_symbol, operand_less_p): Likewise
2921 (compare_values_warnv, compare_values): Likewise.
2922 (value_inside_range, value_range_constant_singleton): Likewise.
2923 (zero_nonzero_bitgs_from_vr): Likewise.
2924 (extract_range_from_binary_expr_1): Likewise.
2925 (overflow_comparison_p): Likewise.
2926 (to_remove_edges, to_update_switch_stmts): Likewise.
2927 (find_case_label-index, find_case_label_range): Likewise.
2928 (switch_update, set_value_range_to_nonnegative): Remove.
2929 (set_value_range_to_truthvalue): Likewise.
2930 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
2931 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
2932 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
2933 (find_case_label_ranges, test_for_singularity): Likewise.
2934 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
2935 (x_vr_values): Move to its remaining use site.
2937 2017-11-10 Jeff Law <law@redhat.com>
2939 * vr-values.h (VR_INITIALIZER): Move #define here.
2940 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
2941 * Makefile.in (OBJS): Add tree-evrp.o
2942 * tree-vrp.h (assert_info): Move structure definition here.
2943 (set_value_range_to_varying): Prototype.
2944 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2945 (infer_value_range, register_edge_assert_for): Likewise.
2946 (stmt_interesting_for_vrp): Likewise.
2947 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
2948 (set_value_range_to_varying): No longer static.
2949 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2950 (infer_value_range, register_edge_assert_for): Likewise.
2952 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2954 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
2955 generate the XXBRD instruction.
2957 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
2958 (rs6000_builtin_type_compatible): Treat _Float128 and long double
2959 as being compatible if -mabi=ieeelongdouble.
2960 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
2961 to setup float128 built-ins with hardware support.
2962 (BU_FLOAT128_HW_2): Likewise.
2963 (BU_FLOAT128_HW_3): Likewise.
2964 (BU_FLOAT128_HW_VSX_1): Likewise.
2965 (BU_FLOAT128_HW_VSX_2): Likewise.
2966 (scalar_extract_expq): Change float128 built-in functions to
2967 accommodate having both KFmode and TFmode functions. Use the
2968 KFmode variant as the default.
2969 (scalar_extract_sigq): Likewise.
2970 (scalar_test_neg_qp): Likewise.
2971 (scalar_insert_exp_q): Likewise.
2972 (scalar_insert_exp_qp): Likewise.
2973 (scalar_test_data_class_qp): Likewise.
2974 (sqrtf128_round_to_odd): Delete processing the round to odd
2975 built-in functions as special built-in functions, and define them
2976 as float128 built-ins. Use the KFmode variant as the default.
2977 (truncf128_round_to_odd): Likewise.
2978 (addf128_round_to_odd): Likewise.
2979 (subf128_round_to_odd): Likewise.
2980 (mulf128_round_to_odd): Likewise.
2981 (divf128_round_to_odd): Likewise.
2982 (fmaf128_round_to_odd): Likewise.
2983 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
2984 support for KFmode and TFmode xststdcqp calls.
2985 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
2986 point, switch the built-in handlers for the get/set float128
2987 exponent, get float128 mantissa, float128 test built-ins, and the
2988 float128 round to odd built-in functions. Eliminate creating the
2989 float128 round to odd built-in functions as special built-ins.
2990 (rs6000_init_builtins): Eliminate special creation of the float128
2991 round to odd built-in functions.
2992 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
2993 function insns to support both TFmode and KFmode variants.
2994 (xsxsigqp_<mode>): Likewise.
2995 (xsiexpqpf_<mode>): Likewise.
2996 (xsiexpqp_<mode>): Likewise.
2997 (xststdcqp_<mode>): Likewise.
2998 (xststdcnegqp_<mode>): Likewise.
2999 (xststdcqp_<mode>): Likewise.
3001 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3003 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3004 unused path_in_freq_ptr parameter.
3005 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3007 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3009 * ipa-inline.c (edge_badness): Dump sreal frequency.
3010 (compute_inlined_call_time): Match natural implementaiton ...
3011 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3012 forgotten division by CGRAPH_FREQ_BASE.
3014 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3016 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3017 Solaris 11. Update comment.
3018 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
3020 * config/sol2.h (STARTFILE_SPEC): Likewise.
3021 * configure: Regenerate.
3023 2017-11-14 Carl Love <cel@us.ibm.com>
3025 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
3026 le_ and be_ prefixes to swap* variables. Remove
3027 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
3030 2017-11-14 Jason Merrill <jason@redhat.com>
3032 Support GTY((cache)) on hash_map.
3033 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
3034 (ggc_cache_remove): Override it instead of ggc_mx.
3035 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
3036 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
3037 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
3038 (simple_cache_map_traits): Override maybe_mx.
3039 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
3040 (hash_map): Friend gt_cleare_cache.
3041 (gt_cleare_cache): New.
3042 * tree.h (tree_cache_traits): New hash_map traits class.
3043 (tree_cache_map): New typedef.
3045 2017-11-14 Richard Biener <rguenther@suse.de>
3047 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
3048 paramter and handling.
3049 (cleanup_control_flow_bb): Likewise.
3050 (cleanup_control_flow_pre): New helper performing a DFS walk
3051 to call cleanup_control_flow_bb in PRE order.
3052 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
3053 via cleanup_control_flow_pre.
3055 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
3057 * config/aarch64/aarch64-simd.md
3058 (aarch64_simd_bsl<mode>_internal): Remove DImode.
3059 (*aarch64_simd_bsl<mode>_alt): Likewise.
3060 (aarch64_simd_bsldi_internal): New.
3061 (aarch64_simd_bsldi_alt): Likewise.
3063 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3065 * tracer.c (better_p): Do not compare frequencies.
3066 * reg-stack.c (better_edge): Likewise.
3067 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
3070 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3072 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
3073 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
3074 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
3075 * lto-streamer-in.c (input_function): Use update_max_bb_count.
3076 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
3077 * predict.c (maybe_hot_frequency_p): Inline to ...
3078 (maybe_hot_count_p): ... here; rewrite to counts.
3079 (counts_to_freqs): Rename to ...
3080 (update_max_bb_count): ... this one.
3081 (expensive_function_p): Use counts.
3082 (estimate_bb_frequencies): Update.
3083 (rebuild_frequencies): Update.
3084 * predict.h (counts_to_freqs): Rename to ...
3085 (update_max_bb_count): ... this one.
3086 * profile.c (compute_branch_probabilities): Add debug info
3087 * tree-inline.c (expand_call_inline): Update debug info.
3088 (optimize_inline_calls): Use update_max_bb_count..
3089 (tree_function_versioning): Use update_max_bb_count..
3090 * value-prof.c (gimple_value_profile_transformations):
3091 Do not use update_max_bb_count.
3093 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3095 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
3096 always use frequencies.
3098 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3100 * bb-reorder.c: Remove frequencies from comments.
3101 (better_edge_p): Use profile counts.
3102 (find_traces): Dump profile counts.
3103 (rotate_loop): Use profile counts.
3104 (find_traces_1_round): Likewise.
3105 (connect_better_edge_p): Use counts instead of probabilities for
3107 (copy_bb_p): Drop early check for non-0 frequency.
3108 (sanitize_hot_paths): Update comments.
3110 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3112 * ipa-split.c (struct split_point): Add count.
3113 (consider_split): Do not compute incoming frequency; compute incoming
3114 count and store it to split_point.
3115 (split_function): Set count of the call to split part correctly.
3117 2017-11-13 Carl Love <cel@us.ibm.com>
3119 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
3121 2017-11-13 Tom Tromey <tom@tromey.com>
3123 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
3125 2017-11-13 Carl Love <cel@us.ibm.com>
3127 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3128 Add support for builtins:
3129 unsigned int vec_first_{,miss}_match_{,or_eos}index,
3130 vector {un,}signed {char,int,short},
3131 vector {un,}signed {char,int,short}) arguments.
3132 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
3133 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
3134 Add BU_P9V_AV_2 expansions for the builtins.
3135 * config/rs6000/altivec.h (vec_first_match_index,
3136 vec_first_mismatch_index, vec_first_match_or_eos_index,
3137 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
3138 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
3139 new extern declaration.
3140 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
3141 * config/rs6000/vsx.md (first_match_index_<mode>,
3142 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
3143 first_mismatch_or_eos_index_<mode>): Add define expand.
3144 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
3145 * doc/extend.texi: Update the built-in documenation file for the new
3148 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
3150 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
3151 into the min/max operations for _Float<N> and _Float<N>X types.
3153 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3156 * dwarf2out.c (do_eh_frame): New static variable.
3157 (dwarf2out_begin_prologue): Set it.
3158 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
3160 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3162 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
3164 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
3166 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3169 * debug.h (dwarf2out_do_eh_frame): Declare.
3170 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
3171 (dwarf2out_do_frame): Use it.
3172 (dwarf2out_do_cfi_asm): Likewise.
3173 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
3174 (dwarf2out_assembly_start): Likewise.
3175 (dwarf2out_begin_prologue): Fix comment.
3176 * toplev.c (compile_file): Always call dwarf2out_frame_finish
3177 if the target needs either debug or unwind DWARF2 info.
3178 * lto-opts.c (lto_write_options): Do not save -fexceptions,
3179 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
3180 -ftrapping-math, -ftrapv and -fwrapv.
3182 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3184 * cgraph.c (cgraph_edge::sreal_frequency): New function.
3185 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
3186 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
3187 (estimate_edge_size_and_time): Likewise.
3188 (ipa_merge_fn_summary_after_inlining): Likewise.
3189 * ipa-inline.c (cgraph_freq_base_rec): Remove.
3190 (compute_uninlined_call_time): Use sreal_frequency.
3191 (compute_inlined_call_time): Likewise.
3192 (ipa_inline): Do not initialize cgraph_freq_base_rec.
3193 * profile-count.c: Include sreal.h.
3194 (profile_count::to_sreal_scale): New.
3195 * profile-count.h: Forward declare sreal.
3196 (profile_count::to_sreal_scale): Declare.
3198 2017-11-13 Nathan Sidwell <nathan@acm.org>
3200 * diagnostic.c (maybe_line_and_column): New.
3201 (diagnostic_get_location_text): Use it.
3202 (diagnostic_report_current_module): Likewise.
3203 (test_diagnostic_get_location_text): Add tests.
3205 2017-11-13 Luis Machado <luis.machado@linaro.org>
3207 * doc/md.texi (Specifying processor pipeline description): Fix
3208 incorrect latency for the div instruction example.
3210 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3212 PR tree-optimization/78821
3213 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
3214 that bit_not_p is the same.
3215 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
3216 (split_group): Count precisely bit_not_p bits in each statement.
3217 (invert_op): New function.
3218 (imm_store_chain_info::output_merged_store): Use invert_op to
3219 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
3220 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
3222 2017-11-13 Martin Liska <mliska@suse.cz>
3224 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
3225 (struct source_info): Likewise.
3226 (add_branch_counts): Likewise.
3227 (add_line_counts): Likewise.
3228 (function_summary): Likewise.
3229 (output_intermediate_line): Likewise.
3230 (generate_results): Likewise.
3232 2017-11-13 Martin Liska <mliska@suse.cz>
3234 * gcov.c (struct block_info): Remove typedef for block_t.
3235 (struct line_info): Likewise.
3236 (line_info::has_block): Likewise.
3237 (EXIT_BLOCK): Likewise.
3238 (unblock): Likewise.
3239 (circuit): Likewise.
3240 (get_cycles_count): Likewise.
3241 (process_file): Likewise.
3242 (read_graph_file): Likewise.
3243 (solve_flow_graph): Likewise.
3244 (find_exception_blocks): Likewise.
3245 (add_line_counts): Likewise.
3246 (accumulate_line_info): Likewise.
3247 (output_line_details): Likewise.
3249 2017-11-13 Martin Liska <mliska@suse.cz>
3251 * gcov.c (struct arc_info): Remove typedef for arc_t.
3252 (struct line_info): Likewise.
3253 (add_branch_counts): Likewise.
3254 (output_branch_count): Likewise.
3255 (function_info::~function_info): Likewise.
3256 (circuit): Likewise.
3257 (output_intermediate_line): Likewise.
3258 (read_graph_file): Likewise.
3259 (solve_flow_graph): Likewise.
3260 (find_exception_blocks): Likewise.
3261 (add_line_counts): Likewise.
3262 (accumulate_line_info): Likewise.
3263 (output_line_details): Likewise.
3264 (output_function_details): Likewise.
3266 2017-11-13 Martin Liska <mliska@suse.cz>
3268 * gcov.c (struct function_info): Remove typedef for function_t.
3269 (struct source_info): Likewise.
3270 (source_info::get_functions_at_location): Likewise.
3271 (solve_flow_graph): Likewise.
3272 (find_exception_blocks): Likewise.
3273 (add_line_counts): Likewise.
3274 (output_intermediate_file): Likewise.
3275 (process_file): Likewise.
3276 (generate_results): Likewise.
3277 (release_structures): Likewise.
3278 (read_graph_file): Likewise.
3279 (read_count_file): Likewise.
3280 (accumulate_line_counts): Likewise.
3281 (output_lines): Likewise.
3283 2017-11-13 Martin Liska <mliska@suse.cz>
3285 * gcov.c (function_info::function_info): Remove num_counts
3286 and add vector<gcov_type>.
3287 (function_info::~function_info): Use the vector.
3288 (process_file): Likewise.
3289 (read_graph_file): Likewise.
3290 (read_count_file): Likewise.
3291 (solve_flow_graph): Likewise.
3293 2017-11-13 Martin Liska <mliska@suse.cz>
3295 * gcov.c (function_info::is_artificial): New function.
3296 (process_file): Erase all artificial early.
3297 (generate_results): Skip as all artificial are already
3300 2017-11-13 Martin Liska <mliska@suse.cz>
3302 * gcov.c (read_graph_file): Store to global vector of functions.
3303 (read_count_file): Iterate the vector.
3304 (process_file): Likewise.
3305 (generate_results): Likewise.
3306 (release_structures): Likewise.
3308 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3310 PR tree-optimization/82954
3311 * gimple-ssa-store-merging.c
3312 (imm_store_chain_info::coalesce_immediate_stores): If
3313 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
3315 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
3317 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
3318 Upddate call to ENDIAN_LANE_N.
3319 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
3320 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
3321 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
3322 and use aarch64_endian_lane_rtx.
3323 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
3325 2017-11-12 Tom de Vries <tom@codesourcery.com>
3327 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
3329 2017-11-12 Tom de Vries <tom@codesourcery.com>
3331 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
3333 2017-11-12 Tom de Vries <tom@codesourcery.com>
3335 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
3337 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3338 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3339 * defaults.h (ASM_OUTPUT_LABELREF): Same.
3341 2017-11-11 Martin Sebor <msebor@redhat.com>
3344 * doc/extend.texi (attribute nonstring): Remove spurious argument.
3347 * prefic.c (translate_name): Replace strncpy with memcpy to
3348 avoid -Wstringop-truncation.
3350 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3352 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
3354 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3356 * predict.c (maybe_hot_frequency_p): Do not use cfun.
3358 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3360 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
3363 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
3365 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
3366 enable generating XXBRH if the value is in a vector register.
3367 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
3368 value is in a vector register.
3369 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
3370 register to register bswap64's instead of doing the GPR sequence
3371 used on previous machines.
3372 (bswapdi2_xxbrd): New insn.
3373 (bswapdi2_reg): Disallow on ISA 3.0.
3374 (register to register bswap64 splitter): Do not split the insn on
3375 ISA 3.0 systems that use XXBRD.
3377 2017-11-10 Martin Sebor <msebor@redhat.com>
3380 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
3382 (find_subframework_file): Same.
3384 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3386 * auto-profile.c (afdo_indirect_call): Drop frequency.
3387 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
3388 (cgraph_node::create_edge): Drop frequency argument.
3389 (cgraph_node::create_indirect_edge): Drop frequency argument.
3390 (cgraph_edge::make_speculative): Drop frequency arguments.
3391 (cgraph_edge::resolve_speculation): Do not update frequencies
3392 (cgraph_edge::dump_edge_flags): Do not dump frequency.
3393 (cgraph_node::dump): Check consistency in IPA mode.
3394 (cgraph_edge::maybe_hot_p): Use IPA counter.
3395 (cgraph_edge::verify_count_and_frequency): Rename to ...
3396 (cgraph_edge::verify_count): ... this one; drop frequency checking.
3397 (cgraph_node::verify_node): Update.
3398 * cgraph.h (struct cgraph_edge): Drop frequency.
3399 (cgraph_edge::frequency): New function.
3400 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
3402 (cgraph_edge::rebuild_edges): Likewise.
3403 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
3404 (duplicate_thunk_for_node): Do not pass frequency.
3405 (cgraph_node::create_clone): Scale only counts.
3406 (cgraph_node::create_virtual_clone): Do not pass frequency.
3407 (cgraph_node::create_edge_including_clones): Do not pass frequency.
3408 (cgraph_node::create_version_clone): Do not pass frequency.
3409 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
3410 (cgraph_node::expand_thunk): Do not pass frequency.
3411 (cgraph_node::create_wrapper): Do not pass frequency.
3412 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
3414 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
3415 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
3416 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
3417 (gather_caller_stats): Use frequency function.
3418 (ipcp_cloning_candidate_p): Use frequency function.
3419 (ipcp_propagate_stage): Use frequency function.
3420 (get_info_about_necessary_edges): Use frequency function.
3421 (update_profiling_info): Update only IPA profile.
3422 (update_specialized_profile): Use frequency functoin.
3423 (perhaps_add_new_callers): Update only IPA profile.
3424 * ipa-devirt.c (ipa_devirt): Use IPA profile.
3425 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
3426 (dump_ipa_call_summary): Use frequency function.
3427 (estimate_edge_size_and_time): Use frequency function.
3428 (ipa_merge_fn_summary_after_inlining): Use frequency function.
3429 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
3430 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
3431 (update_noncloned_counts): ... ths one; scale counts only.
3432 (clone_inlined_nodes): Do not scale frequency.
3433 (inline_call): Do not pass frequency.
3434 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
3435 (compute_inlined_call_time): Use IPA profile.
3436 (want_inline_small_function_p): Use IPA profile.
3437 (want_inline_self_recursive_call_p): Use IPA profile.
3438 (edge_badness): Use IPA profile.
3439 (lookup_recursive_calls): Use IPA profile.
3440 (recursive_inlining): Do not pass frequency.
3441 (resolve_noninline_speculation): Do not update frequency.
3442 (inline_small_functions): Collect max of IPA profile.
3443 (dump_overall_stats): Dump IPA porfile.
3444 (dump_inline_stats): Dump IPA porfile.
3445 (ipa_inline): Collect IPA stats.
3446 * ipa-inline.h (clone_inlined_nodes): Update prototype.
3447 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
3448 (ipa_propagate_frequency): Use frequency function.
3449 (ipa_profile): Cleanup.
3450 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
3451 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
3452 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
3453 (input_node): Do not stream frequency.
3454 (input_edge): Do not stream frequency.
3455 (merge_profile_summaries): Scale only IPA profiles.
3456 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
3457 * predict.c (drop_profile): Do not recompute frequency.
3458 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
3459 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
3460 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
3461 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
3462 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
3463 * tree-inline.c (copy_bb): Do not scale frequency.
3464 (expand_call_inline): Do not scale frequency.
3465 (tree_function_versioning): Do not scale frequency.
3466 * ubsan.c (ubsan_create_edge): Do not pass frequency.
3468 2017-11-10 Julia Koval <julia.koval@intel.com>
3470 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
3471 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
3472 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
3473 (_mm256_maskz_gf2p8affine_epi64_epi8)
3474 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
3475 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
3476 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
3477 (__builtin_ia32_vgf2p8affineqb_v32qi)
3478 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
3479 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
3481 2017-11-10 Tamar Christina <tamar.christina@arm.com>
3485 (arm_option_override): Refactor.
3486 (arm_option_reconfigure_globals): New.
3487 (arm_options_perform_arch_sanity_checks): New.
3488 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
3490 (arm_options_perform_arch_sanity_checks): Likewise
3492 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
3494 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
3495 (power9-qpmul): New.
3496 * rs6000/rs6000.md ("type" attr): Add qmul.
3497 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
3498 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
3499 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
3501 2017-11-10 Martin Sebor <msebor@redhat.com>
3504 * builtins.c (compute_objsize): Handle arrays that
3505 compute_builtin_object_size likes to fail for. Make extern.
3506 * builtins.h (compute_objsize): Declare.
3507 (check_strncpy_sizes): New function.
3508 (expand_builtin_strncpy): Call check_strncpy_sizes.
3509 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
3510 -Wstringop-truncation.
3511 (gimple_fold_builtin_strncat): Same.
3512 * gimple.c (gimple_build_call_from_tree): Set call location.
3513 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
3514 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
3515 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
3516 (handle_builtin_strlen): Use strlen_to_stridx.
3517 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
3519 Use strlen_to_stridx.
3520 (pass_strlen::execute): Release strlen_to_stridx.
3521 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
3522 (-Wstringop-truncation): Document new option.
3524 2017-11-10 Martin Liska <mliska@suse.cz>
3526 PR gcov-profile/82702
3527 * gcov.c (main): Handle intermediate files in a different
3529 (get_gcov_intermediate_filename): New function.
3530 (output_gcov_file): Remove support of intermediate files.
3531 (generate_results): Allocate intermediate file.
3532 (release_structures): Clean-up properly fn_end.
3533 (output_intermediate_file): Start iterating with line 1.
3535 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3537 PR tree-optimization/82929
3538 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3539 ops_swapped_p non-static data member.
3540 (store_immediate_info::store_immediate_info): Clear it.
3541 (imm_store_chain_info::coalesce_immediate_stores): If swapping
3542 ops set ops_swapped_p.
3543 (count_multiple_uses): Handle ops_swapped_p.
3545 2017-11-10 Martin Liska <mliska@suse.cz>
3547 * coverage.c (coverage_init): Stream information about
3548 support of has_unexecuted_blocks.
3549 * doc/gcov.texi: Document that.
3550 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
3551 * gcov.c (read_graph_file): Likewise.
3552 (output_line_beginning): Fix a small issue with
3555 2017-11-10 Bin Cheng <bin.cheng@arm.com>
3557 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
3558 reference of trivial component.
3560 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3563 * gimple-ssa-store-merging.c
3564 (pass_store_merging::terminate_all_aliasing_chains): For
3565 gimple_store_p stmts also call refs_output_dependent_p.
3567 PR rtl-optimization/82913
3568 * compare-elim.c (try_merge_compare): Punt if def_insn is not
3571 2017-11-09 Jeff Law <law@redhat.com>
3573 * vr-values.h: New file with vr_values class.
3574 * tree-vrp.c: Include vr-values.h
3575 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
3576 data objects into the vr_values class.
3577 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
3578 (get_value_range): Make it a member function within vr_values class.
3579 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
3580 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
3581 (op_with_constant_singleton_value_range): Likewise.
3582 (extract_range_for_var_from_comparison_expr): Likewise.
3583 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
3584 (extract_range_from_binary_expr): Likewise.
3585 (extract_range_from_unary_expr): Likewise.
3586 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
3587 (check_for_binary_op_overflow, extract_range_basic): Likewise.
3588 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
3589 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
3590 (compare_name_with_value, compare_names): Likewise.
3591 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
3592 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
3593 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
3594 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
3595 (extract_range_from_phi_node): Likewise.
3596 (simplify_truth_ops_using_ranges): Likewise.
3597 (simplify_div_or_mod_using_ranges): Likewise.
3598 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
3599 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
3600 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
3601 (simplify_float_conversion_using_ranges): Likewise.
3602 (simplify_internal_call_using_ranges): Likewise.
3603 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
3604 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
3605 poitner via x_vr_values for calls into gimple folder.
3606 (vrp_initialize_lattice): Make this the vr_values ctor.
3607 (vrp_free_lattice): Make this the vr_values dtor.
3608 (set_vr_value): New function.
3609 (class vrp_prop): Add vr_values data member. Add various member
3610 functions as well as member functions that delegate to vr_values.
3611 (check_array_ref): Make a member function within vrp_prop class.
3612 (search_for_addr_array, vrp_initialize): Likewise.
3613 (vrp_finalize): Likewise. Revamp to avoid direct access to
3614 vr_value, values_propagated, etc.
3615 (check_array_bounds): Extract vrp_prop class instance pointer from
3616 walk info structure. Use it to call member functions.
3617 (check_all_array_refs): Make a member function within vrp_prop class.
3618 Smuggle class instance pointer via walk info structure.
3619 (x_vr_values): New local static.
3620 (vrp_valueize): Use x_vr_values to get class instance.
3621 (vr_valueize_1): Likewise.
3622 (class vrp_folder): Add vr_values data member. Add various member
3623 functions as well as member functions that delegate to vr_values.
3624 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
3625 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
3626 class instance from vr_values. Use it to call member functions.
3627 (vrp_dom_walker): Add vr_values data member.
3628 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
3629 instance via x_vr_values.
3630 (identify_jump_threads): Accept vr_values as argument. Store
3631 it into the walker structure.
3632 (evrp_dom_walker): Add vr_values class data member. Add various
3634 (evrp_dom_walker::try_find_new_range): Use vr_values data
3635 member to access the memory allocator.
3636 (evrp_dom_walker::before_dom_children): Store vr_values class
3637 instance into the vrp_folder class.
3638 (evrp_dom_walker::push_value_range): Rework to avoid direct
3639 access to num_vr_values and vr_value.
3640 (evrp_dom_walker::pop_value_range): Likewise.
3641 (execute_early_vrp): Remove call to vrp_initialize_lattice.
3642 Use vr_values to get to dump_all_value_ranges member function.
3643 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
3644 and simplify_cond_using_ranges_2 via vrp_prop class instance.
3645 Pass vr_values class instance down to identify_jump_threads.
3646 Remove call to vrp_free_lattice.
3647 (debug_all_value_ranges): Remove.
3649 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
3650 (vrp_folder): Likewise.
3652 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
3653 Get it from the existing bitmap instead.
3654 (vrp_intersect_ranges_1): Likewise.
3656 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3658 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3660 (store_immediate_info::store_immediate_info): Add bitnotp argument,
3661 set bit_not_p to it.
3662 (imm_store_chain_info::coalesce_immediate_stores): Break group
3663 if bit_not_p is different.
3664 (count_multiple_uses, split_group,
3665 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
3666 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
3667 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
3668 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
3669 to store_immediate_info ctor.
3671 2017-11-09 Jim Wilson <jimw@sifive.com>
3673 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
3674 (scan_prog_file): Likewise.
3676 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3678 * bb-reorder.c (max_entry_frequency): Remove.
3679 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
3680 connect_traces, push_to_next_round_p): Remove prototypes.
3681 (find_traces_1_round): Use counts only.
3682 (push_to_next_round_p): Likewise.
3683 (find_traces): Likewise.
3684 (rotate_loop): Likewise.
3685 (find_traces_1_round): Likewise.
3686 (connect_traces): Likewise.
3687 (edge_order): Likewise.
3689 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
3691 * config/arm/arm.c (output_return_instruction): Add comments to
3692 indicate requirement for cmse_nonsecure_entry return to account
3693 for the size of clearing instruction output here.
3694 (thumb_exit): Likewise.
3695 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
3696 return in hardfloat mode.
3698 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
3700 * config/rs6000/rs6000.c (machine_function): Add a bool,
3701 "toc_is_wrapped_separately".
3702 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
3703 if it wasn't explicitly set or unset, we are optimizing for speed, and
3704 doing separate shrink-wrapping.
3705 (rs6000_get_separate_components): Enable the TOC component if
3706 saving the TOC register in the prologue.
3707 (rs6000_components_for_bb): Handle the TOC component.
3708 (rs6000_emit_prologue_components): Store the TOC register where needed.
3709 (rs6000_set_handled_components): Mark TOC as handled, if handled.
3710 (rs6000_emit_prologue): Don't save the TOC if that is already done.
3712 2017-11-09 Martin Jambor <mjambor@suse.cz>
3714 * ipa-param-manipulation.c: New file.
3715 * ipa-param-manipulation.h: Likewise.
3716 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
3717 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
3718 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
3719 (ipa_parm_adjustment): Likewise.
3720 (ipa_parm_adjustment_vec): Likewise.
3721 (ipa_get_vector_of_formal_parms): Moved declaration to
3722 ipa-param-manipulation.h.
3723 (ipa_get_vector_of_formal_parm_types): Likewise.
3724 (ipa_modify_formal_parameters): Likewise.
3725 (ipa_modify_call_arguments): Likewise.
3726 (ipa_combine_adjustments): Likewise.
3727 (ipa_dump_param_adjustments): Likewise.
3728 (ipa_modify_expr): Likewise.
3729 (ipa_get_adjustment_candidate): Likewise.
3730 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
3731 ipa-param-manipulation.c.
3732 (ipa_get_vector_of_formal_parm_types): Likewise.
3733 (ipa_modify_formal_parameters): Likewise.
3734 (ipa_modify_call_arguments): Likewise.
3735 (ipa_modify_expr): Likewise.
3736 (get_ssa_base_param): Likewise.
3737 (ipa_get_adjustment_candidate): Likewise.
3738 (index_in_adjustments_multiple_times_p): Likewise.
3739 (ipa_combine_adjustments): Likewise.
3740 (ipa_dump_param_adjustments): Likewise.
3741 * tree-sra.c: Also include ipa-param-manipulation.h
3742 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
3745 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3746 Alan Hayward <alan.hayward@arm.com>
3747 David Sherwood <david.sherwood@arm.com>
3749 * doc/sourcebuild.texi (vect_masked_store): Document.
3751 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3752 Alan Hayward <alan.hayward@arm.com>
3753 David Sherwood <david.sherwood@arm.com>
3755 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
3757 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3758 Alan Hayward <alan.hayward@arm.com>
3759 David Sherwood <david.sherwood@arm.com>
3761 * doc/sourcebuild.texi (vect_variable_length): Document.
3763 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3764 Alan Hayward <alan.hayward@arm.com>
3765 David Sherwood <david.sherwood@arm.com>
3767 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
3769 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3770 Alan Hayward <alan.hayward@arm.com>
3771 David Sherwood <david.sherwood@arm.com>
3773 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
3775 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3776 Alan Hayward <alan.hayward@arm.com>
3777 David Sherwood <david.sherwood@arm.com>
3779 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
3780 previously undocumented selectors.
3781 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
3783 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3785 * doc/rtl.texi (const_vector): Say that elements can be
3786 const_wide_ints too.
3787 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
3788 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
3789 (gen_vec_duplicate): Use it instead of CONSTANT_P.
3790 * optabs.c (expand_vector_broadcast): Likewise.
3792 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3793 Alan Hayward <alan.hayward@arm.com>
3794 David Sherwood <david.sherwood@arm.com>
3796 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
3797 scaled index even if the unscaled address was invalid.
3798 Don't increase the complexity of using a scale in that case.
3800 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3801 Alan Hayward <alan.hayward@arm.com>
3802 David Sherwood <david.sherwood@arm.com>
3804 * doc/rtl.texi: Rewrite the subreg rules so that they partition
3805 the inner register into REGMODE_NATURAL_SIZE bytes rather than
3806 UNITS_PER_WORD bytes.
3807 * emit-rtl.c (validate_subreg): Divide subregs into blocks
3808 based on REGMODE_NATURAL_SIZE of the inner mode.
3809 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
3810 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
3811 * expmed.c (lowpart_bit_field_p): Divide the value up into
3812 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
3813 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
3814 whether something is likely to occupy more than one register.
3816 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3819 * ipa-inline-transform.c (update_noncloned_frequencies): Use
3820 profile_count::adjust_for_ipa_scaling.
3821 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
3822 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
3824 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
3826 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3828 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
3829 (split_group): Add total_orig and total_new arguments, estimate the
3830 number of statements related to the store group without store merging
3831 and with store merging.
3832 (imm_store_chain_info::output_merged_store): Adjust split_group
3833 callers, punt if estimated number of statements with store merging
3834 is not smaller than estimated number of statements without it.
3836 (handled_load): Remove has_single_use checks.
3837 (pass_store_merging::process_store): Likewise.
3839 2017-11-09 Richard Biener <rguenther@suse.de>
3841 PR tree-optimization/82902
3842 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
3844 2017-11-09 Martin Liska <mliska@suse.cz>
3847 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
3850 2017-11-09 Martin Liska <mliska@suse.cz>
3852 PR tree-optimization/82669
3853 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
3855 2017-11-09 Martin Liska <mliska@suse.cz>
3857 PR gcov-profile/48463
3858 * coverage.c (coverage_begin_function): Output also end locus
3859 of a function and information whether the function is
3861 * gcov-dump.c (tag_function): Parse and print the information.
3862 * gcov.c (INCLUDE_MAP): Add include.
3863 (INCLUDE_SET): Likewise.
3864 (struct line_info): Move earlier in the source file because
3865 of vector<line_info> in function_info structure.
3866 (line_info::line_info): Likewise.
3867 (line_info::has_block): Likewise.
3868 (struct source_info): Add new member index.
3869 (source_info::get_functions_at_location): New function.
3870 (function_info::group_line_p): New function.
3871 (output_intermediate_line): New function.
3872 (output_intermediate_file): Use the mentioned function.
3873 (struct function_start): New.
3874 (struct function_start_pair_hash): Likewise.
3875 (process_file): Add code that identifies group functions.
3876 Assign lines either to global or function scope.
3877 (generate_results): Skip artificial functions.
3878 (find_source): Assign index for each source file.
3879 (read_graph_file): Read new flag artificial and end_line.
3880 (add_line_counts): Assign it either to global of function scope.
3881 (accumulate_line_counts): Isolate core of the function to
3882 accumulate_line_info and call it for both function and global
3884 (accumulate_line_info): New function.
3885 (output_line_beginning): Fix GNU coding style.
3886 (print_source_line): New function.
3887 (output_line_details): Likewise.
3888 (output_function_details): Likewise.
3889 (output_lines): Iterate both source (global) scope and function
3891 (struct function_line_start_cmp): New class.
3892 * doc/gcov.texi: Reflect changes in documentation.
3894 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3897 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
3898 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
3899 and similarly for not instead of neg.
3901 2017-11-08 Andi Kleen <ak@linux.intel.com>
3903 * config/i386/i386.opt: Add -mforce-indirect-call.
3904 * config/i386/predicates.md: Check for flag_force_indirect_call.
3905 * doc/invoke.texi: Document -mforce-indirect-call
3907 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
3909 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
3911 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
3912 config/riscv/riscv.c (predict.h): New include.
3913 (riscv_slow_unaligned_access_p): No longer static.
3914 (riscv_block_move_straight): Add require.
3915 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3917 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3920 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
3921 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
3922 nonimmediate_operand predicate for operand 1 instead of
3925 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3927 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
3929 * config/aarch64/constraints.md (Uml): New constraint.
3930 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
3933 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3935 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
3936 of two vec_duplicates into a vec_concat.
3938 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3940 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3941 Simplify vec_merge of vec_duplicate and vec_concat.
3942 * config/aarch64/constraints.md (Utq): New constraint.
3943 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
3946 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3948 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3949 Simplify vec_merge of vec_duplicate and const_vector.
3950 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
3952 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
3953 mode iterator. Update predicate on operand 1 to
3954 handle non-const_vec constants. Delete constraints.
3955 (*aarch64_combinez_be<mode>): Likewise for operand 2.
3957 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3959 PR tree-optimization/78821
3960 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
3962 (store_operand_info::store_operand_info): Initialize it to false.
3963 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
3964 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
3965 store in the group, and if chain_info is non-NULL, to ignore altogether
3967 (compatible_load_p): Fail if bit_not_p does not match.
3968 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
3969 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
3970 (pass_store_merging::process_store): Adjust
3971 terminate_all_aliasing_chains calls to pass NULL in all current spots,
3972 call terminate_all_aliasing_chains newly when adding a store into
3973 a chain with non-NULL chain_info.
3975 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3977 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
3979 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3981 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3983 (aarch64_layout_frame): Initialise emit_frame_chain.
3984 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
3985 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
3987 2017-11-08 Martin Liska <mliska@suse.cz>
3989 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
3990 of gimple_call_internal_p.
3992 2017-11-07 Tom de Vries <tom@codesourcery.com>
3994 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
3996 2017-11-07 Tom de Vries <tom@codesourcery.com>
3998 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4001 2017-11-08 Martin Liska <mliska@suse.cz>
4004 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4006 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4008 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4010 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4013 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4014 (*cmp<mode>_ccz_1): New insn with $k alternative.
4017 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4018 RTX_COMM_COMPARE as commutative as well.
4019 (ix86_binary_operator_ok): Formatting fix.
4020 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
4021 *<code><mode>3<mask_name><round_saeonly_name>,
4022 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
4023 *<s>mul<mode>3_highpart<mask_name>,
4024 *vec_widen_umult_even_v16si<mask_name>,
4025 *vec_widen_umult_even_v8si<mask_name>,
4026 *vec_widen_umult_even_v4si<mask_name>,
4027 *vec_widen_smult_even_v16si<mask_name>,
4028 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
4029 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
4030 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
4031 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
4032 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
4033 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
4034 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
4035 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
4036 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
4037 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
4038 ix86_binary_operator_ok. Formatting fixes.
4039 (*<plusminus_insn><mode>3<mask_name><round_name>,
4040 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
4043 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4045 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
4046 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
4047 eq and ne if TARGET_ISEL.
4048 (cmp): New code_iterator.
4049 (UNS, UNSU_, UNSIK): New code_attrs.
4050 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
4051 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
4053 ("eq<mode>3"): ... this.
4054 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
4056 ("ne<mode>3"): ... this.
4058 2017-11-07 Julia Koval <julia.koval@intel.com>
4061 * common/config/i386/i386-common.c
4062 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
4063 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
4064 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
4065 * config/i386/i386.opt: Ditto.
4066 * config/i386/i386.c (ix86_target_string): Ditto.
4067 (ix86_option_override_internal): Ditto.
4068 (ix86_init_mpx_builtins): Move MPX to args2.
4069 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
4070 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4071 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4072 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4073 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4074 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4075 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
4077 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
4080 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
4081 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
4082 (zero-extendsidi peephole2): Remove peephole.
4084 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4087 * stor-layout.c: Include attribs.h.
4088 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
4089 explicit lookup of "aligned" attribute.
4091 2017-11-07 Andrew Waterman <andrew@sifive.com>
4093 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
4094 (riscv_expand_block_move): Likewise.
4095 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
4097 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
4098 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
4099 * config/riscv/riscv.c (riscv_block_move_straight): New function.
4100 (riscv_adjust_block_mem): Likewise.
4101 (riscv_block_move_loop): Likewise.
4102 (riscv_expand_block_move): Likewise.
4103 * config/riscv/riscv.md (movmemsi): New pattern.
4105 2017-11-07 Michael Clark <michaeljclark@mac.com>
4107 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
4108 (MUSL_DYNAMIC_LINKER): Likewise.
4110 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
4112 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
4113 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
4116 2017-11-07 Richard Biener <rguenther@suse.de>
4118 * match.pd: Fix build.
4120 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
4121 Jackson Woodruff <jackson.woodruff@arm.com>
4123 PR tree-optimization/71026
4124 * match.pd: Canonicalize negate in division.
4126 2017-11-07 Sudakshina Das <sudi.das@arm.com>
4129 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
4131 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4133 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
4134 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
4136 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4138 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
4139 non-scalar integral types.
4140 * match.pd (negate_expr_p): Handle MINUS_EXPR.
4141 (-(A-B), -(~A)): New transformations.
4143 2017-11-07 Tom de Vries <tom@codesourcery.com>
4145 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
4146 semicolon after "do {} while (0)".
4147 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4148 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4149 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4150 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4151 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4153 2017-11-07 Tom de Vries <tom@codesourcery.com>
4155 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
4156 after "do {} while (0)".
4157 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4158 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4159 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4160 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4161 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4163 2017-11-07 Tom de Vries <tom@codesourcery.com>
4166 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
4168 (arm_rtx_costs_internal): Add missing semicolon after
4169 HANDLE_NARROW_SHIFT_ARITH call.
4171 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4173 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4174 disable isel if it was not set explicitly.
4176 2017-11-06 James Bowman <james.bowman@ftdichip.com>
4178 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
4179 (add_type_attribute) likewise.
4181 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
4183 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
4184 of struct ix86_frame.
4185 (ix86_initial_elimination_offset): Likewise.
4186 (ix86_expand_split_stack_prologue): Likewise.
4188 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
4190 * tree-vrp.h (enum value_range_type): Update stale comment.
4192 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4193 Alan Hayward <alan.hayward@arm.com>
4194 David Sherwood <david.sherwood@arm.com>
4196 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
4197 (aarch64_expand_vec_perm_const): Take the number of units too.
4198 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
4199 (aarch64_expand_vec_perm_const): Likewise.
4200 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
4201 (vec_perm<mode>): Update accordingly.
4203 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4204 Alan Hayward <alan.hayward@arm.com>
4205 David Sherwood <david.sherwood@arm.com>
4207 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
4208 Take the number of units too.
4209 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
4210 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
4211 but check for a vector mode before rather than after the call.
4212 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
4213 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
4214 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
4215 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
4216 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
4217 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
4218 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
4219 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
4220 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
4221 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
4222 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
4223 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
4224 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
4225 (aarch64_sqdmull2_n<mode>): Update accordingly.
4227 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4228 Alan Hayward <alan.hayward@arm.com>
4229 David Sherwood <david.sherwood@arm.com>
4231 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
4232 the number of units too.
4233 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
4234 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
4235 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
4236 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
4237 (vec_store_lanesxi<mode>): Update accordingly.
4239 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4240 Alan Hayward <alan.hayward@arm.com>
4241 David Sherwood <david.sherwood@arm.com>
4243 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
4244 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
4245 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
4246 of units rather than the mode.
4247 * config/aarch64/iterators.md (nunits): New mode attribute.
4248 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4249 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
4250 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
4251 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
4252 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
4253 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
4254 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
4255 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
4256 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
4257 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
4258 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4259 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
4260 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
4261 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
4262 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
4263 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
4264 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
4265 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
4266 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
4267 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
4268 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
4269 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
4270 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
4271 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4272 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
4273 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
4274 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4275 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
4276 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
4277 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
4278 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
4279 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
4280 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
4281 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
4282 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
4283 (aarch64_simd_vec_setv2di): Likewise.
4285 2017-11-06 Carl Love <cel@us.ibm.com>
4287 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
4289 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
4290 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
4292 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
4294 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
4295 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
4296 Add power 8 macro expansions.
4297 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
4298 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
4299 power 8 instructions. (VSX_XXBR): Add iterator.
4301 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
4303 * config/arm/arm.md (predicable_short_it): Change default to "no",
4304 improve documentation, remove uses that are identical to the default.
4305 (enabled_for_depr_it): Rename to enabled_for_short_it.
4306 * gcc/config/arm/arm-fixed.md (predicable_short_it):
4307 Remove default uses.
4308 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
4309 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
4310 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
4311 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
4313 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4316 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
4317 float128 helper macros, which are no longer used after deleting
4318 the old 'q' built-in functions, and moving the round to odd
4319 built-in functions to being special built-in functions.
4320 (BU_FLOAT128_2): Likewise.
4321 (BU_FLOAT128_1_HW): Likewise.
4322 (BU_FLOAT128_2_HW): Likewise.
4323 (BU_FLOAT128_3_HW): Likewise.
4324 (FABSQ): Delete old 'q' built-in functions.
4325 (COPYSIGNQ): Likewise.
4326 (SQRTF128_ODD): Move round to odd built-in functions to be
4327 special built-in functions, so that we can handle
4328 -mabi=ieeelongdouble.
4329 (TRUNCF128_ODD): Likewise.
4330 (ADDF128_ODD): Likewise.
4331 (SUBF128_ODD): Likewise.
4332 (MULF128_ODD): Likewise.
4333 (DIVF128_ODD): Likewise.
4334 (FMAF128_ODD): Likewise.
4335 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
4336 built-in names to 'f128'.
4337 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
4338 old 'q' built-in functions, as the machine independent code for
4339 'f128' built-in functions handles this.
4340 (rs6000_expand_builtin): Add expansion for float128 round to odd
4341 functions, keying off on -mabi=ieeelongdouble of whether to use
4342 the KFmode or TFmode variant.
4343 (rs6000_init_builtins): Initialize the _Float128 round to odd
4345 * doc/extend.texi (PowerPC Built-in Functions): Document the old
4346 _Float128 'q' built-in functions are now mapped into the new
4347 'f128' built-in functions.
4349 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
4351 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
4352 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
4354 2017-11-06 David Malcolm <dmalcolm@redhat.com>
4357 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
4358 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
4359 * toplev.c: Include "ipa-fnsummary.h".
4360 (toplev::finalize): Call ipa_fnsummary_c_finalize.
4362 2017-11-06 Jakub Jelinek <jakub@redhat.com>
4364 PR tree-optimization/82838
4365 * gimple-ssa-store-merging.c
4366 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
4367 on a separate gimple_seq which is then appended to seq.
4369 2017-11-06 Jeff Law <law@redhat.com>
4372 * config/i386/i386.c (PROBE_INTERVAL): Remove.
4373 (get_probe_interval): New functions.
4374 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
4375 (ix86_adjust_stack_and_probe): Likewise.
4376 (output_adjust_stack_and_probe): Likewise.
4377 (ix86_emit_probe_stack_range): Likewise.
4378 (ix86_expand_prologue): Likewise.
4380 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4382 PR tree-optimization/82816
4383 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
4384 if the modes of the two types are the same.
4385 (convert_plusminus_to_widen): Likewise.
4387 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4389 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
4391 (usadv16qi): New define_expand.
4392 (usadv8hi): New define_expand.
4394 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
4397 * ipa-inline-transform.c (update_noncloned_frequencies): Always
4399 (inline_transform): Likewise.
4400 * predict.c (counts_to_freqs): Remove useless conditional.
4401 * profile-count.h (profile_count::apply_scale): Move sanity check.
4402 * tree-inline.c (copy_bb): Always scale.
4403 (copy_cfg_body): Likewise.
4405 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
4408 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
4410 (*cmp_ite0): Add enabled_for_depr_it attribute.
4411 (*cmp_ite1): Likewise.
4413 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4415 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
4418 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4420 * tree-vrp.c (vrp_int_const_binop): Return true on success and
4421 return the value by pointer.
4422 (extract_range_from_multiplicative_op_1): Update accordingly.
4423 Return as soon as an operation fails.
4425 2017-11-05 Tom de Vries <tom@codesourcery.com>
4428 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
4429 (DEF_SANITIZER_BUILTIN): ... here.
4430 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
4431 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
4433 2017-11-05 Tom de Vries <tom@codesourcery.com>
4436 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
4438 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
4439 ASM_OUTPUT_BEFORE_CASE_LABEL call.
4440 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
4442 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4443 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4444 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4446 2017-11-05 Tom de Vries <tom@codesourcery.com>
4449 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
4452 2017-11-04 Michael Clark <michaeljclark@mac.com>
4454 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
4455 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
4457 (*addsi3_extended): Likewise.
4458 (*addsi3_extended2): Likewise.
4459 (<optab>si3): Likewise.
4460 (<optab>di3): Likewise.
4461 (<optab><mode>3): Likewise.
4462 (<*optabe>si3_internal): Likewise.
4463 (zero_extendqi<SUPERQI:mode>2): Likewise.
4464 (*add<mode>hi3): Likewise.
4465 (*xor<mode>hi3): Likewise.
4466 (<optab>di3): Likewise.
4467 (*<optab>si3_extend): Likewise.
4468 (*sge<u>_<X:mode><GPR:mode>): Likewise.
4469 (*slt<u>_<X:mode><GPR:mode>): Likewise.
4470 (*sle<u>_<X:mode><GPR:mode>): Likewise.
4472 2017-11-04 Andrew Waterman <andrew@sifive.com>
4474 * config/riscv/riscv.c (riscv_option_override): Conditionally set
4475 TARGET_STRICT_ALIGN based upon -mtune argument.
4477 2017-11-04 Andrew Waterman <andrew@sifive.com>
4479 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
4481 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
4483 * config/i386/i386.c (choose_basereg): Use optional scratch
4484 register and add assertion.
4485 (x86_emit_outlined_ms2sysv_save): Use scratch register when
4486 needed, and don't allocate stack.
4487 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
4488 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
4489 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
4491 2017-11-03 Jeff Law <law@redhat.com>
4493 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
4494 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
4495 to probe at the start of a noreturn function.
4497 2017-11-03 Jakub Jelinek <jakub@redhat.com>
4499 PR tree-optimization/78821
4500 * gimple-ssa-store-merging.c: Update the file comment.
4501 (MAX_STORE_ALIAS_CHECKS): Define.
4502 (struct store_operand_info): New type.
4503 (store_operand_info::store_operand_info): New constructor.
4504 (struct store_immediate_info): Add rhs_code and ops data members.
4505 (store_immediate_info::store_immediate_info): Add rhscode, op0r
4506 and op1r arguments to the ctor, initialize corresponding data members.
4507 (struct merged_store_group): Add load_align_base and load_align
4509 (merged_store_group::merged_store_group): Initialize them.
4510 (merged_store_group::do_merge): Update them.
4511 (merged_store_group::apply_stores): Pick the constant for
4512 encode_tree_to_bitpos from one of the two operands, or skip
4513 encode_tree_to_bitpos if neither operand is a constant.
4514 (class pass_store_merging): Add process_store method decl. Remove
4515 bool argument from terminate_all_aliasing_chains method decl.
4516 (pass_store_merging::terminate_all_aliasing_chains): Remove
4517 var_offset_p argument and corresponding handling.
4518 (stmts_may_clobber_ref_p): New function.
4519 (compatible_load_p): New function.
4520 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
4521 if there is overlap and rhs_code is not INTEGER_CST. For
4522 non-overlapping stores terminate group if rhs is not mergeable.
4523 (get_alias_type_for_stmts): Change first argument from
4524 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
4525 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
4526 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
4528 (get_location_for_stmts): Change first argument from
4529 auto_vec<gimple *> & to vec<gimple *> &.
4530 (struct split_store): Remove orig_stmts data member, add orig_stores.
4531 (split_store::split_store): Create orig_stores rather than orig_stmts.
4532 (find_constituent_stmts): Renamed to ...
4533 (find_constituent_stores): ... this. Change second argument from
4534 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
4535 to info structures rather than the statements.
4536 (split_group): Rename ALLOW_UNALIGNED argument to
4537 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
4538 it. Adjust find_constituent_stores caller.
4539 (imm_store_chain_info::output_merged_store): Handle rhs_code other
4540 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
4541 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
4542 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
4543 (mem_valid_for_store_merging): New function.
4544 (handled_load): New function.
4545 (pass_store_merging::process_store): New method.
4546 (pass_store_merging::execute): Use process_store method. Adjust
4547 terminate_all_aliasing_chains caller.
4549 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4551 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4552 Return true for more constants, symbols and label references.
4553 (aarch64_valid_floating_const): Remove unused function.
4555 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4558 * config/aarch64/aarch64.c (aarch64_layout_frame):
4559 Undo forcing of LR at bottom of frame.
4561 2017-11-03 Jeff Law <law@redhat.com>
4564 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
4565 for int_registers_saved.
4567 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
4568 extracted from tree-ssa-dom.c.
4569 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
4570 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
4571 (record_equivalences_from_incoming_edge): Add additional argument
4572 to single_pred_edge_ignoring_loop_edges call.
4573 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
4574 (uncprop_dom_walker::before_dom_children): Add additional argument
4575 to single_pred_edge_ignoring_loop_edges call.
4576 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
4577 single_pred_edge_ignoring_loop_edges rather than open coding.
4578 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
4580 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
4582 * match.pd (-(-A)): Rewrite.
4584 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4586 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
4587 (rs6000_emit_int_cmove): New declaration.
4588 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
4589 (rs6000_emit_sISEL): Delete.
4590 (rs6000_emit_int_cmove): Make non-static.
4591 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
4592 instead of rs6000_emit_sISEL.
4594 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
4596 * asan.c (create_cond_insert_point): Maintain profile.
4597 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
4599 * basic-block.h (struct basic_block_def): Remove frequency.
4600 (EDGE_FREQUENCY): Use to_frequency
4601 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
4603 (find_traces): Update to use to_frequency.
4604 (find_traces_1_round): Likewise; use only IPA counts.
4605 (bb_to_key): Likewise.
4606 (connect_traces): Use IPA counts only.
4607 (copy_bb_p): Update to use to_frequency.
4608 (fix_up_crossing_landing_pad): Likewise.
4609 (sanitize_hot_paths): Likewise.
4610 * bt-load.c (basic_block_freq): Likewise.
4611 * cfg.c (init_flow): Set count_max to uninitialized.
4612 (check_bb_profile): Remove frequencies; check counts.
4613 (dump_bb_info): Do not dump frequencies.
4614 (update_bb_profile_for_threading): Update counts only.
4615 (scale_bbs_frequencies_int): Likewise.
4616 (MAX_SAFE_MULTIPLIER): Remove.
4617 (scale_bbs_frequencies_gcov_type): Update counts only.
4618 (scale_bbs_frequencies_profile_count): Update counts only.
4619 (scale_bbs_frequencies): Update counts only.
4620 * cfg.h (struct control_flow_graph): Add count-max.
4621 (update_bb_profile_for_threading): Update prototype.
4622 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
4623 (find_many_sub_basic_blocks): Likewise.
4624 * cfgcleanup.c (try_forward_edges): Likewise.
4625 (try_crossjump_to_edge): Likewise.
4626 * cfgexpand.c (expand_gimple_cond): Likewise.
4627 (expand_gimple_tailcall): Likewise.
4628 (construct_init_block): Likewise.
4629 (construct_exit_block): Likewise.
4630 * cfghooks.c (verify_flow_info): Check consistency of counts.
4631 (dump_bb_for_graph): Do not dump frequencies.
4632 (split_block_1): Do not update frequencies.
4633 (split_edge): Do not update frequencies.
4634 (make_forwarder_block): Do not update frequencies.
4635 (duplicate_block): Do not update frequencies.
4636 (account_profile_record): Do not update frequencies.
4637 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
4638 for global heuristics.
4639 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
4640 (expected_loop_iterations_unbounded): Use counts only.
4641 * cfgloopmanip.c (scale_loop_profile): Simplify.
4642 (create_empty_loop_on_edge): Simplify
4644 (duplicate_loop_to_header_edge): Simplify
4645 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
4646 (update_br_prob_note): Take care of removing note when profile
4648 (relink_block_chain): Do not dump frequency.
4649 (rtl_account_profile_record): Use to_frequency.
4650 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
4651 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
4652 (cgraph_update_edges_for_call_stmt_node): Likewise.
4653 (cgraph_edge::verify_count_and_frequency): Update.
4654 (cgraph_node::verify_node): Temporarily disable frequency verification.
4655 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
4656 to_cgraph_frequency.
4657 (cgraph_edge::rebuild_edges): Convert to ipa counts.
4658 * cgraphunit.c (init_lowered_empty_function): Do not initialize
4660 (cgraph_node::expand_thunk): Update profile.
4661 * except.c (dw2_build_landing_pads): Do not update frequency.
4662 * final.c (compute_alignments): Use to_frequency.
4663 (dump_basic_block_info): Do not dump frequency.
4664 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
4665 (dump_gimple_bb_header): Do not dump frequency.
4666 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
4668 * gimple-streamer-in.c (input_bb): Do not stream frequency.
4669 * gimple-streamer-out.c (output_bb): Do not stream frequency.
4670 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
4671 (init_before_recovery): Do not update frequency.
4672 (sched_create_recovery_edges): Do not update frequency.
4673 * hsa-gen.c (convert_switch_statements): Do not update frequency.
4674 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
4675 (ipa_cp_c_finalize): Set max_count to uninitialized.
4676 * ipa-fnsummary.c (get_minimal_bb): Use counts.
4677 (param_change_prob): Use counts.
4678 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
4680 * ipa-split.c (consider_split): Use to_frequency.
4681 (split_function): Use to_frequency.
4682 * ira-build.c (loop_compare_func): Likewise.
4683 (mark_loops_for_removal): Likewise.
4684 (mark_all_loops_for_removal): Likewise.
4685 * loop-doloop.c (doloop_modify): Do not update frequency.
4686 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
4688 * lto-streamer-in.c (input_function): Update count_max.
4689 * omp-expand.c (expand_omp_taskreg): Update count_max.
4690 * omp-simd-clone.c (simd_clone_adjust): Update profile.
4691 * predict.c (maybe_hot_frequency_p): Use to_frequency.
4692 (maybe_hot_count_p): Use ipa counts only.
4693 (maybe_hot_bb_p): Simplify.
4694 (maybe_hot_edge_p): Simplify.
4695 (probably_never_executed): Do not take frequency argument.
4696 (probably_never_executed_bb_p): Do not pass frequency.
4697 (probably_never_executed_edge_p): Likewise.
4698 (combine_predictions_for_bb): Check that profile is nonzero.
4699 (propagate_freq): Do not set frequency.
4700 (drop_profile): Simplify.
4701 (counts_to_freqs): Simplify.
4702 (expensive_function_p): Use to_frequency.
4703 (propagate_unlikely_bbs_forward): Simplify.
4704 (determine_unlikely_bbs): Simplify.
4705 (estimate_bb_frequencies): Add hack to silence graphite issues.
4706 (compute_function_frequency): Use ipa counts.
4707 (pass_profile::execute): Update.
4708 (rebuild_frequencies): Use counts only.
4709 (force_edge_cold): Use counts only.
4710 * profile-count.c (profile_count::dump): Dump new count types.
4711 (profile_count::differs_from_p): Check compatiblity.
4712 (profile_count::to_frequency): New function.
4713 (profile_count::to_cgraph_frequency): New function.
4714 * profile-count.h (struct function): Declare.
4715 (enum profile_quality): Add profile_guessed_local and
4716 profile_guessed_global0.
4717 (class profile_proability): Decrease number of bits to 29;
4718 update from_reg_br_prob_note and to_reg_br_prob_note.
4719 (class profile_count: Update comment; decrease number of bits
4720 to 61. Check compatibility.
4721 (profile_count::compatible_p): New private member function.
4722 (profile_count::ipa_p): New member function.
4723 (profile_count::operator<): Handle global zero correctly.
4724 (profile_count::operator>): Handle global zero correctly.
4725 (profile_count::operator<=): Handle global zero correctly.
4726 (profile_count::operator>=): Handle global zero correctly.
4727 (profile_count::nonzero_p): New member function.
4728 (profile_count::force_nonzero): New member function.
4729 (profile_count::max): New member function.
4730 (profile_count::apply_scale): Handle IPA scalling.
4731 (profile_count::guessed_local): New member function.
4732 (profile_count::global0): New member function.
4733 (profile_count::ipa): New member function.
4734 (profile_count::to_frequency): Declare.
4735 (profile_count::to_cgraph_frequency): Declare.
4736 * profile.c (OVERLAP_BASE): Delete.
4737 (compute_frequency_overlap): Delete.
4738 (compute_branch_probabilities): Do not use compute_frequency_overlap.
4739 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
4740 * sched-ebb.c (rank): Use counts only.
4741 * shrink-wrap.c (handle_simple_exit): Use counts only.
4742 (try_shrink_wrapping): Use counts only.
4743 (place_prologue_for_one_component): Use counts only.
4744 * tracer.c (find_best_predecessor): Use to_frequency.
4745 (find_trace): Use to_frequency.
4746 (tail_duplicate): Use to_frequency.
4747 * trans-mem.c (expand_transaction): Do not update frequency.
4748 * tree-call-cdce.c: Do not update frequency.
4749 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
4750 (gimple_merge_blocks): Likewise.
4751 (gimple_split_edge): Likewise.
4752 (gimple_duplicate_sese_region): Likewise.
4753 (gimple_duplicate_sese_tail): Likewise.
4754 (move_sese_region_to_fn): Likewise.
4755 (gimple_account_profile_record): Likewise.
4756 (insert_cond_bb): Likewise.
4757 * tree-complex.c (expand_complex_div_wide): Likewise.
4758 * tree-eh.c (lower_resx): Update profile.
4759 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
4761 (initialize_cfun): Do not initialize frequencies
4762 (freqs_to_counts): Delete.
4763 (copy_cfg_body): Ignore count parameter.
4764 (copy_body): Update.
4765 (expand_call_inline): Update count_max.
4766 (optimize_inline_calls): Update count_max.
4767 (tree_function_versioning): Update count_max.
4768 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
4769 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
4771 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
4772 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
4773 (try_peel_loop): Likewise.
4774 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
4776 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
4777 (tree_transform_and_unroll_loop): Do not use frequencies
4778 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
4779 Use reliable prediction only.
4780 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
4781 * tree-ssa-sink.c (select_best_block): Use to_frequency.
4782 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
4783 probability scaling.
4784 * tree-ssa-threadupdate.c (create_block_for_threading): Do
4785 not update frequency
4786 (any_remaining_duplicated_blocks): Likewise.
4787 (update_profile): Likewise.
4788 (estimated_freqs_path): Delete.
4789 (freqs_to_counts_path): Delete.
4790 (clear_counts_path): Delete.
4791 (ssa_fix_duplicate_block_edges): Likewise.
4792 (duplicate_thread_path): Likewise.
4793 * tree-switch-conversion.c (gen_inbound_check): Use counts.
4794 * tree-tailcall.c (decrease_profile): Do not update frequency.
4795 (eliminate_tail_call): Likewise.
4796 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
4797 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
4798 (optimize_mask_stores): Likewise.
4799 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4800 * ubsan.c (ubsan_expand_null_ifn): Update profile.
4801 (ubsan_expand_ptr_ifn): Update profile.
4802 * value-prof.c (gimple_ic): Simplify.
4803 * value-prof.h (gimple_ic): Update prototype.
4804 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
4805 * ipa-inline.c (compute_uninlined_call_time): Be sure that
4807 (want_inline_self_recursive_call_p): Likewise.
4808 (resolve_noninline_speculation): Only cummulate defined counts.
4809 (inline_small_functions): Use nonzero_p.
4810 (ipa_inline): Do not access freed node.
4812 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4814 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4815 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
4817 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
4819 * config/riscv/riscv.c (riscv_legitimize_move): Handle
4820 non-legitimate address.
4822 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4824 * config/rs6000/rs6000.md (*lt0_disi): Delete.
4825 (*lt0_<mode>di, *lt0_<mode>si): New.
4827 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4829 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
4830 TARGET_PAIRED_FLOAT.
4832 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
4833 Jim Wilson <jim.wilson@linaro.org>
4835 * config/aarch64/aarch64-cores.def (saphira): New CPU.
4836 * config/aarch64/aarch64-tune.md: Regenerated.
4837 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
4838 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
4840 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
4842 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
4843 (arc_expand_prologue): Restore blink for millicode.
4844 * config/arc/linux.h (LINK_EH_SPEC): Defined.
4846 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
4849 * config/i386/i386.c (ix86_vector_duplicate_value): Use
4850 gen_vec_duplicate after forcing the scalar into a register.
4852 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
4854 * combine (try_combine): Print the insns input to try_combine to the
4857 2017-11-02 Steve Ellcey <sellcey@cavium.com>
4860 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
4861 Remove second argument from aarch64_process_target_attr call.
4862 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
4864 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
4866 (aarch64_handle_attr_arch): Remove second argument.
4867 (aarch64_handle_attr_cpu): Ditto.
4868 (aarch64_handle_attr_tune): Ditto.
4869 (aarch64_handle_attr_isa_flags): Ditto.
4870 (aarch64_process_one_target_attr): Ditto.
4871 (aarch64_process_target_attr): Ditto.
4872 (aarch64_option_valid_attribute_p): Remove second argument.
4873 on aarch64_process_target_attr call.
4875 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4877 * diagnostic.c: Include "selftest-diagnostic.h".
4878 (selftest::assert_location_text): New function.
4879 (selftest::test_diagnostic_get_location_text): New function.
4880 (selftest::diagnostic_c_tests): Call it.
4882 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4884 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
4885 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
4886 (class selftest::test_diagnostic_context): Move to...
4887 * selftest-diagnostic.c: New file.
4888 * selftest-diagnostic.h: New file.
4890 2017-11-02 James Bowman <james.bowman@ftdichip.com>
4892 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
4893 offset range for FT32B.
4894 * config/ft32/ft32.h: option "mcompress" enables relaxation.
4895 * config/ft32/ft32.md: Add TARGET_NOPM.
4896 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
4897 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
4899 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
4901 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
4903 2017-11-02 Jeff Law <law@redhat.com>
4905 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
4906 virtual keyword on FINAL OVERRIDE members.
4908 * tree-ssa-propagate.h (ssa_propagation_engine): Group
4909 virtuals together. Add virtual destructor.
4910 (substitute_and_fold_engine): Similarly.
4912 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
4914 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
4916 2017-11-02 Richard Biener <rguenther@suse.de>
4918 PR tree-optimization/82795
4919 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
4921 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4923 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
4925 (gcc_GAS_CHECK_FEATURE): Remove.
4926 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
4927 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
4928 * configure: Regenerate.
4930 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
4932 * config/arc/arc.c (hwloop_optimize): Account for empty
4935 2017-11-02 Richard Biener <rguenther@suse.de>
4938 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
4939 Truncate ARRAY_REF index and element size.
4941 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4943 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
4945 2017-11-01 Jeff Law <law@redhat.com>
4947 * tree-ssa-ccp.c (ccp_folder): New class derived from
4948 substitute_and_fold_engine.
4949 (ccp_folder::get_value): New member function.
4950 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
4951 (ccp_fold_stmt): Remove prototype.
4952 (ccp_finalize): Call substitute_and_fold from the ccp_class.
4953 * tree-ssa-copy.c (copy_folder): New class derived from
4954 substitute_and_fold_engine.
4955 (copy_folder::get_value): Renamed from get_value.
4956 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
4957 * tree-vrp.c (vrp_folder): New class derived from
4958 substitute_and_fold_engine.
4959 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
4960 (vrp_folder::get_value): New member function.
4961 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
4962 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
4963 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
4964 provide a class interface to folder/substitute routines.
4965 (ssa_prop_fold_stmt_fn): Remove typedef.
4966 (ssa_prop_get_value_fn): Likewise.
4967 (subsitute_and_fold): Remove prototype.
4968 (replace_uses_in): Likewise.
4969 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
4970 Renamed from replace_uses_in. Call the virtual member function
4971 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
4972 (substitute_and_fold_dom_walker): Remove initialization of
4973 data member entries for calbacks. Add substitute_and_fold_engine
4974 member and initialize it.
4975 (substitute_and_fold_dom_walker::before_dom_children0: Use the
4976 member functions for get_value, replace_phi_args_in c
4977 replace_uses_in, and fold_stmt calls.
4978 (substitute_and_fold_engine::substitute_and_fold): Renamed from
4979 substitute_and_fold. Remove assert. Update ctor call.
4981 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
4982 (ssa_prop_visit_phi_fn): Likewise.
4983 (class ssa_propagation_engine): New class to provide an interface
4985 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
4987 (ssa_prop_visit_phi): Likewise.
4988 (ssa_propagation_engine::simulate_stmt): Moved into class.
4989 Call visit_phi/visit_stmt from the class rather than via
4990 file scoped static variables.
4991 (ssa_propagation_engine::simulate_block): Moved into class.
4992 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
4993 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
4994 set file scoped statics for the visit_stmt/visit_phi callbacks.
4995 * tree-complex.c (complex_propagate): New class derived from
4996 ssa_propagation_engine.
4997 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
4998 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
4999 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5001 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5002 ssa_propagation_engine.
5003 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5004 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5005 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5006 * tree-ssa-copy.c (copy_prop): New class derived from
5007 ssa_propagation_engine.
5008 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5009 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5010 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5011 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5012 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5013 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5014 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5016 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5018 PR rtl-optimization/82778
5019 PR rtl-optimization/82597
5020 * compare-elim.c (struct comparison): Add in_a_setter field.
5021 (find_comparison_dom_walker::before_dom_children): Remove killed
5022 bitmap and df_simulate_find_defs call, instead walk the defs.
5023 Compute last_setter and initialize in_a_setter. Merge definitions
5024 with first initialization for a few variables.
5025 (try_validate_parallel): Use insn_invalid_p instead of
5026 recog_memoized. Return insn rather than just the pattern.
5027 (try_merge_compare): Fix up comment. Don't uselessly test if
5028 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
5030 (execute_compare_elim_after_reload): Remove df_chain_add_problem
5033 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5034 Alan Hayward <alan.hayward@arm.com>
5035 David Sherwood <david.sherwood@arm.com>
5037 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
5038 aarch64_hard_regno_nregs to get the number of registers
5041 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5042 Alan Hayward <alan.hayward@arm.com>
5043 David Sherwood <david.sherwood@arm.com>
5045 * config/aarch64/constraints.md (Upl): Rename to...
5047 * config/aarch64/aarch64.md
5048 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
5051 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5052 Alan Hayward <alan.hayward@arm.com>
5053 David Sherwood <david.sherwood@arm.com>
5055 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
5056 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
5059 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5060 Alan Hayward <alan.hayward@arm.com>
5061 David Sherwood <david.sherwood@arm.com>
5063 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
5064 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
5065 (aarch64_evpc_dup): Generate rtl direcly, rather than using
5067 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
5068 of a single element.
5069 * config/aarch64/iterators.md: Add a comment above the permute
5070 unspecs to say that they are generated directly by
5071 aarch64_expand_vec_perm_const.
5072 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
5074 2017-11-01 Nathan Sidwell <nathan@acm.org>
5076 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
5078 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5080 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
5081 and medany code models, and describe what they do.
5083 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5085 Revert accidental duplicate:
5087 * combine.c (can_change_dest_mode): Reject changes in
5088 REGMODE_NATURAL_SIZE.
5090 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
5092 PR rtl-optimization/64682
5093 PR rtl-optimization/69567
5094 PR rtl-optimization/69737
5095 PR rtl-optimization/82683
5096 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
5097 register mentioned in the note, drop the note, unless it came from I3,
5098 in which case it should go to I3 again.
5100 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5102 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
5103 and return false if not.
5104 (clear_bytes_written_by, live_bytes_read): Update accordingly.
5106 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5108 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
5109 range is known to be empty.
5111 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5112 Alan Hayward <alan.hayward@arm.com>
5113 David Sherwood <david.sherwood@arm.com>
5115 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
5116 and CONST_VECTOR_NUNITS instead of computing the number of units from
5117 the byte sizes of the vector and element.
5118 (simplify_binary_operation_1): Likewise.
5119 (simplify_const_binary_operation): Likewise.
5120 (simplify_ternary_operation): Likewise.
5122 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5123 Alan Hayward <alan.hayward@arm.com>
5124 David Sherwood <david.sherwood@arm.com>
5126 * var-tracking.c (INT_MEM_OFFSET): Replace with...
5127 (int_mem_offset): ...this new function.
5128 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
5129 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
5130 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
5133 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5134 Alan Hayward <alan.hayward@arm.com>
5135 David Sherwood <david.sherwood@arm.com>
5137 * lower-subreg.c (interesting_mode_p): New function.
5138 (compute_costs, find_decomposable_subregs, decompose_register)
5139 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
5140 (resolve_clobber, dump_choices): Use it.
5142 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5143 Alan Hayward <alan.hayward@arm.com>
5144 David Sherwood <david.sherwood@arm.com>
5146 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
5148 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5149 Alan Hayward <alan.hayward@arm.com>
5150 David Sherwood <david.sherwood@arm.com>
5152 * alias.c (find_base_value, find_base_term): Only process integer
5153 truncations. Check the precision rather than the size.
5155 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5156 Alan Hayward <alan.hayward@arm.com>
5157 David Sherwood <david.sherwood@arm.com>
5159 * machmode.h (is_narrower_int_mode): New function
5160 * optabs.c (expand_float, expand_fix): Use it.
5161 * dwarf2out.c (rotate_loc_descriptor): Likewise.
5163 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5164 Alan Hayward <alan.hayward@arm.com>
5165 David Sherwood <david.sherwood@arm.com>
5167 * rtl.h (narrower_subreg_mode): New function.
5168 * ira-color.c (update_costs_from_allocno): Use it.
5170 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5171 Alan Hayward <alan.hayward@arm.com>
5172 David Sherwood <david.sherwood@arm.com>
5174 * optabs-query.h (convert_optab_p): New function, split out from...
5175 (convert_optab_handler): ...here.
5176 (widening_optab_handler): Delete.
5177 (find_widening_optab_handler): Remove permit_non_widening parameter.
5178 (find_widening_optab_handler_and_mode): Likewise. Provide an
5179 override that operates on mode class wrappers.
5180 * optabs-query.c (widening_optab_handler): Delete.
5181 (find_widening_optab_handler_and_mode): Remove permit_non_widening
5182 parameter. Assert that the two modes are the same class and that
5183 the "from" mode is narrower than the "to" mode. Use
5184 convert_optab_handler instead of widening_optab_handler.
5185 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
5186 instead of widening_optab_handler.
5187 * expr.c (expand_expr_real_2): Update calls to
5188 find_widening_optab_handler.
5189 * optabs.c (expand_widen_pattern_expr): Likewise.
5190 (expand_binop_directly): Take the insn_code as a parameter.
5191 (expand_binop): Only call find_widening_optab_handler for
5192 conversion optabs; use optab_handler otherwise. Update calls
5193 to find_widening_optab_handler and expand_binop_directly.
5194 Use convert_optab_handler instead of widening_optab_handler.
5195 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
5196 find_widening_optab_handler and use scalar_mode rather than
5198 (convert_plusminus_to_widen): Likewise.
5200 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5201 Alan Hayward <alan.hayward@arm.com>
5202 David Sherwood <david.sherwood@arm.com>
5204 * machmode.h (fixed_size_mode): New class.
5205 * rtl.h (get_pool_mode): Return fixed_size_mode.
5206 * gengtype.c (main): Add fixed_size_mode.
5207 * target.def (get_raw_result_mode): Return a fixed_size_mode.
5208 (get_raw_arg_mode): Likewise.
5209 * doc/tm.texi: Regenerate.
5210 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
5211 * targhooks.c (default_get_reg_raw_mode): Likewise.
5212 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
5213 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
5214 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
5215 (msp430_get_raw_result_mode): Likewise.
5216 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
5217 * dbxout.c (dbxout_parms): Require fixed-size modes.
5218 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
5219 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5220 * omp-low.c (lower_oacc_reductions): Likewise.
5221 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
5222 (simplify_subreg): Update accordingly.
5223 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
5224 (force_const_mem): Update accordingly. Return NULL_RTX for modes
5225 that aren't fixed-size.
5226 (get_pool_mode): Return a fixed_size_mode.
5227 (output_constant_pool_2): Take a fixed_size_mode.
5229 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5230 Alan Hayward <alan.hayward@arm.com>
5231 David Sherwood <david.sherwood@arm.com>
5233 * doc/rtl.texi (vec_series): Document.
5234 (const): Say that the operand can be a vec_series.
5235 * rtl.def (VEC_SERIES): New rtx code.
5236 * rtl.h (const_vec_series_p_1): Declare.
5237 (const_vec_series_p): New function.
5238 * emit-rtl.h (gen_const_vec_series): Declare.
5239 (gen_vec_series): Likewise.
5240 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
5241 (gen_vec_series): Likewise.
5242 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
5243 * simplify-rtx.c (simplify_unary_operation): Handle negations
5245 (simplify_binary_operation_series): New function.
5246 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
5247 (test_vector_ops_series): New function.
5248 (test_vector_ops): Call it.
5249 * config/powerpcspe/altivec.md (altivec_lvsl): Use
5250 gen_const_vec_series.
5251 (altivec_lvsr): Likewise.
5252 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
5254 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5255 Alan Hayward <alan.hayward@arm.com>
5256 David Sherwood <david.sherwood@arm.com>
5258 * doc/rtl.texi (const): Update description of address constants.
5259 Say that vector constants are allowed too.
5260 * common.md (E, F): Use CONSTANT_P instead of checking for
5262 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
5263 checking for CONST_VECTOR.
5264 * expmed.c (make_tree): Use build_vector_from_val for a CONST
5266 * expr.c (expand_expr_real_2): Check for vector modes instead
5267 of checking for CONST_VECTOR.
5268 * rtl.h (const_vec_p): New function.
5269 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
5270 (unwrap_const_vec_duplicate): Handle them here too.
5272 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5273 David Malcolm <dmalcolm@redhat.com>
5274 Alan Hayward <alan.hayward@arm.com>
5275 David Sherwood <david.sherwood@arm.com>
5277 * rtl.h (vec_duplicate_p): New function.
5278 * selftest-rtl.c (assert_rtx_eq_at): New function.
5279 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
5280 (assert_rtx_eq_at): Declare.
5281 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
5282 * selftest-run-tests.c (selftest::run_tests): Call it.
5283 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
5284 (simplify_unary_operation_1): Recursively handle vector duplicates.
5285 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
5287 (simplify_subreg): Handle subregs of vector duplicates.
5288 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
5289 (selftest::simplify_rtx_c_tests): New functions.
5291 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5292 Alan Hayward <alan.hayward@arm.com>
5293 David Sherwood <david.sherwood@arm.com>
5295 * emit-rtl.h (gen_const_vec_duplicate): Declare.
5296 (gen_vec_duplicate): Likewise.
5297 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
5299 (gen_const_vector): ...here.
5300 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
5301 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
5302 whose elements are all equal.
5303 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
5304 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
5305 (simplify_relational_operation): Likewise.
5306 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5308 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
5309 (aarch64_expand_vector_init): Likewise.
5310 * config/arm/arm.c (neon_vdup_constant): Likewise.
5311 (neon_expand_vector_init): Likewise.
5312 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
5313 (arm_block_set_unaligned_vect): Likewise.
5314 (arm_block_set_aligned_vect): Likewise.
5315 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
5316 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
5317 (expand_vec_perm_even_odd_pack): Likewise.
5318 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
5319 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
5320 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
5321 gen_const_vec_duplicate.
5322 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
5323 * config/mips/mips.c (mips_gen_const_int_vector): Use
5324 gen_const_vec_duplicate.
5325 (mips_expand_vector_init): Use CONST0_RTX.
5326 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
5327 (define_split): Use gen_const_vec_duplicate.
5328 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
5329 (define_split): Use gen_const_vec_duplicate.
5330 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
5331 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
5332 * config/spu/spu.c (spu_const): Likewise.
5334 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5335 Alan Hayward <alan.hayward@arm.com>
5336 David Sherwood <david.sherwood@arm.com>
5338 * combine.c (can_change_dest_mode): Reject changes in
5339 REGMODE_NATURAL_SIZE.
5341 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
5343 * configure.ac (--enable-libssp): New.
5344 (gcc_cv_libc_provides_ssp): Check for explicit setting before
5345 trying to determine target-specific default. Adjust indentation.
5346 * configure: Regenerated.
5347 * doc/install.texi (Configuration): Expand --disable-libssp
5350 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
5352 config/i386/i386.c (ix86_expand_epilogue): Correct stack
5355 2017-10-31 Martin Jambor <mjambor@suse.cz>
5358 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
5360 2017-10-31 David Malcolm <dmalcolm@redhat.com>
5362 * auto-profile.c (autofdo_source_profile::read): Use
5363 UNKNOWN_LOCATION rather than 0.
5364 * diagnostic-core.h (warning_at_rich_loc): Rename to...
5365 (warning_at): ...this overload.
5366 (warning_at_rich_loc_n): Rename to...
5367 (warning_n): ...this overload.
5368 (error_at_rich_loc): Rename to...
5369 (error_at): ...this overload.
5370 (pedwarn_at_rich_loc): Rename to...
5371 (pedwarn): ...this overload.
5372 (permerror_at_rich_loc): Rename to...
5373 (permerror): ...this overload.
5374 (inform_at_rich_loc): Rename to...
5375 (inform): ...this overload.
5376 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
5377 (diagnostic_n_impl_richloc): Rename to...
5378 (diagnostic_n_impl): ...this rich_location *-based decl.
5379 (inform_at_rich_loc): Rename to...
5380 (inform): ...this, and add an assertion.
5381 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
5382 (warning_at_rich_loc): Rename to...
5383 (warning_at): ...this, and add an assertion.
5384 (warning_at_rich_loc_n): Rename to...
5385 (warning_n): ...this, and add an assertion.
5386 (warning_n): Update location_t-based implementation for removal of
5387 location_t-based diagnostic_n_impl.
5388 (pedwarn_at_rich_loc): Rename to...
5389 (pedwarn): ...this, and add an assertion.
5390 (permerror_at_rich_loc): Rename to...
5391 (permerror): ...this, and add an assertion.
5392 (error_n): Update for removal of location_t-based diagnostic_n_impl.
5393 (error_at_rich_loc): Rename to...
5394 (error_at): ...this, and add an assertion.
5395 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
5396 (driver::do_spec_on_infiles): Likewise.
5397 * substring-locations.c (format_warning_va): Update for renaming
5398 of inform_at_rich_loc.
5400 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
5402 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
5403 _Float<N>X built-in functions so that the variant without the
5404 "__builtin_" prefix is only enabled for the GNU C and Objective C
5405 languages when they are in non-strict ANSI/ISO mode.
5406 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
5407 * target.def (floatn_builtin_p): Add a target hook to control
5408 whether _Float<N> and _Float<N>X built-in functions without the
5409 "__builtin_" prefix are enabled, and return true for C and
5410 Objective C in the default hook. Include langhooks.h in
5412 * targhooks.h (default_floatn_builtin_p): Likewise.
5413 * targhooks.c (default_floatn_builtin_p): Likewise.
5414 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
5415 floatn_builtin_p target hook.
5416 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
5418 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
5419 Eric Botcazou <ebotcazou@adacore.com>
5421 PR rtl-optimization/81803
5422 * lra-constraints.c (curr_insn_transform): Also reload the whole
5423 register for a strict subreg no wider than a word if this is for
5424 a WORD_REGISTER_OPERATIONS target.
5426 2017-10-31 Jason Merrill <jason@redhat.com>
5428 * gdbinit.in: Skip over inlines from timevar.h.
5430 2017-10-31 Martin Liska <mliska@suse.cz>
5432 * doc/gcov.texi: Document new option.
5433 * gcov.c (print_usage): Likewise print it.
5434 (process_args): Support the argument.
5435 (format_count): New function.
5436 (format_gcov): Use the function.
5438 2017-10-31 Martin Liska <mliska@suse.cz>
5440 * gcov.c (struct name_map): do not use typedef.
5441 Define operator== and operator<.
5442 (name_search): Remove.
5443 (name_sort): Remove.
5444 (main): Do not allocate names.
5445 (process_file): Add vertical space.
5446 (generate_results): Use std::find.
5447 (release_structures): Do not release memory.
5448 (find_source): Use std::find.
5450 2017-10-31 Martin Liska <mliska@suse.cz>
5452 * gcov.c (struct line_info): Remove it's typedef.
5453 (line_info::line_info): Add proper ctor.
5454 (line_info::has_block): Do not use a typedef.
5455 (struct source_info): Do not use typedef.
5456 (circuit): Likewise.
5457 (get_cycles_count): Likewise.
5458 (output_intermediate_file): Iterate via vector iterator.
5459 (add_line_counts): Use std::vector methods.
5460 (accumulate_line_counts): Likewise.
5461 (output_lines): Likewise.
5463 2017-10-31 Martin Liska <mliska@suse.cz>
5465 * gcov.c (struct source_info): Remove typedef.
5466 (source_info::source_info): Add proper ctor.
5467 (accumulate_line_counts): Use struct, not it's typedef.
5468 (output_gcov_file): Likewise.
5469 (output_lines): Likewise.
5470 (main): Do not allocate an array.
5471 (output_intermediate_file): Use size of vector container.
5472 (process_file): Resize the vector.
5473 (generate_results): Do not preallocate, use newly added vector
5475 (release_structures): Do not release sources.
5476 (find_source): Use vector methods.
5477 (add_line_counts): Do not use typedef.
5479 2017-10-31 Martin Liska <mliska@suse.cz>
5481 * doc/gcov.texi: Document that.
5482 * gcov.c (add_line_counts): Mark lines with a non-executed
5484 (output_line_beginning): Handle such lines.
5485 (output_lines): Pass new argument.
5486 (output_intermediate_file): Print it in intermediate format.
5488 2017-10-31 Martin Liska <mliska@suse.cz>
5490 * color-macros.h: New file.
5491 * diagnostic-color.c: Factor out color related to macros to
5493 * doc/gcov.texi: Document -k option.
5494 * gcov.c (INCLUDE_STRING): Include string.h.
5495 (print_usage): Add -k option.
5496 (process_args): Parse it.
5497 (pad_count_string): New function.
5498 (output_line_beginning): Likewise.
5499 (DEFAULT_LINE_START): New macro.
5500 (output_lines): Support color output.
5502 2017-10-31 Martin Liska <mliska@suse.cz>
5504 PR gcov-profile/82633
5505 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
5506 their interaction with GCOV infrastructure.
5507 * configure.ac: Add -fkeep-{inline,static}-functions to
5509 * configure: Regenerate.
5511 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
5514 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
5516 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
5519 * config/rs6000/rs6000.md (allocate_stack): Force update interval
5520 into a register if it does not fit into an immediate offset field.
5522 2017-10-31 Olivier Hainque <hainque@adacore.com>
5524 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
5526 2017-10-31 Julia Koval <julia.koval@intel.com>
5528 * config.gcc: Add gfniintrin.h.
5529 * config/i386/gfniintrin.h: New.
5530 * config/i386/i386-builtin-types.def
5531 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5532 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5533 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5534 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5535 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5536 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
5537 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
5538 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
5539 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
5540 V64QI_FTYPE_V64QI_V64QI_INT): New types.
5541 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5542 * config/i386/immintrin.h: Include gfniintrin.h.
5543 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
5545 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
5547 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
5549 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5551 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
5552 (arm_ashldi3_1bit): Remove pattern.
5553 (ashrdi3): Remove shift by 1 expansion.
5554 (arm_ashrdi3_1bit): Remove pattern.
5555 (lshrdi3): Remove shift by 1 expansion.
5556 (arm_lshrdi3_1bit): Remove pattern.
5557 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
5558 cost of ashldi3 by 1.
5559 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
5560 (<shift>di3_neon): Likewise.
5562 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5564 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
5565 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
5566 and (*aarch64_simd_mov<VQ:mode>).
5567 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
5568 pattern alternative.
5569 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
5570 attributes to match pattern alternative.
5572 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
5574 * config.gcc (powerpc*-*-*): Add emmintrin.h.
5575 * config/rs6000/emmintrin.h: New file.
5576 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
5578 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5580 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
5581 * (movdi_vfp_cortexa8): Remove pattern.
5583 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5585 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
5587 (Specific, alpha*-dec-osf5.1): Remove.
5588 (Specific, mips-sgi-irix5): Remove.
5589 (Specific, mips-sgi-irix6): Remove.
5591 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5594 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
5595 arguments to clear_bit_region_be.
5597 2017-10-30 Jim Wilson <wilson@tuliptree.org>
5599 * gimplify.c: Include memmodel.h.
5601 2017-10-30 Martin Jambor <mjambor@suse.cz>
5603 * omp-grid.c (grid_attempt_target_gridification): Also insert a
5604 condition whether loop should be executed at all.
5606 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
5608 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5609 gimple folding of vec_madd() intrinsics.
5610 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
5611 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
5612 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
5614 2017-10-30 Richard Biener <rguenther@suse.de>
5616 PR tree-optimization/82762
5618 2017-10-23 Richard Biener <rguenther@suse.de>
5620 PR tree-optimization/82129
5622 2017-08-01 Richard Biener <rguenther@suse.de>
5624 PR tree-optimization/81181
5625 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5626 (compute_antic): ... end of iteration here.
5628 2017-10-30 Joseph Myers <joseph@codesourcery.com>
5630 * doc/invoke.texi (C Dialect Options): Document -std=c17,
5631 -std=iso9899:2017 and -std=gnu17.
5632 * doc/standards.texi (C Language): Document C17 support.
5633 * doc/cpp.texi (Overview): Mention -std=c17.
5634 (Standard Predefined Macros): Document C11 and C17 values of
5635 __STDC_VERSION__. Do not refer to C99 support as incomplete.
5636 * doc/extend.texi (Inline): Do not list individual options for
5637 standards newer than C99.
5638 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
5640 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
5643 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5645 * asan.c (asan_finish_file): Align asan globals array by shadow
5648 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5651 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
5652 (struct store_immediate_info): Add bitregion_start and bitregion_end
5654 (store_immediate_info::store_immediate_info): Add brs and bre
5655 arguments and initialize bitregion_{start,end} from those.
5656 (struct merged_store_group): Add bitregion_start, bitregion_end,
5657 align_base and mask fields. Drop unnecessary struct keyword from
5658 struct store_immediate_info. Add do_merge method.
5659 (clear_bit_region_be): Use memset instead of loop storing zeros.
5660 (merged_store_group::do_merge): New method.
5661 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
5662 stores as long as the surrounding bitregions have no gaps.
5663 (merged_store_group::merge_overlapping): Use do_merge.
5664 (merged_store_group::apply_stores): Test that bitregion_{start,end}
5665 is byte aligned, rather than requiring that start and width are
5666 byte aligned. Drop unnecessary struct keyword from
5667 struct store_immediate_info. Allocate and populate also mask array.
5668 Make start of the arrays relative to bitregion_start rather than
5669 start and size them according to bitregion_{end,start} difference.
5670 (struct imm_store_chain_info): Drop unnecessary struct keyword from
5671 struct store_immediate_info.
5672 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
5673 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
5674 struct keyword from struct store_immediate_info.
5675 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
5676 between stores as long as the surrounding bitregions have no gaps.
5678 (struct split_store): Add orig non-static data member.
5679 (split_store::split_store): Initialize orig to false.
5680 (find_constituent_stmts): Return store_immediate_info *, non-NULL
5681 if there is exactly a single original stmt. Change stmts argument
5682 to pointer from reference, if NULL, don't push anything to it. Add
5683 first argument, use it to optimize skipping over orig stmts that
5684 are known to be before bitpos already. Simplify.
5685 (split_group): Return unsigned int count how many stores are or
5686 would be needed rather than a bool. Add allow_unaligned argument.
5687 Change split_stores argument from reference to pointer, if NULL,
5688 only do a dry run computing how many stores would be produced.
5689 Rewritten algorithm to use both alignment and misalign if
5690 !allow_unaligned and handle bitfield stores with gaps.
5691 (imm_store_chain_info::output_merged_store): Set start_byte_pos
5692 from bitregion_start instead of start. Compute allow_unaligned
5693 here, if true, do 2 split_group dry runs to compute which one
5694 produces fewer stores and prefer aligned if equal. Punt if
5695 new count is bigger or equal than original before emitting any
5696 statements, rather than during that. Remove no longer needed
5697 new_ssa_names tracking. Replace num_stmts with
5698 split_stores.length (). Use 32-bit stack allocated entries
5699 in split_stores auto_vec. Try to reuse original store lhs/rhs1
5700 if possible. Handle bitfields with gaps.
5701 (pass_store_merging::execute): Ignore bitsize == 0 stores.
5702 Compute bitregion_{start,end} for the stores and construct
5703 store_immediate_info with that. Formatting fixes.
5705 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
5708 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
5709 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
5711 2017-10-29 Jim Wilson <wilson@tuliptree.org>
5713 * gimplify.c: Include tm_p.h.
5715 * common.opt (gcoff): Re-add as ignored option.
5716 (gcoff1, gcoff2, gcoff3): Likewise.
5718 * Makefile.in (OBJS): Delete sdbout.o.
5719 (GTFILES): Delete $(srcdir)/sdbout.c.
5720 * debug.h: Delete sdb_debug_hooks.
5721 * final.c: Delete sdbout.h include.
5722 (final_scan_insn): Delete SDB_DEBUG check.
5723 (rest_of_clean_state): Likewise.
5724 * output.h: Delete sdb_begin_function_line.
5727 * toplev.c: Delete sdbout.h include.
5728 (process_options): Delete SDB_DEBUG check.
5729 * tree-core.h (tree_type_common): Delete pointer field of
5731 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
5732 TYPE_SYMTAB_POINTER.
5733 * tree.h (TYPE_SYMTAB_POINTER): Delete.
5734 (TYPE_SYMTAB_IS_POINTER): Delete.
5735 (TYPE_SYMTAB_IS_DIE): Renumber.
5736 * xcoffout.c: Refer to former sdbout.c file.
5737 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
5739 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
5740 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
5741 (Debugging Options): Delete -gcoff.
5742 (-gstabs): Delete SDB reference.
5744 (-gcoff@var{level}): Delete.
5745 * doc/passes.texi (Debugging information output): Delete SDB and
5746 sdbout.c references.
5747 * doc/tm.texi: Regenerate.
5748 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
5749 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
5751 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
5752 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
5753 (SDB_DEBUGGING_INFO): Delete.
5754 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
5755 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
5756 * target.def (output_source_filename): Delete COFF reference.
5758 * common.opt (gcoff): Delete.
5759 (gxcoff+): Update Negative chain.
5760 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
5762 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
5763 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
5764 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
5766 (expand_function_start): Change sdb reference to past tense.
5767 (expand_function_end): Change sdb reference to past tense.
5768 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
5769 * opts.c (debug_type_names): Delete coff entry.
5770 (common_handle_option): Delete OPT_gcoff case.
5771 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
5773 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
5774 * config/cris/cris.h: Delete SDB reference in comment.
5775 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
5776 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
5777 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
5778 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
5780 (ix86_expand_prologue): Likewise.
5781 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
5782 * config/ia64/ia64.h: Likewise.
5783 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
5784 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
5786 * config/mmix/mmix.h: Likewise.
5787 * config/nds32/nds32.c: Likewise.
5788 * config/stormy/storym16.h: Likewise.
5789 * config/visium/visium.h: Likewise.
5790 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
5792 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
5794 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
5795 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
5796 FRAME_POINTER_REGNUM point at high end of local var area.
5798 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
5800 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
5801 Move comment around. Do not reset best_edge for a copiable
5802 destination if the copy would cause a partition change.
5803 (better_edge_p): Remove redundant check.
5805 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5807 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
5809 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5811 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
5812 for math functions that have _Float<N> and _Float<N>X variants.
5813 (mathfn_built_in_2): Add support for math functions that have
5814 _Float<N> and _Float<N>X variants.
5815 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
5816 (expand_builtin_mathfn_ternary): Add support for fma with
5817 _Float<N> and _Float<N>X variants.
5818 (expand_builtin): Likewise.
5819 (fold_builtin_3): Likewise.
5820 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
5821 create math function _Float<N> and _Float<N>X variants as external
5823 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
5824 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
5825 the __builtin_ prefix and if not strict ansi, without the prefix.
5826 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
5827 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
5828 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
5829 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
5830 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
5831 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
5832 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
5833 function signatures for fma _Float<N> and _Float<N>X variants.
5834 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
5835 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
5836 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
5837 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
5838 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
5839 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
5840 * gencfn-macros.c (print_case_cfn): Add support for math functions
5841 that have _Float<N> and _Float<N>X variants.
5842 (print_define_operator_list): Likewise.
5843 (fltfn_suffixes): Likewise.
5845 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
5846 for math functions that have _Float<N> and _Float<N>X variants.
5847 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
5848 and _Float<N>X variants.
5849 (COPYSIGN): Likewise.
5852 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
5853 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
5855 (integer_valued_read_call_p): Likewise.
5856 * fold-const-call.c (fold_const_call_ss): Likewise.
5857 (fold_const_call_sss): Add support for copysign, fmin, and fmax
5858 _Float<N> and _Float<N>X variants.
5859 (fold_const_call_ssss): Add support for fma _Float<N> and
5860 _Float<N>X variants.
5861 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
5862 support for copysign and fma _Float<N> and _Float<N>X variants.
5863 (backprop::process_builtin_call_use): Likewise.
5864 * tree-call-cdce.c (can_test_argument_range); Add support for
5865 sqrt _Float<N> and _Float<N>X variants.
5866 (edom_only_function): Likewise.
5867 (get_no_error_domain): Likewise.
5868 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
5869 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
5870 copysign _Float<N> and _Float<N>X variants.
5871 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
5872 handled by machine independent code.
5873 (FMAF128): Likewise.
5874 * doc/cpp.texi (Common Predefined Macros): Document defining
5875 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
5876 fma _Float<N> and _Float<N>X variants.
5878 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5881 * config/i386/i386-modes.def (CCFPU): Remove definition.
5882 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
5883 (ix86_cc_modes_compatible): Ditto.
5884 (ix86_expand_carry_flag_compare): Ditto.
5885 (ix86_expand_int_movcc): Ditto.
5886 (ix86_expand_int_addcc): Ditto.
5887 (ix86_reverse_condition): Ditto.
5888 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
5889 Return true/false for unordered/ordered fp comparisons.
5890 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
5891 (ix86_prepare_fp_compare_args): Update for rename.
5892 (ix86_expand_fp_compare): Update for rename. Generate unordered
5893 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
5894 (ix86_expand_sse_compare_and_jump): Ditto.
5895 * config/i386/predicates.md (fcmov_comparison_operator):
5896 Remove CCFPU mode handling.
5897 (ix86_comparison_operator): Ditto.
5898 (ix86_carry_flag_operator): Ditto.
5899 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
5900 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
5901 (*cmpu<mode>_cc_i387): Ditto.
5902 (FPCMP): Remove mode iterator.
5903 (unord): Remove mode attribute.
5904 (unord_subst): New define_subst transformation
5905 (unord): New define_subst attribute.
5907 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
5908 (*cmpi<unord>xf_i387): Ditto.
5909 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
5910 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
5911 using unord_subst transformation.
5912 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
5913 (round_saeonly): Also handle CCFP mode.
5914 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
5915 Remove UNSPEC_SAHF unspec handling.
5917 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5919 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
5921 2017-10-27 Jeff Law <law@redhat.com>
5923 * gimple-ssa-sprintf.c: Include domwalk.h.
5924 (class sprintf_dom_walker): New class, derived from dom_walker.
5925 (sprintf_dom_walker::before_dom_children): New function.
5926 (struct call_info): Moved into sprintf_dom_walker class
5927 (compute_formath_length, handle_gimple_call): Likewise.
5928 (sprintf_length::execute): Call the dominator walker rather
5929 than walking the statements.
5931 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
5932 gimple statement locations.
5933 (check_array_bounds): Corresponding changes. Get the statement's
5934 location directly from wi->stmt.
5936 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
5939 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
5941 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5943 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
5944 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
5946 2017-10-27 Jakub Jelinek <jakub@redhat.com>
5949 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
5950 * config/i386/i386.c (maybe_get_pool_constant): Removed.
5951 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
5952 maybe_get_pool_constant.
5953 * config/i386/predicates.md (zero_extended_scalar_load_operand):
5956 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5958 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
5959 2.26 caveat. Update gas and gld versions.
5960 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
5963 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5965 * cgraph.h (set_malloc_flag): Declare.
5966 * cgraph.c (set_malloc_flag_1): New function.
5967 (set_malloc_flag): Likewise.
5968 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
5969 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
5971 (read_ipa_call_summary): Add support for reading is_return_callee.
5972 (write_ipa_call_summary): Stream is_return_callee.
5973 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
5974 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
5975 ipa-prop.h, ipa-fnsummary.h.
5976 (pure_const_names): Change to static.
5977 (malloc_state_e): Define.
5978 (malloc_state_names): Define.
5979 (funct_state_d): Add field malloc_state.
5980 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
5981 (check_retval_uses): New function.
5982 (malloc_candidate_p): Likewise.
5983 (analyze_function): Add support for malloc attribute.
5984 (pure_const_write_summary): Stream malloc_state.
5985 (pure_const_read_summary): Add support for reading malloc_state.
5986 (dump_malloc_lattice): New function.
5987 (propagate_malloc): New function.
5988 (warn_function_malloc): New function.
5989 (ipa_pure_const::execute): Call propagate_malloc and
5990 ipa_free_fn_summary.
5991 (pass_local_pure_const::execute): Add support for malloc attribute.
5992 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
5993 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
5995 2017-10-27 Martin Liska <mliska@suse.cz>
5997 PR gcov-profile/82457
5998 * doc/invoke.texi: Document that one needs a non-strict ISO mode
5999 for fork-like functions to be properly instrumented.
6001 2017-10-27 Richard Biener <rguenther@suse.de>
6004 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6005 info when we redirected EH.
6007 2017-10-26 Michael Collison <michael.collison@arm.com>
6009 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6011 (<optab>_trunchf<GPI:mode>2: New pattern.
6012 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6013 * config/aarch64/iterators.md (wv): New mode attribute.
6014 (vf, VF): New mode attributes.
6015 (vgp, VGP): New mode attributes.
6016 (s): Update attribute with SImode and DImode prefixes.
6018 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6020 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
6021 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
6022 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
6023 (nios2_option_overide): Initialize it. Don't allow R0-relative
6024 addressing with PIC.
6025 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
6026 (nios2_symbolic_constant_p): Likewise.
6027 (nios2_legitimate_address_p): Likewise.
6028 (nios2_r0rel_section_name_p): New.
6029 (nios2_symbol_ref_in_r0rel_data_p): New.
6030 (nios2_emit_move_sequence): Handle r0rel_constant_p.
6031 (r0rel_constant_p): New.
6032 (nios2_print_operand_address): Handle r0rel_constant_p.
6033 (nios2_cdx_narrow_form_p): Likewise.
6034 * config/nios2/nios2.opt (mr0rel-sec=): New option.
6035 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
6036 (Nios II Options): Document -mr0rel-sec.
6038 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6040 * config/nios2/nios2.c: Include xregex.h.
6041 (nios2_gprel_sec_regex): New.
6042 (nios2_option_overide): Initialize it. Don't allow GP-relative
6043 addressing with PIC.
6044 (nios2_small_section_name_p): Check for regex match.
6045 * config/nios2/nios2.opt (mgprel-sec=): New option.
6046 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
6047 (Nios II Options): Document -mgprel-sec.
6049 2017-10-26 Jim Wilson <wilson@tuliptree.org>
6051 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
6053 2017-10-26 Tom de Vries <tom@codesourcery.com>
6055 PR tree-optimization/82707
6056 * gimple.c (gimple_copy): Fix unsharing of
6057 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
6059 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
6061 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
6062 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
6063 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
6064 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
6065 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
6066 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
6067 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
6068 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
6069 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
6070 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
6071 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
6072 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
6073 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
6074 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
6075 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
6076 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
6077 _mm512_mask_cmpunord_ps_mask): New intrinsics.
6079 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6081 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
6083 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
6084 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
6085 warning to rs6000.c. Remove the Undocumented flag, since it has
6087 (-mabi=ibmlongdouble): Likewise.
6088 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
6089 already set, set the default format for long double.
6090 (rs6000_debug_reg_global): Print whether long double is IBM or
6092 (rs6000_option_override_internal): Rework setting long double
6093 format. Only warn if the user is changing the long double default
6094 and they did not use -Wno-psabi.
6095 * doc/invoke.texi (PowerPC options): Update the documentation for
6096 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
6098 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6099 Alan Hayward <alan.hayward@arm.com>
6100 David Sherwood <david.sherwood@arm.com>
6102 * rtl.h (wider_subreg_mode): New function.
6103 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
6104 rather than an unsigned int *.
6105 * ira-color.c (regno_max_ref_width): Replace with...
6106 (regno_max_ref_mode): ...this new variable.
6107 (coalesced_pseudo_reg_slot_compare): Update accordingly.
6108 Use wider_subreg_mode.
6109 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
6110 rather than an unsigned int *.
6111 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
6112 (process_alt_operands): Likewise.
6113 (invariant_p): Likewise.
6114 * lra-spills.c (assign_mem_slot): Likewise.
6115 (add_pseudo_to_slot): Likewise.
6116 * lra.c (collect_non_operand_hard_regs): Likewise.
6117 (add_regs_to_insn_regno_info): Likewise.
6118 * reload1.c (regno_max_ref_width): Replace with...
6119 (regno_max_ref_mode): ...this new variable.
6120 (reload): Update accordingly. Update call to
6121 ira_sort_regnos_for_alter_reg.
6122 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
6123 (init_eliminable_invariants): Update to use regno_max_ref_mode.
6124 (scan_paradoxical_subregs): Likewise.
6126 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6128 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
6129 (aarch64_frame): Add emit_frame_chain boolean.
6130 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6131 Move eh_return case to aarch64_layout_frame.
6132 (aarch64_layout_frame): Initialize emit_frame_chain.
6133 (aarch64_expand_prologue): Use emit_frame_chain.
6135 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6137 * config/aarch64/aarch64.c (aarch64_layout_frame):
6138 Ensure LR is always stored at the bottom of the callee-saves.
6139 Remove rarely used frame layout which saves callee-saves at top of
6140 frame, so the store of LR can be used as a valid probe in all cases.
6142 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6144 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
6145 Improve unaligned TImode/TFmode base/offset split.
6147 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6148 Alan Hayward <alan.hayward@arm.com>
6149 David Sherwood <david.sherwood@arm.com>
6151 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
6152 * combine.c (find_single_use_1): Likewise.
6153 (expand_field_assignment): Likewise.
6154 (move_deaths): Likewise.
6155 * lra-constraints.c (simplify_operand_subreg): Likewise.
6156 (curr_insn_transform): Likewise.
6157 * lra.c (collect_non_operand_hard_regs): Likewise.
6158 (add_regs_to_insn_regno_info): Likewise.
6159 * rtlanal.c (reg_referenced_p): Likewise.
6160 (covers_regno_no_parallel_p): Likewise.
6162 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6164 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
6165 Don't print any bits outside the precision of the value.
6166 * wide-int.cc (test_printing): Add some new tests.
6168 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6170 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
6171 supports -xbrace_comment option.
6172 * configure: Regenerate.
6173 * config.in: Regenerate.
6174 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
6175 (ASM_CPU_SPEC): Use it.
6177 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6179 * target.def (static_rtx_alignment): New hook.
6180 * targhooks.h (default_static_rtx_alignment): Declare.
6181 * targhooks.c (default_static_rtx_alignment): New function.
6182 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
6183 * doc/tm.texi: Regenerate.
6184 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
6185 instead of targetm.constant_alignment. Remove call to
6187 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6188 (cris_preferred_mininum_alignment): New function, split out from...
6189 (cris_constant_alignment): ...here.
6190 (cris_static_rtx_alignment): New function.
6191 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
6193 (ix86_constant_alignment): ...here.
6194 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6195 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6196 (mmix_static_rtx_alignment): New function.
6197 * config/spu/spu.c (spu_static_rtx_alignment): New function.
6198 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6200 2017-10-26 Tamar Christina <tamar.christina@arm.com>
6203 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
6204 Add flag_trapping_math and flag_fp_int_builtin_inexact.
6206 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
6208 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
6209 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
6210 mark as a sign-extending load.
6211 (local_pic_load_u): Define.
6213 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
6216 * fold-const.c (operand_equal_for_comparison_p): Also return true
6217 if ARG0 is a simple variant of ARG1 with narrower precision.
6218 (fold_ternary_loc): Always pass unstripped operands to the predicate.
6220 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
6222 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
6224 * i386.h (processor_costs): Add gather_static, gather_per_elt,
6225 scatter_static, scatter_per_elt.
6226 * x86-tune-costs.h: Add new cost entries.
6228 2017-10-25 Richard Biener <rguenther@suse.de>
6230 * tree-ssa-sccvn.h (vn_eliminate): Declare.
6231 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
6232 class pass_fre): Move to ...
6233 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
6234 class pass_fre): ... here and adjust for statistics.
6236 2017-10-25 Jakub Jelinek <jakub@redhat.com>
6239 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
6240 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
6241 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
6242 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
6245 2017-10-25 Richard Biener <rguenther@suse.de>
6247 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
6248 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
6249 eliminate_push_avail, eliminate_insert): Move inside...
6250 (class eliminate_dom_walker): ... this class in preparation
6252 (fini_eliminate): Remove by merging with ...
6253 (eliminate): ... this function. Adjust for class changes.
6254 (pass_pre::execute): Remove fini_eliminate call.
6255 (pass_fre::execute): Likewise.
6257 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6260 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
6261 (VPERMI2, VPERMI2I): New mode iterators.
6262 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
6263 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
6265 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
6266 mode iterator. Remove 3 old define_insn patterns.
6267 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
6268 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
6269 VPERMI2 mode iterator, remove the other two expanders.
6270 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
6271 to use VPERMI2 mode iterator, add another alternative for vpermi2*
6272 instructions, remove the other two patterns.
6273 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
6274 mode iterator, remove the other two patterns.
6275 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
6276 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
6277 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
6278 and adjust argument order accordingly.
6279 (ix86_expand_vec_perm): Adjust caller.
6280 (expand_vec_perm_1): Likewise.
6281 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
6282 (expand_vec_perm_vpermt2_vpshub2): ... this.
6283 (ix86_expand_vec_perm_const_1): Adjust caller.
6284 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
6287 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
6288 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
6289 (vec_shl_<mode>): Remove unused expander.
6290 (avx512bw_<shift_insn><mode>3): New define_insn.
6291 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
6292 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
6294 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
6297 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
6300 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6302 PR rtl-optimization/82396
6303 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
6304 (autopref_multipass_init): Simplify initialization.
6305 (autopref_rank_data): Simplify sort order.
6306 * gcc/sched-int.h (autopref_multipass_data_): Remove
6307 multi_mem_insn_p, min_offset and max_offset.
6309 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6312 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6313 Check special value of flag_omit_frame_pointer.
6314 (aarch64_can_eliminate): Likewise.
6315 (aarch64_override_options_after_change_1): Simplify handling of
6316 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
6318 2017-10-24 Richard Biener <rguenther@suse.de>
6320 PR tree-optimization/82697
6321 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
6322 zero for conditional load and unconditional store.
6324 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6326 * doc/install.texi: Document bootstrap-cet.
6328 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6331 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
6332 ENDBR instruction at function entrance if function is only
6335 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6338 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
6339 patterns to better describe from which operation the CF is computed.
6340 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
6341 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
6342 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
6343 is 0, use _0 suffixed expanders instead of emitting a comparison
6346 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
6348 * config/i386/i386.md(*movsf_internal, *movdf_internal):
6349 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
6351 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
6354 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
6355 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
6356 * loop-iv.c (iv_get_reaching_def): Likewise.
6357 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
6358 variable is promoted and the partition contains undefined values.
6360 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6362 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
6364 (nios2_address_cost): Define.
6365 (nios2_legitimize_address): Recognize (exp + constant) directly.
6366 (TARGET_ADDRESS_COST): Define.
6368 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6370 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
6371 (nios2_symbolic_memory_operand_p): Declare.
6372 (nios2_split_large_constant): Declare.
6373 (nios2_split_symbolic_memory_operand): Declare.
6374 * config/nios2/nios2.c: Adjust includes.
6375 (nios2_symbolic_constant_allowed): New.
6376 (nios2_symbolic_constant_p): New.
6377 (nios2_plus_symbolic_constant_p): New.
6378 (nios2_valid_addr_expr_p): Recognize addresses involving
6380 (nios2_legitimate_address_p): Likewise, also LO_SUM.
6381 (nios2_symbolic_memory_operand_p): New.
6382 (nios2_large_constant_p): New.
6383 (nios2_split_large_constant): New.
6384 (nios2_split_plus_large_constant): New.
6385 (nios2_split_symbolic_memory_operand): New.
6386 (nios2_legitimize_address): Code refactoring. Handle addresses
6387 involving symbolic constants.
6388 (nios2_emit_move_sequence): Likewise.
6389 (nios2_print_operand): Improve error output.
6390 (nios2_print_operand_address): Handle LO_SUM.
6391 (nios2_cdx_narrow_form_p): Likewise.
6392 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
6393 operands involving symbolic constants.
6394 (movhi_internal, movsi_internal): Likewise.
6395 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
6396 (extendhisi2, extendqi<mode>2): Likewise.
6398 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6400 * tree-pass.h (PROP_rtl_split_insns): Define.
6401 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
6403 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6405 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
6407 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6410 * target.def (const_not_ok_for_debug_p): Default to
6411 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
6412 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
6413 * targhooks.c (default_const_not_ok_for_debug_p): New function.
6414 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
6415 which targetm.const_not_ok_for_debug_p returned true.
6416 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
6418 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
6420 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
6421 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
6422 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
6424 (ix86_const_not_ok_for_debug_p): New function.
6425 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
6426 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
6428 2017-10-23 David Malcolm <dmalcolm@redhat.com>
6431 * system.h: Conditionally include "unique-ptr.h" if
6432 INCLUDE_UNIQUE_PTR is defined.
6433 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
6434 of defining INCLUDE_UNIQUE_PTR before including "system.h".
6436 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
6438 * config/rl78/rl78.md: New define_expand "subdi3".
6440 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
6443 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
6444 DF_REF_INSN if DF_REF_INSN_INFO is false.
6446 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
6448 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
6449 xmm_move instead of sse_move.
6450 (sse_store_index): New function.
6451 (ix86_register_move_cost): Be more sensible about mismatch stall;
6452 model AVX moves correctly; make difference between sse->integer and
6454 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
6455 moves; make difference between SSE and AVX.
6456 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
6457 and zmm_move. Increase size of sse load and store tables;
6458 add unaligned load and store tables; add ssemmx_to_integer.
6459 * x86-tune-costs.h: Update all entries according to real
6460 move latencies from Agner Fog's manual and chip documentation.
6462 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6465 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
6466 * config/i386/constraints.md (Wf): New constraint.
6467 * config/i386/i386.md (UNSPEC_SBB): New unspec.
6468 (cmp<dwi>_doubleword): Removed.
6469 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
6470 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
6471 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
6472 expand with cmp<dwi>_doubleword. For LTU and GEU use
6473 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
6475 * common.opt (gcolumn-info): Enable by default.
6476 * doc/invoke.texi (gcolumn-info): Document new default.
6478 2017-10-23 Richard Biener <rguenther@suse.de>
6480 PR tree-optimization/82672
6481 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
6482 Fold the stmt if we propagated into it.
6484 2017-10-23 Richard Biener <rguenther@suse.de>
6486 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
6487 (bitmap_remove_expr_from_set): ... this. All callers call this
6488 for non-constant values.
6489 (bitmap_set_subtract): Rename to...
6490 (bitmap_set_subtract_expressions): ... this. Adjust and
6492 (bitmap_set_contains_value): Remove superfluous check.
6493 (bitmap_set_replace_value): Inline into single caller ...
6494 (bitmap_value_replace_in_set): ... here and simplify.
6495 (dependent_clean): Merge into ...
6496 (clean): ... this using an overload. Adjust.
6497 (prune_clobbered_mems): Adjust.
6498 (compute_antic_aux): Likewise.
6499 (compute_partial_antic_aux): Likewise.
6501 2017-10-23 Richard Biener <rguenther@suse.de>
6503 PR tree-optimization/82129
6505 2017-08-01 Richard Biener <rguenther@suse.de>
6507 PR tree-optimization/81181
6508 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6509 (compute_antic): ... end of iteration here.
6511 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6513 * target.def (starting_frame_offset): New hook.
6514 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
6515 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
6516 * doc/tm.texi.in: Regenerate.
6517 * hooks.h (hook_hwi_void_0): Declare.
6518 * hooks.c (hook_hwi_void_0): New function.
6519 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
6520 STARTING_FRAME_OFFSET.
6521 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
6522 * reload1.c (reload): Likewise.
6523 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
6524 instead of STARTING_FRAME_OFFSET.
6525 * function.c (try_fit_stack_local): Likewise.
6526 (assign_stack_local_1): Likewise
6527 (instantiate_virtual_regs): Likewise.
6528 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
6529 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
6530 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
6531 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
6532 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
6533 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
6534 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
6535 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
6536 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
6537 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
6538 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
6539 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
6540 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
6541 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
6542 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
6543 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
6544 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
6545 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
6546 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
6547 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
6548 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
6549 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
6550 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
6551 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
6552 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
6553 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
6554 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
6555 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
6556 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
6557 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
6558 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
6559 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
6560 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
6561 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
6562 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
6563 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
6564 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
6565 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
6566 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
6567 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
6568 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
6569 * config/avr/avr.c (avr_starting_frame_offset): Make static and
6570 return a HOST_WIDE_INT.
6571 (avr_builtin_setjmp_frame_value): Use it instead of
6572 STARTING_FRAME_OFFSET.
6573 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6574 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
6575 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
6577 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6578 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
6579 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
6580 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6581 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
6582 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
6583 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6584 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
6585 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
6586 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6587 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
6588 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
6590 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6591 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
6592 * config/mips/mips.c (mips_compute_frame_info): Refer to
6593 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6594 (mips_starting_frame_offset): New function.
6595 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6596 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
6597 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
6598 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
6599 and return a HOST_WIDE_INT.
6600 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6601 (mmix_initial_elimination_offset): Refer to
6602 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6603 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
6604 * config/pa/pa.c (pa_starting_frame_offset): New function.
6605 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
6606 (pa_expand_prologue): Likewise.
6607 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6608 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
6609 !FRAME_GROWS_DOWNWARD handling to...
6610 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6611 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
6612 !FRAME_GROWS_DOWNWARD handling to...
6613 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6614 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
6615 !FRAME_GROWS_DOWNWARD handling to...
6616 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6617 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
6619 (rs6000_starting_frame_offset): New function.
6620 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
6621 !FRAME_GROWS_DOWNWARD handling to...
6622 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6623 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
6624 !FRAME_GROWS_DOWNWARD handling to...
6625 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6626 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
6627 !FRAME_GROWS_DOWNWARD handling to...
6628 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6629 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
6630 (rs6000_starting_frame_offset): New function.
6631 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
6632 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
6633 * config/vax/vax.c (vax_starting_frame_offset): New function.
6634 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
6635 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6636 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
6637 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
6638 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6639 * system.h (STARTING_FRAME_OFFSET): Poison.
6641 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6643 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
6644 SCALAR_TYPE_MODE instead of TYPE_MODE.
6646 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6647 Alan Hayward <alan.hayward@arm.com>
6648 David Sherwood <david.sherwood@arm.com>
6650 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
6652 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6653 Alan Hayward <alan.hayward@arm.com>
6654 David Sherwood <david.sherwood@arm.com>
6656 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
6658 2017-10-23 Richard Biener <rguenther@suse.de>
6660 PR tree-optimization/82129
6661 * tree-ssa-pre.c (bitmap_set_and): Remove.
6662 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
6663 canonicalizing expressions in the set to those with lowest
6664 ID rather than taking that from the first edge.
6666 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6668 * combine.c (rtx_equal_for_field_assignment_p): Use
6669 byte_lowpart_offset.
6671 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6672 Alan Hayward <alan.hayward@arm.com>
6673 David Sherwood <david.sherwood@arm.com>
6675 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
6676 to a SUBREG_PROMOTED_VAR.
6678 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6679 Alan Hayward <alan.hayward@arm.com>
6680 David Sherwood <david.sherwood@arm.com>
6682 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
6683 (expand_debug_source_expr): Likewise.
6684 * combine.c (combine_simplify_rtx): Likewise.
6685 * cse.c (fold_rtx): Likewise.
6686 * optabs.c (expand_float): Likewise.
6687 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6688 (simplify_binary_operation_1): Likewise.
6690 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6691 Alan Hayward <alan.hayward@arm.com>
6692 David Sherwood <david.sherwood@arm.com>
6694 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
6695 (record_promoted_value): Likewise.
6696 * expr.c (expand_expr_real_2): Likewise.
6697 * ree.c (update_reg_equal_equiv_notes): Likewise.
6698 (combine_set_extension): Likewise.
6699 * rtlanal.c (low_bitmask_len): Likewise.
6700 * simplify-rtx.c (neg_const_int): Likewise.
6701 (simplify_binary_operation_1): Likewise.
6703 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6704 Alan Hayward <alan.hayward@arm.com>
6705 David Sherwood <david.sherwood@arm.com>
6707 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
6708 * regcprop.c (maybe_mode_change): Likewise.
6709 * reload1.c (alter_reg): Likewise.
6711 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6713 * inchash.h (inchash::hash::add_wide_int): New function.
6714 * lto-streamer-out.c (hash_tree): Use it.
6716 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6718 * inchash.h (inchash::hash::add_wide_int): Rename to...
6719 (inchash::hash::add_hwi): ...this.
6720 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
6721 (polymorphic_call_target_hasher::hash): Likewise.
6722 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
6723 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
6724 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
6725 * lto-streamer-out.c (hash_tree): Likewise.
6726 * optc-save-gen.awk: Likewise.
6727 * tree.c (add_expr): Likewise.
6729 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6732 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
6733 for ordered inequality comparisons even with TARGET_IEEE_FP.
6735 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6738 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
6739 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6740 Expand with cmp<dwi>_doubleword.
6742 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6744 * extend.texi: Add x86 specific to 'nocf_check' attribute.
6745 List CET intrinsics.
6746 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
6747 specific to -fcf-protection option.
6749 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6751 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
6752 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
6753 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
6754 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
6755 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
6756 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
6757 (extra_objs): Add cet.o for Linux/x86 targets.
6758 (tmake_file): Add i386/t-cet for Linux/x86 targets.
6759 * config/i386/cet.c: New file.
6760 * config/i386/cetintrin.h: Likewise.
6761 * config/i386/t-cet: Likewise.
6762 * config/i386/cpuid.h (bit_SHSTK): New.
6763 (bit_IBT): Likewise.
6764 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
6765 pass IBT and SHSTK bits.
6766 * config/i386/i386-builtin-types.def
6767 (VOID_FTYPE_UNSIGNED_PVOID): New.
6768 (VOID_FTYPE_UINT64_PVOID): Likewise.
6769 * config/i386/i386-builtin.def: Add CET intrinsics.
6770 * config/i386/i386-c.c (ix86_target_macros_internal): Add
6771 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
6772 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
6773 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
6775 * config/i386/i386.c (rest_of_insert_endbranch): New.
6776 (pass_data_insert_endbranch): Likewise.
6777 (pass_insert_endbranch): Likewise.
6778 (make_pass_insert_endbranch): Likewise.
6779 (ix86_notrack_prefixed_insn_p): Likewise.
6780 (ix86_target_string): Add -mibt, -mshstk flags.
6781 (ix86_option_override_internal): Add flag_cf_protection
6783 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
6784 (ix86_print_operand): Add 'notrack' prefix output.
6785 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
6786 (ix86_expand_builtin): Expand CET intrinsics.
6787 (x86_output_mi_thunk): Add 'endbranch' instruction.
6788 * config/i386/i386.h (TARGET_IBT): New.
6789 (TARGET_IBT_P): Likewise.
6790 (TARGET_SHSTK): Likewise.
6791 (TARGET_SHSTK_P): Likewise.
6792 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
6793 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
6794 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
6795 (builtin_setjmp_setup): New pattern.
6796 (builtin_longjmp): Likewise.
6797 (rdssp<mode>): Likewise.
6798 (incssp<mode>): Likewise.
6799 (saveprevssp): Likewise.
6800 (rstorssp): Likewise.
6801 (wrss<mode>): Likewise.
6802 (wruss<mode>): Likewise.
6803 (setssbsy): Likewise.
6804 (clrssbsy): Likewise.
6805 (nop_endbr): Likewise.
6806 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
6808 * config/i386/immintrin.h: Include <cetintrin.h>.
6809 * config/i386/linux-common.h
6810 (file_end_indicate_exec_stack_and_cet): New prototype.
6811 (TARGET_ASM_FILE_END): New.
6813 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6815 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
6816 latencies instead of having separate table; make difference between
6817 integer and float costs.
6818 * i386.h (processor_costs): Remove scalar_stmt_cost,
6819 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
6820 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
6822 * x86-tune-costs.h: Remove entries which has been removed in
6823 procesor_costs from all tables; make cond_taken_branch_cost
6824 and cond_not_taken_branch_cost COST_N_INSNS based.
6826 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6828 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
6830 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6832 * config/i386/i386.md (isa): Remove fma_avx512f.
6833 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
6834 <avx512>_fmadd_<mode>_mask3<round_name>,
6835 <avx512>_fmsub_<mode>_mask<round_name>,
6836 <avx512>_fmsub_<mode>_mask3<round_name>,
6837 <avx512>_fnmadd_<mode>_mask<round_name>,
6838 <avx512>_fnmadd_<mode>_mask3<round_name>,
6839 <avx512>_fnmsub_<mode>_mask<round_name>,
6840 <avx512>_fnmsub_<mode>_mask3<round_name>,
6841 <avx512>_fmaddsub_<mode>_mask<round_name>,
6842 <avx512>_fmaddsub_<mode>_mask3<round_name>,
6843 <avx512>_fmsubadd_<mode>_mask<round_name>,
6844 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
6845 (*vec_widen_umult_even_v16si<mask_name>,
6846 *vec_widen_smult_even_v16si<mask_name>): Likewise.
6847 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
6849 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6851 * extend.texi: Add 'nocf_check' documentation.
6852 * gimple.texi: Add second parameter to
6853 gimple_build_call_from_tree.
6854 * invoke.texi: Add -fcf-protection documentation.
6855 * rtl.texi: Add REG_CALL_NOTRACK documenation.
6857 2017-10-20 Richard Biener <rguenther@suse.de>
6859 PR tree-optimization/82473
6860 * tree-vect-loop.c (vectorizable_reduction): Properly get at
6861 the largest input type.
6863 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6865 * c-attribs.c (handle_nocf_check_attribute): New function.
6866 (c_common_attribute_table): Add 'nocf_check' handling.
6867 * gimple-parser.c: Add second argument NULL to
6868 gimple_build_call_from_tree.
6869 * attrib.c (comp_type_attributes): Check nocf_check attribute.
6870 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
6872 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
6873 * common.opt: Add fcf-protection flag.
6874 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
6875 * flag-types.h: Add enum cf_protection_level.
6876 * gimple.c (gimple_build_call_from_tree): Add second parameter.
6877 Add 'nocf_check' attribute propagation to gimple call.
6878 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
6879 (gimple_build_call_from_tree): Update prototype.
6880 (gimple_call_nocf_check_p): New function.
6881 (gimple_call_set_nocf_check): Likewise.
6882 * gimplify.c: Add second argument to gimple_build_call_from_tree.
6883 * ipa-icf.c: Add nocf_check attribute in statement hash.
6884 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
6885 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
6886 * toplev.c (process_options): Add flag_cf_protection handling.
6888 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6890 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
6892 2017-10-20 Richard Biener <rguenther@suse.de>
6894 PR tree-optimization/82603
6895 * tree-if-conv.c (predicate_mem_writes): Make sure to only
6896 remove false predicated stores.
6898 2017-10-20 Richard Biener <rguenther@suse.de>
6900 * graphite-isl-ast-to-gimple.c
6901 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
6902 Remove return value and simplify, dump copied stmt after lhs
6904 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
6905 Reduce dump verbosity.
6906 (gsi_insert_earliest): Likewise.
6907 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
6908 * graphite.c (print_global_statistics): Adjust dumping.
6909 (print_graphite_scop_statistics): Likewise.
6910 (print_graphite_statistics): Do not dump loops here.
6911 (graphite_transform_loops): But here.
6913 2017-10-20 Nicolas Roche <roche@adacore.com>
6915 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
6916 * configure: Regenerate.
6918 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6921 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
6922 functions when optimizing replace GIMPLE_RETURN stmts with
6923 calls to __builtin_unreachable ().
6926 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
6927 for -fsanitize=thread link of executables.
6928 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
6929 link of executables.
6932 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
6934 (<shift_insn><mode>3<mask_name>): Change the last of the 3
6935 define_insns for logical vector shifts to use VI248_AVX512BW
6936 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
6937 condition, useless isa and prefix attributes. Change the first
6938 2 of these define_insns to ...
6939 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
6940 define_insn for avx512vl.
6941 (<shift_insn><mode>3): ... and this, new define_insn without
6942 masking for non-avx512vl.
6945 * config/i386/sse.md (*andnot<mode>3,
6946 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
6947 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
6948 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
6949 not applied use empty suffix even for TARGET_AVX512VL.
6950 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
6951 is applied, supply evex,evex or evex,evex,evex instead of just
6954 2017-10-20 Julia Koval <julia.koval@intel.com>
6956 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
6957 (OPTION_MASK_ISA_GFNI_UNSET): New.
6958 (ix86_handle_option): Handle OPT_mgfni.
6959 * config/i386/cpuid.h (bit_GFNI): New.
6960 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
6961 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
6962 * config/i386/i386.c (ix86_target_string): Add -mgfni.
6963 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
6964 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
6965 * config/i386/i386.opt: Add mgfni.
6967 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
6969 * config/msp430/msp430.c (msp430_option_override): Disable
6970 -fdelete-null-pointer-checks.
6971 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
6973 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6975 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
6976 of x87 and SSE instructions.
6978 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6980 * asan.c (create_cond_insert_point): Do not update edge count.
6981 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
6982 (afdo_propagate_circuit): Likewise.
6983 (afdo_calculate_branch_prob): Likewise.
6984 (afdo_annotate_cfg): Likewise.
6985 * basic-block.h (struct edge_def): Remove count.
6986 (edge_def::count): New accessor.
6987 * bb-reorder.c (rotate_loop): Update.
6988 (find_traces_1_round): Update.
6989 (connect_traces): Update.
6990 (sanitize_hot_paths): Update.
6991 * cfg.c (unchecked_make_edge): Update.
6992 (make_single_succ_edge): Update.
6993 (check_bb_profile): Update.
6994 (dump_edge_info): Update.
6995 (update_bb_profile_for_threading): Update.
6996 (scale_bbs_frequencies_int): Update.
6997 (scale_bbs_frequencies_gcov_type): Update.
6998 (scale_bbs_frequencies_profile_count): Update.
6999 (scale_bbs_frequencies): Update.
7000 * cfganal.c (connect_infinite_loops_to_exit): Update.
7001 * cfgbuild.c (compute_outgoing_frequencies): Update.
7002 (find_many_sub_basic_blocks): Update.
7003 * cfgcleanup.c (try_forward_edges): Update.
7004 (try_crossjump_to_edge): Update
7005 * cfgexpand.c (expand_gimple_cond): Update
7006 (expand_gimple_tailcall): Update
7007 (construct_exit_block): Update
7008 * cfghooks.c (verify_flow_info): Update
7009 (redirect_edge_succ_nodup): Update
7010 (split_edge): Update
7011 (make_forwarder_block): Update
7012 (duplicate_block): Update
7013 (account_profile_record): Update
7014 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7015 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7016 * cfgloopmanip.c (scale_loop_profile): Update.
7018 (lv_adjust_loop_entry_edge): Update.
7019 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7020 (force_nonfallthru_and_redirect): Update.
7021 (purge_dead_edges): Update.
7022 (rtl_flow_call_edges_add): Update.
7023 * cgraphunit.c (init_lowered_empty_function): Update.
7024 (cgraph_node::expand_thunk): Update.
7025 * gimple-pretty-print.c (dump_probability): Update.
7026 (dump_edge_probability): Update.
7027 * gimple-ssa-isolate-paths.c (isolate_path): Update.
7028 * haifa-sched.c (sched_create_recovery_edges): Update.
7029 * hsa-gen.c (convert_switch_statements): Update.
7030 * ifcvt.c (dead_or_predicable): Update.
7031 * ipa-inline-transform.c (inline_transform): Update.
7032 * ipa-split.c (split_function): Update.
7033 * ipa-utils.c (ipa_merge_profiles): Update.
7034 * loop-doloop.c (add_test): Update.
7035 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
7036 * lto-streamer-in.c (input_cfg): Update.
7037 (input_function): Update.
7038 * lto-streamer-out.c (output_cfg): Update.
7039 * modulo-sched.c (sms_schedule): Update.
7040 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7041 * predict.c (maybe_hot_edge_p): Update.
7042 (unlikely_executed_edge_p): Update.
7043 (probably_never_executed_edge_p): Update.
7044 (dump_prediction): Update.
7045 (drop_profile): Update.
7046 (propagate_unlikely_bbs_forward): Update.
7047 (determine_unlikely_bbs): Update.
7048 (force_edge_cold): Update.
7049 * profile.c (compute_branch_probabilities): Update.
7050 * reg-stack.c (better_edge): Update.
7051 * shrink-wrap.c (handle_simple_exit): Update.
7052 * tracer.c (better_p): Update.
7053 * trans-mem.c (expand_transaction): Update.
7054 (split_bb_make_tm_edge): Update.
7055 * tree-call-cdce.c: Update.
7056 * tree-cfg.c (gimple_find_sub_bbs): Update.
7057 (gimple_split_edge): Update.
7058 (gimple_duplicate_sese_region): Update.
7059 (gimple_duplicate_sese_tail): Update.
7060 (gimple_flow_call_edges_add): Update.
7061 (insert_cond_bb): Update.
7062 (execute_fixup_cfg): Update.
7063 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
7064 * tree-complex.c (expand_complex_div_wide): Update.
7065 * tree-eh.c (lower_resx): Update.
7066 (unsplit_eh): Update.
7067 (cleanup_empty_eh_move_lp): Update.
7068 * tree-inline.c (copy_edges_for_bb): Update.
7069 (freqs_to_counts): Update.
7070 (copy_cfg_body): Update.
7071 * tree-ssa-dce.c (remove_dead_stmt): Update.
7072 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7073 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7074 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7075 (unloop_loops): Update.
7076 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7077 * tree-ssa-loop-split.c (connect_loops): Update.
7078 (split_loop): Update.
7079 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7080 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7081 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7082 * tree-ssa-reassoc.c (branch_fixup): Update.
7083 * tree-ssa-tail-merge.c (replace_block_by): Update.
7084 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7085 (compute_path_counts): Update.
7086 (update_profile): Update.
7087 (recompute_probabilities): Update.
7088 (update_joiner_offpath_counts): Update.
7089 (estimated_freqs_path): Update.
7090 (freqs_to_counts_path): Update.
7091 (clear_counts_path): Update.
7092 (ssa_fix_duplicate_block_edges): Update.
7093 (duplicate_thread_path): Update.
7094 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7095 (case_bit_test_cmp): Update.
7096 (collect_switch_conv_info): Update.
7097 (gen_inbound_check): Update.
7098 (do_jump_if_equal): Update.
7099 (emit_cmp_and_jump_insns): Update.
7100 * tree-tailcall.c (decrease_profile): Update.
7101 (eliminate_tail_call): Update.
7102 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7103 (vect_do_peeling): Update.
7104 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7105 * ubsan.c (ubsan_expand_null_ifn): Update.
7106 (ubsan_expand_ptr_ifn): Update.
7107 * value-prof.c (gimple_divmod_fixed_value): Update.
7108 (gimple_mod_pow2): Update.
7109 (gimple_mod_subtract): Update.
7110 (gimple_ic): Update.
7111 (gimple_stringop_fixed_value): Update.
7113 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7116 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
7118 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
7120 PR rtl-optimization/82395
7121 * ira-color.c (allocno_priority_compare_func): Fix comparison step
7122 based on non_spilled_static_chain_regno_p.
7124 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7126 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
7127 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
7128 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
7130 2017-10-19 Martin Sebor <msebor@redhat.com>
7132 PR tree-optimization/82596
7133 * tree.c (array_at_struct_end_p): Handle STRING_CST.
7135 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7137 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
7138 (get_mem_refs_of_builtin_call): Likewise.
7139 * builtins.c (expand_builtin_apply): Adjust call to
7140 allocate_dynamic_stack_space.
7141 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
7142 the third argument to allocate_dynamic_stack_space, otherwise -1.
7143 (expand_builtin): Deal with all alloca variants.
7144 (is_inexpensive_builtin): Likewise.
7145 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
7146 * calls.c (special_function_p): Deal with all alloca variants.
7147 (initialize_argument_information): Adjust call to
7148 allocate_dynamic_stack_space.
7149 (expand_call): Likewise.
7150 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
7151 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
7152 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
7153 use it for the stack usage computation.
7154 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
7155 * function.c (gimplify_parameters): Call build_alloca_call_expr.
7156 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
7157 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
7158 (in_loop_p): Remove first argument and useless check.
7159 (pass_walloca::execute): Remove useless test and adjust call to above.
7160 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
7161 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
7162 (gimplify_call_expr): Deal with all alloca variants.
7163 * hsa-gen.c (gen_hsa_alloca): Likewise.
7164 (gen_hsa_insns_for_call): Likewise.
7165 * ipa-pure-const.c (special_builtin_state): Likewise.
7166 * tree-chkp.c (chkp_build_returned_bound): Likewise.
7167 * tree-object-size.c (alloc_object_size): Likewise.
7168 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
7169 (call_may_clobber_ref_p_1): Likewise.
7170 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
7171 (ccp_fold_stmt): Likewise.
7172 (optimize_stack_restore): Likewise.
7173 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
7174 (mark_all_reaching_defs_necessary_1): Likewise.
7175 (propagate_necessity): Likewise.
7176 (eliminate_unnecessary_stmts): Likewise.
7177 * tree.c (build_common_builtin_nodes): Build
7178 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
7179 (build_alloca_call_expr): New function.
7180 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
7181 (CASE_BUILT_IN_ALLOCA): Likewise.
7182 (build_alloca_call_expr): Declare.
7183 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
7185 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7188 * dwarf2out.c (new_die_raw): New static inline function.
7189 (new_die): Use it to create the DIE.
7190 (add_AT_external_die_ref): Likewise.
7191 (clone_die): Likewise.
7192 (clone_as_declaration): Likewise.
7193 (dwarf2out_vms_debug_main_pointer): Likewise.
7194 (base_type_die): Likewise. Remove early return for corner cases.
7195 Do not call add_pubtype on the DIE here.
7196 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
7197 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
7198 typedefs for base types with DW_AT_endianity. Make sure a DIE with
7199 native order exists for base types, attach the DIE manually and call
7200 add_pubtype on it. Do not equate a reverse order DIE to the type.
7202 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
7204 * config/arm/arm.c (align_ok_ldrd_strd): New function.
7205 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
7207 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
7209 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7211 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
7212 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
7213 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
7214 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
7215 * opts.c (sanitizer_opts): Add builtin.
7216 * ubsan.c (instrument_builtin): New function.
7217 (pass_ubsan::execute): Call it.
7218 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
7219 * doc/invoke.texi: Document -fsanitize=builtin.
7221 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
7222 builtins, store max (log2 (align), 0) into uchar field instead of
7223 align into uptr field.
7224 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
7225 store uchar 0 field instead of uptr 0 field.
7226 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
7227 instead of passing one address of struct with 2 locations pass
7228 two addresses of structs with 1 location each.
7229 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
7230 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
7231 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
7232 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
7233 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
7234 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
7235 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
7236 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
7238 2017-10-19 Martin Liska <mliska@suse.cz>
7241 * file-find.c (remove_prefix): Remove.
7242 * file-find.h (remove_prefix): Likewise.
7243 * gcc-ar.c: Remove smartness of lookup.
7245 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
7247 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
7248 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
7249 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
7251 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7254 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
7255 (setcc + and to xor + setcc): New peephole2.
7257 2017-10-19 Tom de Vries <tom@codesourcery.com>
7259 * doc/sourcebuild.texi (Test Directives, Variants of
7260 dg-require-support): Add dg-require-stack-size.
7262 2017-10-19 Martin Liska <mliska@suse.cz>
7265 * gimplify.c (gimplify_decl_expr): Do not instrument variables
7266 that have a large alignment.
7267 (gimplify_target_expr): Likewise.
7269 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
7271 PR rtl-optimization/82602
7272 * ira.c (rtx_moveable_p): Return false for volatile asm.
7274 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
7277 * config/i386/i386-modes.def (CCGZ): New CC mode.
7278 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
7279 * config/i386/predicates.md (ix86_comparison_operator):
7281 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7282 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
7283 with double-word subtraction.
7284 (put_condition_code): Handle CCGZmode.
7286 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
7288 * wide-int.cc (debug (const wide_int &)): New.
7289 (debug (const wide_int *)): New.
7290 (debug (const widest_int &)): New.
7291 (debug (const widest_int *)): New.
7293 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
7296 * lra-constraints.c (curr_insn_transform): Use non-input operand
7297 instead of output one for matched reload.
7299 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7301 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
7302 (tree-ssa-loop-ivopts.h): New header file.
7303 (struct builtin_info): New fields.
7304 (classify_builtin_1): Compute and record base and offset parts for
7305 memset builtin partition by calling strip_offset.
7306 (offset_cmp, fuse_memset_builtins): New functions.
7307 (finalize_partitions): Fuse adjacent memset partitions by calling
7309 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
7310 Expose the interface.
7311 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
7313 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7315 PR tree-optimization/82574
7316 * tree-loop-distribution.c (find_single_drs): New parameter. Check
7317 that data reference must be executed exactly once per iteration
7318 against the outermost loop in nest.
7319 (classify_partition): Update call to above function.
7321 2017-10-18 Richard Biener <rguenther@suse.de>
7323 PR tree-optimization/82591
7324 * graphite.c (graphite_transform_loops): Move code gen message
7326 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7327 Here. Handle scop_to_isl_ast failing.
7328 (scop_to_isl_ast): Limit the number of ISL operations.
7330 2017-10-18 Richard Biener <rguenther@suse.de>
7332 * graphite-isl-ast-to-gimple.c
7333 (translate_isl_ast_to_gimple::set_rename): Simplify.
7334 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
7335 (graphite_copy_stmts_from_block): ... here.
7336 (copy_bb_and_scalar_dependences): Simplify.
7337 (add_parameters_to_ivs_params): Canonicalize.
7338 (generate_entry_out_of_ssa_copies): Simplify.
7339 * graphite-sese-to-poly.c (extract_affine_name): Simplify
7340 by passing in ISL dimension.
7341 (parameter_index_in_region_1): Rename to ...
7342 (parameter_index_in_region): ... this.
7343 (extract_affine): Adjust assert, pass down parameter index.
7344 (add_param_constraints): Use range-info when available.
7345 (build_scop_context): Adjust.
7346 * sese.c (new_sese_info): Adjust.
7347 (free_sese_info): Likewise.
7348 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
7349 Remove unused typedefs.
7350 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
7352 2017-10-18 Martin Liska <mliska@suse.cz>
7354 * combine.c (simplify_compare_const): Add gcc_fallthrough.
7356 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7358 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
7359 (s390_sched_init): Do not reset s390_sched_state if we entered the
7360 current basic block via a fallthru edge and all others are unlikely.
7362 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7364 * config/s390/s390.c (NUM_SIDES): New variable.
7365 (LONGRUNNING_THRESHOLD): New variable.
7366 (LATENCY_FACTOR): New variable.
7367 (s390_sched_score): Decrease score for long-running instructions on
7369 (s390_sched_variable_issue): Perform bookkeeping for long-running
7372 2017-10-18 Richard Biener <rguenther@suse.de>
7374 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
7375 Simplify with removal of the parameter rename map.
7376 (set_rename): Likewise.
7377 (should_copy_to_new_region): Likewise.
7378 (graphite_copy_stmts_from_block): Likewise.
7379 (copy_bb_and_scalar_dependences): Remove initialization of
7380 unused copied_bb_map.
7382 (copy_internal_parameters): Likewise.
7383 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
7384 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
7385 Use INTEGRAL_TYPE_P.
7386 (parameter_index_in_region_1): Rename to ...
7387 (assign_parameter_index_in_region): ... this. Assert we have
7388 a parameter we handle.
7389 (scan_tree_for_params): Adjust.
7390 * sese.h (parameter_rename_map_t): Remove.
7391 (struct sese_info_t): Remove unused parameter_rename_map and
7392 copied_bb_map members.
7393 * sese.c (new_sese_info): Adjust.
7394 (free_sese_info): Likewise.
7396 2017-10-18 Martin Liska <mliska@suse.cz>
7399 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
7400 on an abnormal edge.
7402 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
7404 * doc/invoke.texi (ffunction-sections and fdata-sections):
7407 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7409 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
7410 the use statement can throw internally.
7412 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7414 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
7415 any RTX present on the RHS of a SET.
7416 * compare-elim.c (try_eliminate_compare): Restore comment.
7418 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7420 * langhooks.h (struct lang_hooks): Document that tree_size langhook
7421 may be also called on tcc_type nodes.
7422 * langhooks.c (lhd_tree_size): Likewise.
7424 2017-10-17 David Malcolm <dmalcolm@redhat.com>
7426 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
7427 format_warning_at_substring.
7428 (maybe_warn): Convert source_range * param to a location_t. Pass
7429 UNKNOWN_LOCATION rather than NULL to fmtwarn.
7430 (format_directive): Remove code to extract source_ranges and
7431 source_range * in favor of just a location_t.
7432 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
7434 * substring-locations.c (format_warning_va): Convert
7435 source_range * param to a location_t.
7436 (format_warning_at_substring): Likewise.
7437 * substring-locations.h (format_warning_va): Likewise.
7438 (format_warning_at_substring): Likewise.
7440 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
7442 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
7444 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
7445 and scatter/gather ops.
7447 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
7448 vec_gather_load and vec_scatter_store.
7449 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
7450 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
7451 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
7452 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
7453 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
7454 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
7456 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
7458 * reg-stack.c (compare_for_stack_reg): Add bool argument.
7459 Detect FTST instruction and handle its register pops. Only pop
7460 second operand if can_pop_second_op is true.
7461 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
7462 set can_pop_second_op to false in the compare_for_stack_reg call.
7464 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
7465 output_fp_compare for stack register operands.
7466 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
7467 instructions here. Do not emit stack register pops here. Assert
7468 that FCOMPP pops next to top stack register. Rewrite function.
7470 2017-10-17 Nathan Sidwell <nathan@acm.org>
7473 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
7474 use DECL_ASSEMBLER_NAME_RAW.
7477 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
7480 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
7481 Wilco Dijkstra <wilco.dijkstra@arm.com>
7483 * builtins.c (expand_builtin_update_setjmp_buf): Add a
7484 converstion to Pmode from the buf_addr.
7486 2017-10-17 Richard Biener <rguenther@suse.de>
7488 * graphite-dependences.c (scop_get_reads_and_writes): Change
7489 output parameters to references.
7491 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
7493 PR 71026/tree-optimization
7494 * fold-const.c (distribute_real_division): Removed.
7495 (fold_binary_loc): Remove calls to distribute_real_divison.
7497 2017-10-17 Richard Biener <rguenther@suse.de>
7499 * graphite-scop-detection.c
7500 (scop_detection::stmt_has_simple_data_refs_p): Always use
7501 the full nest as region.
7502 (try_generate_gimple_bb): Likewise.
7503 * sese.c (scalar_evolution_in_region): Simplify now that
7504 SCEV can handle instantiation in regions.
7505 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
7506 in the non-loop part of a function if requested.
7508 2017-10-17 Richard Biener <rguenther@suse.de>
7510 PR tree-optimization/82563
7511 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
7513 (graphite_regenerate_ast_isl): Call it.
7514 * graphite-scop-detection.c (build_scops): Remove entry edge split.
7516 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7518 PR tree-optimization/82549
7519 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
7520 Formatting fixes. Instead of calling make_bit_field_ref with negative
7523 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
7525 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
7526 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
7527 _mm_maskz_reduce_ss): New.
7528 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
7529 __builtin_ia32_reducess_mask): Ditto..
7530 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
7531 * config/i386/sse.md (reduces<mode>): Renamed to ...
7532 (reduces<mode><mask_scalar_name>): ... this.
7533 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
7535 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
7536 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
7538 2017-10-16 David Malcolm <dmalcolm@redhat.com>
7540 * Makefile.in (OBJS): Add unique-ptr-tests.o.
7541 * selftest-run-tests.c (selftest::run_tests): Call
7542 selftest::unique_ptr_tests_cc_tests.
7543 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
7544 * unique-ptr-tests.cc: New file.
7546 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
7549 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7551 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7552 (make_hard_regno_born, make_hard_regno_dead): Update
7553 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
7555 2017-10-16 Jeff Law <law@redhat.com>
7557 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7559 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
7561 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
7563 2017-10-16 Olivier Hainque <hainque@adacore.com>
7565 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
7566 with_cpu if we were configured for an e500v2 target cpu name.
7568 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
7570 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
7571 * doc/invoke.texi: Document +nodsp as a valid extension for
7574 2017-10-16 Martin Liska <mliska@suse.cz>
7576 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
7577 (test_set_range): Likewise.
7578 (test_range_functions): Rename to ...
7579 (test_bit_in_range): ... this.
7580 (sbitmap_c_tests): Add new test.
7582 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7584 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
7586 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
7587 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
7589 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7591 * config/aarch64/aarch64-builtins.c
7592 (aarch64_types_quadopu_lane_qualifiers): New.
7593 (TYPES_QUADOPU_LANE): New.
7594 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
7595 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
7596 (aarch64_<sur>dot_laneq<vsi2qi>): New.
7597 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
7598 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
7599 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
7600 (Vdottype, DOTPROD): New.
7601 (sur): Add SDOT and UDOT.
7603 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7605 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
7606 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
7607 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
7609 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
7610 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
7611 Enable TARGET_DOTPROD.
7612 (cortex-a75.cortex-a55): Likewise.
7613 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
7615 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7617 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
7618 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
7620 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
7622 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
7623 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
7624 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
7625 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
7626 * config/arm/types.md (neon_dot, neon_dot_q): New.
7627 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
7629 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7631 * config/arm/arm.h (TARGET_DOTPROD): New.
7632 * config/arm/arm.c (arm_arch_dotprod): New.
7633 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
7634 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
7635 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
7636 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
7637 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
7638 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
7639 * doc/invoke.texi (armv8.2-a): Document dotprod
7641 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7643 * i386.c (ix86_vec_cost): New function.
7644 (ix86_rtx_costs): Handle vector operations better.
7645 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
7646 * x86-tune-costs.h: Add new costs to all tables.
7648 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7650 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
7652 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
7653 divsd, sqrtss and sqrtsd
7654 * x86-tune-costs.h: Add new entries to all costs.
7655 (znver1_cost): Fix to match real instruction latencies.
7657 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7658 Michael Collison <michael.collison@arm.com>
7660 * compare-elim.c: Include emit-rtl.h.
7661 (can_merge_compare_into_arith): New function.
7662 (try_validate_parallel): Likewise.
7663 (try_merge_compare): Likewise.
7664 (try_eliminate_compare): Call the above when no previous clobber
7666 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
7669 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7673 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
7674 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
7678 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
7679 to be any operand_equal_p operands. For & (B - 1) require
7680 B to be power of 2. Recognize
7681 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
7683 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7686 * optabs.c (expand_memory_blockage): Fix call of
7687 targetm.have_memory_blockage.
7689 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7692 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
7693 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
7694 objects to extra_objs instead of overwriting it.
7696 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7698 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
7699 Use any_fp_register_operand as operand[3] predicate. Simplify
7700 equality test for operands[2] and operands[4] memory location.
7701 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
7702 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
7703 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
7704 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
7705 any_fp_register_operand as operand[1] predicate. Simplify
7706 equality test for operands[0] and operands[3] memory location.
7707 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
7708 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
7709 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
7711 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7713 * target-insns.def: Add memory_blockage.
7714 * optabs.c (expand_memory_blockage): New function.
7715 (expand_asm_memory_barrier): Rename ...
7716 (expand_asm_memory_blockage): ... to this.
7717 (expand_mem_thread_fence): Call expand_memory_blockage
7718 instead of expand_asm_memory_barrier.
7719 (expand_mem_singnal_fence): Ditto.
7720 (expand_atomic_load): Ditto.
7721 (expand_atomic_store): Ditto.
7722 * doc/md.texi (Standard Pattern Names For Generation):
7723 Document memory_blockage instruction pattern.
7725 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
7727 * config/rl78/rl78.c (rl78_emit_libcall): New function.
7728 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
7729 * config/rl78/rl78.md: New define_expand "adddi3".
7731 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7733 * cfghooks.c (verify_flow_info): Disable check that all probabilities
7736 2017-10-13 Jeff Law <law@redhat.com>
7738 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
7740 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7743 * internal-fn.c (expand_mul_overflow): If both operands have
7744 the same highpart of -1 or 0 and the topmost bit of lowpart
7745 is different, overflow is if res <= 0 rather than res < 0.
7747 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
7749 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
7750 TARGET_P9_VECTOR code for unaligned_load case.
7752 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7754 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
7756 2017-10-13 Nathan Sidwell <nathan@acm.org>
7758 * tree-core.h (tree_contains_struct): Make bool.
7759 * tree.c (tree_contains_struct): Likewise.
7760 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
7761 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
7762 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
7763 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
7764 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
7766 2017-10-13 Richard Biener <rguenther@suse.de>
7768 * graphite-isl-ast-to-gimple.c
7769 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
7770 parameters and dominance check.
7771 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
7772 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
7773 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
7774 Do not update SSA form here or do intermediate IL verification.
7775 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
7776 (graphite_initialize): Remove check on the number of loops in
7777 the function and inline into graphite_transform_loops.
7778 (graphite_finalize): Inline into graphite_transform_loops.
7779 (graphite_transform_loops): Perform SSA update and IL verification
7781 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
7783 2017-10-13 Richard Biener <rguenther@suse.de>
7785 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
7786 graphite_expression_type_precision): Avoid global constructor
7788 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
7789 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
7790 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
7791 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
7792 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
7793 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
7795 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
7798 * config/i386/i386.h (ix86_red_zone_size): New.
7799 * config/i386/i386.md (push peephole2s): Replace
7800 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
7802 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7803 Alan Hayward <alan.hayward@arm.com>
7804 David Sherwood <david.sherwood@arm.com>
7806 * combine.c (can_change_dest_mode): Reject changes in
7807 REGMODE_NATURAL_SIZE.
7809 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7810 Alan Hayward <alan.hayward@arm.com>
7811 David Sherwood <david.sherwood@arm.com>
7813 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
7814 (expand_debug_source_expr): Likewise.
7815 * combine.c (combine_simplify_rtx): Likewise.
7816 * cse.c (fold_rtx): Likewise.
7817 * fwprop.c (canonicalize_address): Likewise.
7818 * targhooks.c (default_shift_truncation_mask): Likewise.
7820 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7821 Alan Hayward <alan.hayward@arm.com>
7822 David Sherwood <david.sherwood@arm.com>
7824 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
7825 (widened_mode): Likewise.
7826 (expand_unop): Likewise.
7827 * ree.c (transform_ifelse): Likewise.
7828 (merge_def_and_ext): Likewise.
7829 (combine_reaching_defs): Likewise.
7830 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7832 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7833 Alan Hayward <alan.hayward@arm.com>
7834 David Sherwood <david.sherwood@arm.com>
7836 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
7837 * combine.c (gen_lowpart_for_combine): Likewise.
7838 * dwarf2out.c (rtl_for_decl_location): Likewise.
7839 * final.c (alter_subreg): Likewise.
7840 * rtlhooks.c (gen_lowpart_general): Likewise.
7841 (gen_lowpart_if_possible): Likewise.
7843 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7844 Alan Hayward <alan.hayward@arm.com>
7845 David Sherwood <david.sherwood@arm.com>
7847 * calls.c (expand_call): Use subreg_lowpart_offset.
7848 * cse.c (cse_insn): Likewise.
7849 * regcprop.c (copy_value): Likewise.
7850 (copyprop_hardreg_forward_1): Likewise.
7852 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7855 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
7856 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
7857 =Q constraints to +Q and into insn condition add check
7858 that operands[0] and operands[1] are equal.
7859 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
7860 =Q constraints to +Q and into insn condition add check
7861 that operands[0] is equal to either operands[1] or operands[2].
7864 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
7865 instead of handling MINUS_EXPR twice (once for each argument),
7866 canonicalize operand order and handle just once, use rtype where
7867 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
7870 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
7871 any values of __C while still being pattern recognizable as a simple
7874 2017-10-13 Richard Biener <rguenther@suse.de>
7876 PR tree-optimization/82451
7878 2017-10-02 Richard Biener <rguenther@suse.de>
7880 PR tree-optimization/82355
7881 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7882 a mapping for the enclosing loop but avoid generating one for
7884 (copy_bb_and_scalar_dependences): Remove premature codegen
7885 error on PHIs in blocks duplicated into multiple places.
7886 * graphite-scop-detection.c
7887 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7888 in the region use it as loop and nest to analyze the DR in.
7889 (try_generate_gimple_bb): Likewise.
7890 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7891 (add_loop_constraints): For blocks in a loop not in the region
7892 create a dimension with a single iteration.
7893 * sese.h (gbb_loop_at_index): Remove assert.
7895 * cfgloop.c (loop_preheader_edge): For the loop tree root
7896 return the single successor of the entry block.
7897 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7898 Reset the SCEV hashtable and niters.
7899 * graphite-scop-detection.c
7900 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
7901 assert that we only have POLYNOMIAL_CHREC that vary in loops
7902 contained in the region.
7903 (scop_detection::graphite_can_represent_expr): Adjust.
7904 (scop_detection::stmt_has_simple_data_refs_p): For loops
7905 not in the region set loop to NULL. The nest is now the
7906 entry edge to the region.
7907 (try_generate_gimple_bb): Likewise.
7908 * sese.c (scalar_evolution_in_region): Adjust for
7909 instantiate_scev change.
7910 * tree-data-ref.h (graphite_find_data_references_in_stmt):
7911 Make nest parameter the edge into the region.
7912 (create_data_ref): Likewise.
7913 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
7914 entry edge into a region and adjust instantiate_scev calls.
7915 (create_data_ref): Likewise.
7916 (graphite_find_data_references_in_stmt): Likewise.
7917 (find_data_references_in_stmt): Pass the loop preheader edge
7918 from the nest argument.
7919 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
7920 parameter the edge into the region.
7921 (instantiate_parameters): Use the loop preheader edge as entry.
7922 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
7924 (get_instantiated_value_entry): Make instantiate_below parameter
7925 the edge into the region.
7926 (instantiate_scev_name): Likewise. Adjust dominance checks,
7927 when we cannot use loop-based instantiation instantiate by
7928 walking use-def chains.
7929 (instantiate_scev_poly): Adjust.
7930 (instantiate_scev_binary): Likewise.
7931 (instantiate_scev_convert): Likewise.
7932 (instantiate_scev_not): Likewise.
7933 (instantiate_array_ref): Remove.
7934 (instantiate_scev_3): Likewise.
7935 (instantiate_scev_2): Likewise.
7936 (instantiate_scev_1): Likewise.
7937 (instantiate_scev_r): Do not blindly handle N-operand trees.
7938 Do not instantiate array-refs. Handle all constants and invariants.
7939 (instantiate_scev): Make instantiate_below parameter
7940 the edge into the region.
7941 (resolve_mixers): Use the loop preheader edge for the region
7942 parameter to instantiate_scev_r.
7943 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
7945 2017-10-13 Richard Biener <rguenther@suse.de>
7947 PR tree-optimization/82525
7948 * graphite-isl-ast-to-gimple.c
7949 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
7951 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
7952 Fail code generation when we cannot represent the isl integer.
7953 (binary_op_to_tree): Elide modulo operations that are no-ops
7954 in the type we code generate. Remove now superfluous code
7957 2017-10-13 Richard Biener <rguenther@suse.de>
7959 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
7960 (scop_detection::harmful_loop_in_region): Remove premature
7961 IV type restriction.
7962 (scop_detection::graphite_can_represent_scev): We can handle
7963 pointer IVs just fine.
7965 2017-10-13 Alan Modra <amodra@gmail.com>
7967 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
7968 "Clobbers and Scratch Registers". Add paragraph on
7969 alternative to clobbers for scratch registers and OpenBLAS
7972 2017-10-13 Alan Modra <amodra@gmail.com>
7974 * doc/extend.texi (Clobbers): Correct vax example. Delete old
7975 example of a memory input for a string of known length. Move
7976 commentary out of table. Add a number of new examples
7977 covering array memory inputs.
7979 2017-10-12 Martin Liska <mliska@suse.cz>
7981 PR tree-optimization/82493
7982 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
7983 (test_range_functions): New function.
7984 (sbitmap_c_tests): Likewise.
7985 * selftest-run-tests.c (selftest::run_tests): Run new tests.
7986 * selftest.h (sbitmap_c_tests): New function.
7988 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7990 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7992 * config/rs6000/amo.h: Fix spacing issue.
7994 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7997 * config/i386/i386.md (*ashl<mode>3_mask_1,
7998 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
7999 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8002 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8004 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8005 (profile_probability): Set max_probability
8006 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8008 * profile-count.c (profile_probability::differs_from_p): Do not
8009 rely on max_probaiblity == 10000
8011 2017-10-12 Jeff Law <law@redhat.com>
8013 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8016 2017-10-12 Martin Sebor <msebor@redhat.com>
8020 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
8021 (handle_alias_pairs): Call it.
8022 * common.opt (-Wattribute-alias): New option.
8023 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
8024 * doc/invoke.texi (-Wattribute-alias): Document.
8026 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
8029 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8031 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8033 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8034 (make_hard_regno_born, make_hard_regno_dead): Update
8035 bb_killed_pseudos and bb_gen_pseudos.
8037 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8039 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
8041 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
8043 * config/alpha/alpha.c (alpha_split_conditional_move):
8044 Use std::swap instead of manually swapping.
8045 (alpha_stdarg_optimize_hook): Ditto.
8046 (alpha_canonicalize_comparison): Ditto.
8048 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8050 * tree-loop-distribution.c (struct builtin_info): New struct.
8051 (struct partition): Refactor fields into struct builtin_info.
8052 (partition_free): Free struct builtin_info.
8053 (build_size_arg_loc, build_addr_arg_loc): Delete.
8054 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
8055 information from struct builtin_info.
8056 (find_single_drs): New function refactored from classify_partition.
8057 Also moved builtin validity checks to this function.
8058 (compute_access_range, alloc_builtin): New functions.
8059 (classify_builtin_st, classify_builtin_ldst): New functions.
8060 (classify_partition): Refactor code into functions find_single_drs,
8061 classify_builtin_st and classify_builtin_ldst.
8062 (distribute_loop): Don't do runtime alias check when distributing
8064 (find_seed_stmts_for_distribution): New function.
8065 (pass_loop_distribution::execute): Refactor code finding seed
8066 stmts into above function. Support distribution for the innermost
8067 two-level loop nest. Adjust dump information.
8069 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8071 * tree-loop-distribution.c: Adjust the general comment.
8072 (NUM_PARTITION_THRESHOLD): New macro.
8073 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
8074 (classify_partition): Skip builtin pattern of loop nest's inner loop.
8075 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
8076 in call to build_partition_graph.
8077 (finalize_partitions): New parameter. Make loop distribution more
8078 conservative by fusing more partitions.
8079 (distribute_loop): Don't do runtime alias check in case of loop nest
8081 (find_seed_stmts_for_distribution): New function.
8082 (prepare_perfect_loop_nest): New function.
8083 (pass_loop_distribution::execute): Refactor code finding seed stmts
8084 and loop nest into above functions. Support loop nest distribution.
8085 Adjust dump information accordingly.
8087 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8089 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
8090 and set PTYPE_SEQUENTIAL for merged partition.
8092 2017-10-12 Richard Biener <rguenther@suse.de>
8094 PR tree-optimization/69728
8096 2017-09-19 Richard Biener <rguenther@suse.de>
8098 PR tree-optimization/69728
8099 * graphite-sese-to-poly.c (schedule_error): New global.
8100 (add_loop_schedule): Handle empty domain by failing the
8102 (build_original_schedule): Handle schedule_error.
8104 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
8105 domain by returning an unchanged schedule.
8107 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8109 * genrecog.c (validate_pattern): For VEC_SELECT verify that
8110 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
8112 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
8114 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
8117 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8120 * expr.c (store_field): Don't optimize away bitsize == 0 store
8121 from CALL_EXPR with addressable return type.
8123 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8125 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
8126 * config/rs6000/rs6000.md (sel): Delete mode attribute.
8127 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
8128 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
8129 TARGET_ISEL instead of TARGET_ISEL<sel>.
8131 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
8133 * config/rs6000/rs6000.c
8134 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
8136 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8138 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
8140 (reg_or_cint_operand): Fix comment.
8141 (reg_or_zero_operand): New predicate.
8142 * config/rs6000/rs6000-protos.h (output_isel): Delete.
8143 * config/rs6000/rs6000.c (output_isel): Delete.
8144 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
8145 instead of reg_or_cint_operand. Output instruction directly (not via
8147 (isel_unsigned_<mode>): Ditto.
8148 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
8149 gpc_reg_operand. Add an instruction alternative for this. Output
8150 instruction directly.
8151 (*isel_reversed_unsigned_<mode>): Ditto.
8153 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8155 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
8156 (TARGET_CANONICALIZE_COMPARISON): Define.
8158 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
8161 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8162 Check whether the dest is REG before adding REG_EQUIV note.
8164 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8167 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8169 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8170 (make_hard_regno_born, make_hard_regno_dead): Update
8171 bb_killed_pseudos and bb_gen_pseudos.
8173 2017-10-11 Nathan Sidwell <nathan@acm.org>
8175 * incpath.h (enum incpath_kind): Name enum, prefix values.
8176 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
8177 * incpath.c (heads, tails): Use INC_MAX.
8178 (add_env_var_paths, add_standard_paths): Use incpath_kind.
8179 (merge_include_chains, split_quote_chain,
8180 register_include_chains): Update incpath_kind names.
8181 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
8182 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
8184 (add_framework_path, darwin_register_objc_includes): Likewise.
8185 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
8187 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8189 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
8190 Do not use float_operator operator predicate.
8191 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
8192 * config/i386/predicates.md (float_operator): Remove predicate.
8194 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8196 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
8197 (*jccxf_i387): Ditto.
8198 (*jcc<mode>_i387): Ditto.
8199 (*jccu<mode>_i387): Ditto.
8200 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
8201 (*jcc_*_i387 splitters): Remove.
8202 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
8203 * config/i386/i386.c (ix86_split_fp_branch): Remove.
8204 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
8207 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8209 * profile-count.h (slow_safe_scale_64bit): New function.
8210 (safe_scale_64bit): New inline.
8211 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
8212 * profile-count.c: Include wide-int.h
8213 (slow_safe_scale_64bit): New.
8215 2017-10-11 Nathan Sidwell <nathan@acm.org>
8217 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
8218 HAS_DECL_ASSEMBLER_NAME_P.
8219 * gimple-expr.c (gimple_decl_printable_name: Check
8220 HAS_DECL_ASSEMBLER_NAME_P too.
8221 * ipa-utils.h (type_in_anonymous_namespace_p): Check
8222 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
8223 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
8224 * passes.c (rest_of_decl_compilation): Check
8225 HAS_DECL_ASSEMBLER_NAME_P too.
8226 * recog.c (verify_changes): Likewise.
8227 * tree-pretty-print.c (dump_decl_name): Likewise.
8228 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
8230 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
8231 (SET_DECL_ASSEMBLER_NAME): Use it.
8232 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
8233 (COPY_DECL_ASSEMBLER_NAME): Likewise.
8234 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
8236 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8238 * config.gcc (i386, x86_64): Add extra objects.
8239 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
8240 (ix86_min_insn_size): Declare.
8241 (ix86_issue_rate): Declare.
8242 (ix86_adjust_cost): Declare.
8243 (ia32_multipass_dfa_lookahead): Declare.
8244 (ix86_macro_fusion_p): Declare.
8245 (ix86_macro_fusion_pair_p): Declare.
8246 (ix86_bd_has_dispatch): Declare.
8247 (ix86_bd_do_dispatch): Declare.
8248 (ix86_core2i7_init_hooks): Declare.
8249 (ix86_atom_sched_reorder): Declare.
8250 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
8251 (COSTS_N_BYTES): Move to x86-tune-costs.h.
8252 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
8253 (rip_relative_addr_p): Rename to ...
8254 (ix86_rip_relative_addr_p): ... this one; export.
8255 (memory_address_length): Update.
8256 (ix86_issue_rate): Move to x86-tune-sched.c.
8257 (ix86_flags_dependent): Move to x86-tune-sched.c.
8258 (ix86_agi_dependent): Move to x86-tune-sched.c.
8259 (exact_dependency_1): Move to x86-tune-sched.c.
8260 (exact_store_load_dependency): Move to x86-tune-sched.c.
8261 (ix86_adjust_cost): Move to x86-tune-sched.c.
8262 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
8263 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
8264 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
8265 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
8266 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
8267 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
8268 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
8269 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
8270 (min_insn_size): Rename to ...
8271 (ix86_min_insn_size): ... this one; export.
8272 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
8273 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
8274 (core2i7_first_cycle_multipass_backtrack): Move to
8275 x86-tune-sched-core.c.
8276 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
8277 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
8278 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
8279 (ix86_avoid_jump_mispredicts): Update.
8280 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
8281 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
8282 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
8283 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
8284 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
8285 (MAX_INSN): Move to ix86-tune-sched-bd.c.
8286 (MAX_IMM): Move to ix86-tune-sched-bd.c.
8287 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
8288 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
8289 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
8290 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
8291 (MAX_STORE): Move to ix86-tune-sched-bd.c.
8292 (BIG): Move to ix86-tune-sched-bd.c.
8293 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
8294 (enum insn_path): Move to ix86-tune-sched-bd.c.
8295 (get_mem_group): Move to ix86-tune-sched-bd.c.
8296 (is_cmp): Move to ix86-tune-sched-bd.c.
8297 (dispatch_violation): Move to ix86-tune-sched-bd.c.
8298 (is_branch): Move to ix86-tune-sched-bd.c.
8299 (is_prefetch): Move to ix86-tune-sched-bd.c.
8300 (init_window): Move to ix86-tune-sched-bd.c.
8301 (allocate_window): Move to ix86-tune-sched-bd.c.
8302 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
8303 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
8304 (process_end_window): Move to ix86-tune-sched-bd.c.
8305 (allocate_next_window): Move to ix86-tune-sched-bd.c.
8306 (find_constant): Move to ix86-tune-sched-bd.c.
8307 (get_num_immediates): Move to ix86-tune-sched-bd.c.
8308 (has_immediate): Move to ix86-tune-sched-bd.c.
8309 (get_insn_path): Move to ix86-tune-sched-bd.c.
8310 (get_insn_group): Move to ix86-tune-sched-bd.c.
8311 (count_num_restricted): Move to ix86-tune-sched-bd.c.
8312 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
8313 (add_insn_window): Move to ix86-tune-sched-bd.c.
8314 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
8315 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
8316 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
8317 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
8318 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
8319 (do_dispatch): Move to ix86-tune-sched-bd.c.
8320 (has_dispatch): Move to ix86-tune-sched-bd.c.
8321 * config/i386/t-i386: Add new object files.
8322 * config/i386/x86-tune-costs.h: New file.
8323 * config/i386/x86-tune-sched-atom.c: New file.
8324 * config/i386/x86-tune-sched-bd.c: New file.
8325 * config/i386/x86-tune-sched-core.c: New file.
8326 * config/i386/x86-tune-sched.c: New file.
8328 2017-10-11 Liu Hao <lh_mouse@126.com>
8330 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
8331 the generic version below instead.
8332 (should_colorize): Recognize Windows consoles as terminals
8334 * pretty-print.c [__MINGW32__] (write_all): New function.
8335 [__MINGW32__] (find_esc_head): Likewise.
8336 [__MINGW32__] (find_esc_terminator): Likewise.
8337 [__MINGW32__] (eat_esc_sequence): Likewise.
8338 [__MINGW32__] (mingw_ansi_fputs): New function that handles
8340 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
8343 2017-10-11 Richard Biener <rguenther@suse.de>
8345 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
8346 Properly call analyze_scalar_evolution with the loop of the stmt.
8348 2017-10-11 Richard Biener <rguenther@suse.de>
8350 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
8351 * tree-core.h (tree_base): Add chrec_var union member.
8352 * tree.h (CHREC_VAR): Remove.
8353 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
8354 * tree-chrec.h (build_polynomial_chrec): Adjust.
8355 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
8356 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
8358 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
8360 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
8361 * match.pd: ... here.
8362 ((T) X == (T) Y): Relax condition.
8364 2017-10-11 Bin Cheng <bin.cheng@arm.com>
8366 PR tree-optimization/82472
8367 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
8369 (break_alias_scc_partitions): Update postorder number.
8371 2017-10-11 Martin Liska <mliska@suse.cz>
8374 * opts.c (parse_no_sanitize_attribute): Do not use error_value
8376 * opts.h (parse_no_sanitize_attribute): Remove last argument.
8378 2017-10-11 Martin Liska <mliska@suse.cz>
8380 * print-rtl.c (print_insn): Move declaration of idbuf
8381 to same scope as name.
8383 2017-10-11 Martin Liska <mliska@suse.cz>
8388 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8391 2017-10-11 Martin Liska <mliska@suse.cz>
8394 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8397 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
8399 * config.gcc (default_gnu_indirect_function): Default to yes for
8400 arm*-*-linux* with glibc.
8402 2017-10-11 Richard Biener <rguenther@suse.de>
8404 * tree-scalar-evolution.c (get_scalar_evolution): Handle
8405 default-defs and types we do not want to analyze.
8406 (interpret_loop_phi): Replace unreachable code with an assert.
8407 (compute_scalar_evolution_in_loop): Remove and inline ...
8408 (analyze_scalar_evolution_1): ... here, replacing condition with
8409 what makes the intent clearer. Remove handling of cases
8410 get_scalar_evolution now handles.
8412 2017-10-10 Jim Wilson <wilson@tuliptree.org>
8414 PR rtl-optimization/81434
8415 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
8416 comment for main loop. In sched_group_found if, also add checks for
8417 pass and min_cost_group.
8419 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
8421 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
8422 (rs6000_insn_cost): New function.
8423 * config/rs6000/rs6000.md (cost): New attribute.
8425 2017-10-10 Jakub Jelinek <jakub@redhat.com>
8426 H.J. Lu <hongjiu.lu@intel.com>
8430 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
8431 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
8432 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
8433 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
8434 to OPTION_MASK_ISA_AVX512VL - builtins that have both
8435 OPTION_MASK_ISA_MMX and some other bit set require both
8436 mmx and the ISAs without the mmx bit.
8437 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
8438 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
8439 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
8440 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
8441 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
8442 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
8443 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
8444 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
8445 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
8446 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
8447 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
8448 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
8449 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
8450 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
8451 Add OPTION_MASK_ISA_MMX.
8453 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
8455 * config.gcc (armv7*-*-freebsd*): New target.
8456 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
8458 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
8460 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
8461 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
8464 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
8466 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
8467 a template parameter.
8468 (WIDE_INT_REF_FOR): Update accordingly.
8469 * tree.h (wi::int_traits <const_tree>): Delete.
8470 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
8471 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
8472 (wi::tree_to_wide_ref): New typedef.
8473 (wi::to_wide): New function.
8474 * calls.c (get_size_range): Use wi::to_wide when operating on
8476 * cgraph.c (cgraph_node::create_thunk): Likewise.
8477 * config/i386/i386.c (ix86_data_alignment): Likewise.
8478 (ix86_local_alignment): Likewise.
8479 * dbxout.c (stabstr_O): Likewise.
8480 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
8481 * expr.c (const_vector_from_tree): Likewise.
8482 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
8483 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
8484 (fold_negate_expr_1, int_const_binop_1, const_binop)
8485 (fold_convert_const_int_from_real, optimize_bit_field_compare)
8486 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
8487 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
8488 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
8489 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
8490 (fold_not_const, round_up_loc): Likewise.
8491 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
8492 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
8493 (alloca_call_type): Likewise.
8494 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
8495 * godump.c (go_output_typedef): Likewise.
8496 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
8497 * internal-fn.c (get_min_precision): Likewise.
8498 * ipa-cp.c (ipcp_store_vr_results): Likewise.
8499 * ipa-polymorphic-call.c
8500 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
8501 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
8502 (ipa_modify_call_arguments): Likewise.
8503 * match.pd: Likewise.
8504 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
8505 * print-tree.c (print_node_brief, print_node): Likewise.
8506 * stmt.c (expand_case): Likewise.
8507 * stor-layout.c (layout_type): Likewise.
8508 * tree-affine.c (tree_to_aff_combination): Likewise.
8509 * tree-cfg.c (group_case_labels_stmt): Likewise.
8510 * tree-data-ref.c (dr_analyze_indices): Likewise.
8511 (prune_runtime_alias_test_list): Likewise.
8512 * tree-dump.c (dequeue_and_dump): Likewise.
8513 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
8514 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
8515 * tree-pretty-print.c (dump_generic_node): Likewise.
8516 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
8517 (simple_iv_with_niters): Likewise.
8518 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
8519 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
8520 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
8521 * tree-ssa-loop-niter.c (split_to_var_and_offset)
8522 (refine_value_range_using_guard, number_of_iterations_ne_max)
8523 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
8524 (get_cst_init_from_scev, record_nonwrapping_iv)
8525 (scev_var_range_cant_overflow): Likewise.
8526 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
8527 * tree-ssa-pre.c (compute_avail): Likewise.
8528 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
8529 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
8530 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
8531 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
8532 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
8533 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
8534 (dump_case_nodes, try_switch_expansion): Likewise.
8535 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
8536 (vect_do_peeling): Likewise.
8537 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
8538 * tree-vect-stmts.c (vectorizable_load): Likewise.
8539 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
8540 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
8541 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
8542 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
8543 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
8544 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
8545 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
8546 (evrp_dom_walker::before_dom_children): Likewise.
8547 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
8548 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
8549 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
8550 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
8551 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
8552 (get_range_pos_neg): Likewise.
8553 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
8554 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
8555 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
8556 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
8557 * config/avr/avr.c (avr_fold_builtin): Likewise.
8558 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
8559 * config/msp430/msp430.c (msp430_attr): Likewise.
8560 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
8561 * config/powerpcspe/powerpcspe-c.c
8562 (altivec_resolve_overloaded_builtin): Likewise.
8563 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
8564 (rs6000_expand_ternop_builtin): Likewise.
8565 * config/rs6000/rs6000-c.c
8566 (altivec_resolve_overloaded_builtin): Likewise.
8567 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
8568 (rs6000_expand_ternop_builtin): Likewise.
8569 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
8571 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8573 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
8574 when copying loop nest with only one inner loop.
8576 2017-10-10 Richard Biener <rguenther@suse.de>
8578 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
8579 blocks if SCEV is active.
8580 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
8582 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
8583 (scev_initialize): Assert we are not yet initialized.
8585 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8587 * tree-loop-distribution.c (generate_loops_for_partition): Remove
8588 inner loop's exit stmt by making it always exit the loop, otherwise
8589 we would generate an infinite empty loop.
8591 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8593 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
8594 renaming variables in new preheader if it's deleted.
8596 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8598 * tree-loop-distribution.c (struct partition): Remove unused field
8599 loops of the structure.
8600 (partition_alloc, partition_free): Ditto.
8601 (build_rdg_partition_for_vertex): Ditto.
8603 2017-10-09 Jeff Law <law@redhat.com>
8605 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
8606 return type to match prototype and documentation.
8608 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8610 * config/rs6000/rs6000.c (processor_costs): Move to ...
8611 * config/rs6000/rs6000.h: ... here.
8612 (rs6000_cost): Declare.
8614 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
8616 * except.c (setjmp_fn): New global variable.
8617 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
8618 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
8619 if DONT_USE_BUILTIN_SETJMP is defined.
8621 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8623 * target.def (insn_cost): New hook.
8624 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
8625 * doc/tm.texi: Regenerate.
8626 * rtlanal.c (insn_cost): Use the new hook.
8628 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8630 * combine.c (combine_validate_cost): Compute the new insn_cost,
8631 not just pattern_cost.
8632 (try_combine): Adjust comment.
8634 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8636 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
8638 * combine.c (uid_insn_cost): Adjust comment.
8639 (combine_validate_cost): Adjust comment. Use pattern_cost instead
8641 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
8642 * dse.c (find_shift_sequence): Ditto.
8643 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
8644 (bb_valid_for_noce_process_p): Use pattern_cost.
8645 * rtl.h (insn_rtx_cost): Delete.
8646 (pattern_cost): New prototype.
8647 (insn_cost): New prototype.
8648 * rtlanal.c (insn_rtx_cost): Rename to...
8649 (pattern_cost): ... this.
8652 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
8654 * config/i386/i386.md (*jcc_2): Remove insn pattern.
8655 (*jcc<mode>_0_r_i387): Ditto.
8656 (*jccxf_r_i387): Ditto.
8657 (*jcc<mode>_r_i387): Ditto.
8658 (*jccu<mode>_r_i387): Ditto.
8659 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
8660 (*jcc): Rename from *jcc_1.
8662 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8664 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
8665 deferred rescans after the lvx/stvx recombination pre-pass.
8667 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8669 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
8670 memory operation instruction support.
8671 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
8672 (rs6000-ibm-aix[789]*): Likewise.
8673 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
8674 Document new functions.
8676 2017-10-09 Richard Biener <rguenther@suse.de>
8678 PR tree-optimization/82397
8679 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
8680 equality only for semantically equal trees.
8682 2017-10-09 Richard Biener <rguenther@suse.de>
8684 PR tree-optimization/82449
8685 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
8686 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
8687 allow constant addresses.
8688 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
8691 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8693 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
8696 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8699 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
8702 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8705 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
8707 2017-10-09 Jakub Jelinek <jakub@redhat.com>
8710 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
8711 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
8713 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
8715 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
8716 (WI_BINARY_PREDICATE_RESULT): Likewise.
8717 (wi::binary_traits::operator_result): New type.
8718 (wi::binary_traits::predicate_result): Likewise.
8719 (generic_wide_int::operator~, unary generic_wide_int::operator-)
8720 (generic_wide_int::operator==, generic_wide_int::operator!=)
8721 (generic_wide_int::operator&, generic_wide_int::and_not)
8722 (generic_wide_int::operator|, generic_wide_int::or_not)
8723 (generic_wide_int::operator^, generic_wide_int::operator+
8724 (binary generic_wide_int::operator-, generic_wide_int::operator*):
8726 (operator~, unary operator-, operator==, operator!=, operator&)
8727 (operator|, operator^, operator+, binary operator-, operator*): New
8729 * expr.c (get_inner_reference): Use wi::bit_and_not.
8730 * fold-const.c (fold_binary_loc): Likewise.
8731 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
8732 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
8733 (bit_value_binop): Likewise.
8734 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
8735 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
8736 (extract_range_from_binary_expr_1): Likewise.
8737 (masked_increment): Likewise.
8738 (simplify_bit_ops_using_ranges): Likewise.
8740 2017-10-09 Martin Jambor <mjambor@suse.cz>
8743 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
8744 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
8745 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
8747 (hsa_fixup_mov_insn_type): New function.
8748 (hsa_op_with_type::get_in_type): Use it.
8749 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
8750 immediates in an assert.
8751 (hsa_op_with_type::extend_int_to_32bit): New method.
8752 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
8753 types. Convert to dest type if necessary.
8754 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
8755 (reg_for_gimple_ssa): Pass false as min32int to
8756 hsa_type_for_scalar_tree_type.
8757 (gen_hsa_addr): Fixup type when creating addresable temporary.
8758 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
8759 (gen_hsa_unary_operation): Extend operands and convert to dest type if
8760 necessary. Call hsa_fixup_mov_insn_type.
8761 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
8762 extend operands and convert to dest type if necessary.
8763 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
8764 to dest type if necessary.
8765 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
8766 if conversion nt necessary and size matches.
8767 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
8768 to dest type if necessary.
8769 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
8770 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
8772 (gen_hsa_clrsb): Likewise.
8773 (gen_hsa_ffs): Likewise.
8774 (gen_hsa_divmod): Extend operands and convert to dest type if
8776 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
8778 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8780 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
8781 Remove empty default arguments. Use a brace block as output
8783 (conditional return): Ditto.
8785 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
8786 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
8788 (group_ending_nop): Ditto.
8789 (doloop_end): Ditto.
8790 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
8791 (splitters for those): Ditto.
8793 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8795 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
8796 a conditional jump (and the compare for it) so that pc_rtx is the
8798 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
8799 for the deleted and renamed ctr<mode>_internal[234] patterns.
8800 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
8801 Delete second conditional return pattern.
8802 (ctr<mode>_internal2): Delete this second bdnz pattern.
8803 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
8804 (ctr<mode>_internal4): Delete this second bdz pattern.
8806 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8808 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
8809 (always_initialized_rtx_for_ssa_name_p): New predicate.
8810 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
8811 (finish_out_of_ssa): Free new field of SA.
8812 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
8813 * tree-ssa-coalesce.c: Include tree-ssa.h.
8814 (get_parm_default_def_partitions): Remove extern keyword.
8815 (get_undefined_value_partitions): New function.
8816 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
8817 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
8819 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
8821 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8823 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
8825 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8827 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
8828 for targets that preffer 128bit.
8830 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8832 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
8834 2017-10-08 Olivier Hainque <hainque@adacore.com>
8836 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
8837 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
8839 (thumb_set_return_address): Likewise.
8841 2017-10-08 Olivier Hainque <hainque@adacore.com>
8843 * common/config/arm/arm-common.c (arm_except_unwind_info):
8844 Handle DWARF2_UNWIND_INFO.
8846 2017-10-07 Michael Collison <michael.collison@arm.com>
8848 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
8851 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
8853 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
8854 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
8855 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
8856 defined, force the creation of a new block for a dispatch label.
8858 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
8860 * invoke.texi (Wsuggest-attribute=cold): Document.
8861 * common.opt (Wsuggest-attribute=cold): New
8862 * ipa-pure-const.c (warn_function_cold): New function.
8863 * predict.c (compute_function_frequency): Use it.
8864 * predict.h (warn_function_cold): Declare.
8866 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
8868 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
8871 2017-10-06 Martin Liska <mliska@suse.cz>
8873 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
8874 keyword for member functions.
8875 (struct sanopt_tree_couple): New struct.
8876 (struct sanopt_tree_couple_hash): New function.
8877 (struct sanopt_ctx): Add new hash_map.
8878 (has_dominating_ubsan_ptr_check): New function.
8879 (record_ubsan_ptr_check_stmt): Likewise.
8880 (maybe_optimize_ubsan_ptr_ifn): Likewise.
8881 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
8882 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
8884 2017-10-06 Sudakshina Das <sudi.das@arm.com>
8887 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
8888 aarch64_simd_valid_immediate on CONST_VECTORs.
8889 (aarch64_reg_or_bic_imm): Likewise.
8891 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
8893 PR rtl-optimization/82396
8894 * haifa-sched.c (ready_sort_real): Disable qsort checking.
8896 2017-10-06 Sebastian Pop <sebpop@gmail.com>
8898 * graphite-dependences.c (scop_get_reads): Move code to...
8899 (scop_get_must_writes): Move code to...
8900 (scop_get_may_writes): Move code to...
8901 (scop_get_reads_and_writes): ... here.
8902 (scop_get_dependences): Call scop_get_reads_and_writes.
8904 2017-10-06 Jakub Jelinek <jakub@redhat.com>
8906 PR tree-optimization/82434
8907 * fold-const.h (can_native_encode_type_p,
8908 can_native_encode_string_p): Remove.
8909 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
8910 don't encode anything, just return what would be otherwise returned.
8911 (native_encode_fixed, native_encode_complex, native_encode_vector):
8913 (native_encode_string): Likewise. Inline by hand
8914 can_native_encode_string_p.
8915 (can_native_encode_type_p): Remove.
8916 (can_native_encode_string_p): Remove.
8917 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
8918 STRING_CSTs using can_native_encode_string_p, test all
8919 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
8920 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
8921 argument from native_encode_expr.
8922 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
8923 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
8926 2017-10-06 Richard Biener <rguenther@suse.de>
8928 PR tree-optimization/82397
8929 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
8930 operand_equal_p but rely on data_ref_compare_tree for detecting
8932 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
8933 to match up with dr_group_sort_cmp.
8935 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8938 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
8940 * config/s390/s390-builtin-types.def: Regenerate.
8942 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8945 * config/s390/s390-builtin-types.def: Regenerate.
8946 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
8947 Change flag from B_VXE to B_VX.
8948 (s390_vec_min_dbl): Remove B_VXE flag.
8950 2017-10-06 Richard Biener <rguenther@suse.de>
8952 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
8953 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
8954 translate_isl_ast_to_gimple::is_valid_rename,
8955 translate_isl_ast_to_gimple::get_rename,
8956 translate_isl_ast_to_gimple::get_def_bb_for_const,
8957 translate_isl_ast_to_gimple::get_new_name,
8958 translate_isl_ast_to_gimple::collect_all_ssa_names,
8959 translate_isl_ast_to_gimple::copy_loop_phi_args,
8960 translate_isl_ast_to_gimple::collect_all_ssa_names,
8961 translate_isl_ast_to_gimple::copy_loop_phi_args,
8962 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
8963 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
8964 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
8965 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
8966 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
8967 translate_isl_ast_to_gimple::copy_cond_phi_args,
8968 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
8969 translate_isl_ast_to_gimple::edge_for_new_close_phis,
8970 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
8971 translate_isl_ast_to_gimple::rename_uses,
8972 translate_isl_ast_to_gimple::rename_all_uses): Remove.
8973 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
8974 (set_rename_for_each_def): Likewise.
8975 (graphite_copy_stmts_from_block): Handle debug stmt resetting
8976 here. Handle rewriting SCEV analyzable uses here.
8977 (copy_bb_and_scalar_dependences): Generate code for PHI
8979 (graphite_regenerate_ast_isl): Adjust.
8980 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
8981 (add_write, add_read): New functions.
8982 (build_cross_bb_scalars_def): Use it and simplify.
8983 (build_cross_bb_scalars_use): Likewise.
8984 (graphite_find_cross_bb_scalar_vars): Inline into...
8985 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
8986 simulating out-of-SSA. Compute liveout and add dependencies.
8987 (build_scops): Force an empty entry block.
8988 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
8990 (sese_build_liveouts): Declare.
8991 (sese_trivially_empty_bb_p): Likewise.
8992 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
8993 compute liveout and debug_liveout.
8994 (sese_bad_liveouts_use): Remove.
8995 (sese_reset_debug_liveouts_bb): Likewise.
8996 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
8997 (sese_build_liveouts): Build liveout and debug_liveout and store
8999 (new_sese_info): Adjust.
9000 (free_sese_info): Likewise.
9001 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9002 do not build liveout here.
9003 (move_sese_in_condition): Adjust region entry.
9004 (scev_analyzable_p): Match up with chrec_apply requirements.
9005 (sese_trivially_empty_bb_p): New.
9006 * tree-into-ssa.c (get_reaching_def): Properly support generating
9007 default-defs for incremental rewrite of anonymous names.
9009 2017-10-06 Richard Biener <rguenther@suse.de>
9011 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9012 precision do not perform modulo reduction.
9014 2017-10-06 Richard Biener <rguenther@suse.de>
9016 PR tree-optimization/82436
9017 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9018 conservatively choose the vectorization factor when checking
9019 whether we can perform the required load permutation.
9020 (vect_transform_slp_perm_load): Assert when we may not fail.
9022 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
9024 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
9025 message for incompatible -msdata=* and -mcall-* options.
9027 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9029 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
9030 rate for post-reload scheduling.
9032 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9034 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
9036 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9038 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
9039 to improve monte carlo in scimark.
9041 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9043 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
9044 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
9045 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
9046 pentium4_cost, nocona_cost): Set reassociation width to 1.
9047 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
9048 width to 2 for fp operations and 1 otherwise.
9049 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
9051 (atom_cost): Set reassociation width to 2.
9052 (slm_cost, generic_cost): Set fp reassociation width
9053 to 2 and 1 otherwise.
9054 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
9055 (core_cost): Set fp reassociation width to 4 and vector to 2.
9056 (ix86_reassociation_width): Rewrite using cost table; special case
9057 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
9058 and TARGET_AVX128_OPTIMAL.
9059 * config/i386/i386.h (processor_costs): Add
9060 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
9061 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
9062 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
9063 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
9064 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
9065 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
9067 2017-10-05 Nathan Sidwell <nathan@acm.org>
9069 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
9071 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9073 * config/arm/arm.c (arm_test_fpu_data): New.
9074 (arm_run_selftests): Call arm_test_fpu_data.
9076 2017-10-04 Nathan Sidwell <nathan@acm.org>
9078 * toplev.c (toplev::main): Remove excess parens on pretty_printer
9080 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
9082 2017-10-04 Sudakshina Das <sudi.das@arm.com>
9084 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
9085 check type for aarch64_simd_valid_immediate.
9086 (aarch64_output_simd_mov_immediate): Update prototype.
9087 (aarch64_simd_valid_immediate): Update prototype.
9088 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
9089 support for ORR-immediate.
9090 (and<mode>3): modified pattern to add support for BIC-immediate.
9091 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
9092 now checks for valid immediate for BIC and ORR based on new enum
9094 (aarch64_output_simd_mov_immediate): Function now used to output
9095 BIC/ORR imm as well based on new enum argument.
9096 * config/aarch64/constraints.md (Do): New vector immediate constraint.
9098 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
9099 (aarch64_reg_or_bic_imm): Likewise.
9101 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9103 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
9104 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
9106 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9110 PR rtl-optimization/82396
9111 * haifa-sched.c (autopref_multipass_init): Simplify
9113 (autopref_rank_data): Simplify sort order.
9114 * sched-int.h (autopref_multipass_data_): Remove
9115 multi_mem_insn_p, min_offset and max_offset.
9117 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9119 * doc/sourcebuild.texi: Document vect_peeling_profitable.
9121 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9123 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
9126 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9128 * doc/sourcebuild.texi: Document vect_long_mult.
9130 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
9132 PR tree-optimization/82413
9133 * fold-const.c (build_range_check): Use widest_int when comparing
9134 the maximum ETYPE value with HIGH.
9136 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9138 PR rtl-optimization/82396
9139 * haifa-sched.c (autopref_multipass_init): Simplify
9141 (autopref_rank_data): Simplify sort order.
9142 * sched-int.h (autopref_multipass_data_): Remove
9143 multi_mem_insn_p, min_offset and max_offset.
9145 2017-10-04 Jakub Jelinek <jakub@redhat.com>
9147 PR tree-optimization/82381
9148 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
9149 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
9152 PR tree-optimization/82374
9153 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
9154 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
9155 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
9156 current_function_decl to the new decl.
9158 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9160 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
9161 helper macro for IEEE float128 hardware built-in functions.
9162 (SQRTF128_ODD): Add built-in functions with the round-to-odd
9164 (TRUNCF128_ODD): Likewise.
9165 (ADDF128_ODD): Likewise.
9166 (SUBF128_ODD): Likewise.
9167 (MULF128_ODD): Likewise.
9168 (DIVF128_ODD): Likewise.
9169 (FMAF128_ODD): Likewise.
9170 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
9171 UNSPEC_TRUNC_ROUND_TO_ODD.
9172 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
9173 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
9174 floating point round to odd instructions.
9175 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
9176 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
9177 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
9178 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
9179 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
9180 (trunc<mode>sf2_hw): Change the truncate with round to odd
9181 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
9182 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
9183 to odd hardware instructions.
9184 (sub<mode>3_odd): Likewise.
9185 (mul<mode>3_odd): Likewise.
9186 (div<mode>3_odd): Likewise.
9187 (sqrt<mode>2_odd): Likewise.
9188 (fma<mode>4_odd): Likewise.
9189 (fms<mode>4_odd): Likewise.
9190 (nfma<mode>4_odd): Likewise.
9191 (nfms<mode>4_odd): Likewise.
9192 (trunc<mode>df2_odd): Change the truncate with round to odd
9193 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
9195 * doc/extend.texi (PowerPC built-in functions): Update documentation
9196 for existing IEEE float128-bit built-in functions. Add built-in
9197 functions that generate the IEEE 128-bit floating point round to
9200 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
9202 PR rtl-optimization/77729
9203 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
9204 to (X&(C1&~C2))|C2 transformations.
9206 2017-10-03 Martin Jambor <mjambor@suse.cz>
9208 PR tree-optimization/82363
9209 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
9210 mismatch, mark lacc written regardless of racc.
9212 2017-10-03 Jakub Jelinek <jakub@redhat.com>
9214 PR tree-optimization/82381
9215 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
9216 stmt_to_insert nor wheather SSA_NAMEs are default defs.
9217 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
9218 fallthrough into reassoc_stmt_dominates_stmt_p.
9221 * combine.c (combine_instructions): Don't combine in unreachable
9224 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9227 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
9228 function to not use the have_cpu variable. Do not set cpu_index,
9229 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
9231 (rs6000_valid_attribute_p): Remove duplicate initializations of
9232 old_optimize and func_optimize.
9233 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
9234 (rs6000_activate_target_options): Make global.
9235 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
9238 2017-10-02 Jakub Jelinek <jakub@redhat.com>
9240 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
9241 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
9242 Set *poffset to 0 and *psize and *pmax_size to -1 if
9243 *poffset + *psize overflows in HOST_WIDE_INT.
9245 PR tree-optimization/82387
9246 PR tree-optimization/82388
9247 PR tree-optimization/82389
9248 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
9249 instead of live_bytes non-NULL.
9251 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
9254 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
9257 2017-10-02 Richard Biener <rguenther@suse.de>
9259 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
9260 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
9261 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
9263 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9265 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
9266 requested precision matches the type's.
9267 * calls.c (alloc_max_size): Calculate the new candidate size as
9268 a widest_int and use wi::to_widest when comparing it with the
9269 current candidate size.
9270 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
9271 zero rather than integer_zero_node.
9272 * match.pd: Check for a no-op conversion before using wi::add
9273 rather than after. Use tree_to_uhwi when summing small shift
9274 counts into an unsigned int.
9276 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9277 Alan Hayward <alan.hayward@arm.com>
9278 David Sherwood <david.sherwood@arm.com>
9281 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
9282 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
9283 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
9284 POINTER_AND_FP_REGS.
9286 2017-10-02 Richard Biener <rguenther@suse.de>
9288 PR tree-optimization/82355
9289 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9290 a mapping for the enclosing loop but avoid generating one for
9292 (copy_bb_and_scalar_dependences): Remove premature codegen
9293 error on PHIs in blocks duplicated into multiple places.
9294 * graphite-scop-detection.c
9295 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9296 in the region use it as loop and nest to analyze the DR in.
9297 (try_generate_gimple_bb): Likewise.
9298 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9299 (add_loop_constraints): For blocks in a loop not in the region
9300 create a dimension with a single iteration.
9301 * sese.h (gbb_loop_at_index): Remove assert.
9303 2017-10-01 Kevin Buettner <kevinb@redhat.com>
9305 * omp-expand.c (adjust_context_scope): New function.
9306 (expand_parallel_call): Call adjust_context_scope.
9308 2017-10-01 Jeff Law <law@redhat.com>
9310 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
9311 dom_opt_dom_walker class with direct access to private members.
9312 Add comments. Call test_for_singularity.
9313 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
9314 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
9315 m_dummy_cond anymore.
9316 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
9318 (pass_dominator:execute): Build the dummy_cond here and pass it
9319 to the dom_opt_dom_walker ctor.
9320 (test_for_singularity): New function.
9322 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
9323 Maya Rashish <coypu@sdf.org>
9325 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
9326 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
9327 (alpha*-*-netbsd*) Use nbsd_tm_file.
9328 (arm*-*-netbsdelf*) Likewise.
9329 (i[34567]86-*-netbsdelf*) Likewise.
9330 (x86_64-*-netbsd*) Likewise.
9331 (mips*-*-netbsd*) Likewise.
9332 (powerpc-*-netbsd*) Likewise.
9333 (sh*-*-netbsd*) Likewise.
9334 (sparc-*-netbsdelf*) Likewise.
9335 (sparc64-*-netbsd*) Likewise.
9336 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
9338 (vax-*-netbsdelf*) Likewise.
9339 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
9340 (UINT_FAST8_TYPE) Likewise.
9341 (INT_FAST16_TYPE) Check CHAR_FAST16.
9342 (UINT_FAST16_TYPE) Likewise.
9344 2017-09-30 Jakub Jelinek <jakub@redhat.com>
9347 * config/i386/i386.md
9348 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
9349 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
9350 *divmodsi4_zext_2): New define_insn_and_split.
9351 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
9352 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
9353 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
9354 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
9355 New define_insn_and_split.
9356 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
9357 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
9358 operands[1] having DImode when mode is SImode.
9360 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
9361 always SImode for DIV and MOD in REG_EQUAL notes.
9363 2017-09-29 Yury Gribov <tetra2005@gmail.com>
9366 * match.pd: Fix handling of NaNs in pattern.
9368 2017-09-29 Jeff Law <law@redhat.com>
9370 * sbitmap.c (bitmap_bit_in_range_p): New function.
9371 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
9372 * tree-ssa-dse.c (live_bytes_read): New function.
9373 (dse_classify_store): Ignore reads of dead bytes.
9375 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
9376 typos and whitespace errors.
9377 * config/i386/predicates.md (address_no_seg_operand): Likewise.
9378 * config/s390/s390.c (s390_emit_prologue): Likewise.
9380 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9383 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
9384 with a symbol for LRA.
9386 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9388 PR rtl-optimization/82338
9389 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
9391 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
9393 * genmodes.c (calc_wider_mode): Suppress qsort macro.
9394 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
9395 (qsort_chk): Declare.
9396 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
9397 (qsort_chk): New function.
9399 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9401 PR tree-optimization/82337
9402 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9403 phi definition if the PHI result appears in an abnormal PHI.
9404 (find_basis_for_base_expr): Don't record a basis if the LHS of the
9405 basis appears in an abnormal PHI.
9407 2017-09-29 Richard Biener <rguenther@suse.de>
9409 * graphite-isl-ast-to-gimple.c
9410 (translate_isl_ast_to_gimple::set_codegen_error): New function.
9411 (binary_op_to_tree): Use it.
9412 (get_rename_from_scev): Likewise.
9413 (copy_loop_phi_nodes): Likewise.
9414 (copy_bb_and_scalar_dependences): Likewise.
9415 (translate_pending_phi_nodes): Likewise.
9417 2017-09-29 Jakub Jelinek <jakub@redhat.com>
9420 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
9421 for movabsq $(i32 << shift), r64.
9423 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
9425 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
9426 index when encoding %esp as %rsp to avoid 0x67 prefix.
9428 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
9430 * config/i386/i386.md (*movsf_internal, *movdf_internal):
9431 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
9433 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
9435 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
9436 Extensions with more than 16 double VFP registers.
9437 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
9438 to_clear_mask and all code related to it. Replace the remaining
9439 entry by a sbitmap and adapt code accordingly.
9441 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
9443 * brig-builtins.def: Change pure attributes to const.
9445 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9447 * config.gcc (default_gnu_indirect_function): Default to yes for
9448 sparc*-*-linux* with glibc.
9450 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9452 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
9453 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
9454 when creating .init_array and .fini_array sections with priority
9457 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
9460 * config/aarch64/aarch64.c
9461 (aarch64_builtin_support_vector_misalignment): Always return false
9462 when misalignment is unknown.
9464 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9466 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
9467 this function to return false if the definition used by the swap
9468 instruction is artificial, or if the memory address from which the
9469 constant value is loaded is not represented by a base address held
9470 in a register or if the base address register is a frame or stack
9471 pointer. Additionally, return false if the base address of the
9472 loaded constant is a SYMBOL_REF but is not considered to be a
9474 (replace_swapped_load_constant): New function.
9475 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
9476 loaded constant vector with a load of a swapped constant vector.
9478 2017-09-27 Carl Love <cel@us.ibm.com>
9480 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
9481 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
9482 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
9485 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
9487 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
9488 first, always call autopref_rank_data otherwise.
9490 2017-09-27 Richard Biener <rguenther@suse.de>
9492 * graphite-scop-detection.c (find_scop_parameters): Move
9493 loop bound handling ...
9494 (gather_bbs::before_dom_children): ... here, avoiding the need
9495 to build scop_info->loop_nest.
9496 (record_loop_in_sese): Remove.
9497 * sese.h (sese_info_t::loop_nest): Remove.
9498 * sese.c (new_sese_info): Do not allocate loop_nest.
9499 (free_sese_info): Do not free loop_nest.
9501 2017-09-27 Jakub Jelinek <jakub@redhat.com>
9504 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
9505 lhs from calls if the lhs has addressable type.
9507 2017-09-27 Richard Biener <rguenther@suse.de>
9509 * graphite.h (scop::max_alias_set): New member.
9510 * graphite-scop-detection.c: Remove references to non-existing
9511 --param in comments.
9512 (build_alias_sets): Record the maximum alias set used for drs.
9513 (build_scops): Support zero as unlimited for
9514 --param graphite-max-arrays-per-scop.
9515 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
9517 (build_poly_sr_1): ... here. Compute alias set based on the
9518 maximum alias set used for drs rather than
9519 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
9521 2017-09-27 Richard Biener <rguenther@suse.de>
9523 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
9524 --param loop-block-tile-size=0 to disable tiling.
9526 2017-09-27 Richard Biener <rguenther@suse.de>
9528 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
9529 (graphite-max-nb-scop-params): Document special value zero.
9530 * domwalk.h (dom_walker::STOP): New symbolical constant.
9531 (dom_walker::dom_walker): Add optional parameter for bb to
9533 (dom_walker::~dom_walker): Declare.
9534 (dom_walker::before_dom_children): Document STOP return value.
9535 (dom_walker::m_user_bb_to_rpo): New member.
9536 (dom_walker::m_bb_to_rpo): Likewise.
9537 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
9538 mapping here if not provided by the user.
9539 (dom_walker::~dom_walker): Free bb to RPO mapping if not
9540 provided by the user.
9541 (dom_walker::STOP): Define.
9542 (dom_walker::walk): Do not compute bb to RPO mapping here.
9543 Support STOP return value from before_dom_children to stop
9545 * graphite-optimize-isl.c (optimize_isl): If the schedule
9546 is the same still generate code if -fgraphite-identity
9547 or -floop-parallelize-all are given.
9548 * graphite-scop-detection.c: Include cfganal.h.
9549 (gather_bbs::gather_bbs): Get and pass through bb to RPO
9551 (gather_bbs::before_dom_children): Return STOP for BBs
9553 (build_scops): Compute bb to RPO mapping and pass it to
9554 the domwalk. Treat --param graphite-max-nb-scop-params=0
9555 as not limiting the number of params.
9556 * graphite.c (graphite_initialize): Remove limit on the
9557 number of basic-blocks in a function.
9558 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
9559 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
9560 default value of 10.
9562 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9564 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
9565 Adjust code to eliminate needing to do the shift right 32-bits
9566 operation after XSCVDPSPN.
9568 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9570 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
9571 ((X / Y) != 0 -> X >= Y): Likewise.
9573 2017-09-26 Carl Love <cel@us.ibm.com>
9575 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
9576 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
9577 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
9578 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
9579 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
9580 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
9581 definitions and overloading.
9582 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
9583 statement for P9V_BUILTIN_XST_LEN_R.
9584 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
9585 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
9586 define_expand and define_insn for the instructions and builtins.
9587 * doc/extend.texi: Update the built-in documentation file for the new
9589 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
9590 define_insn for the instructions
9592 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
9595 * gcc/config/netbsd-protos.h: New file.
9596 * gcc/config/netbsd.c: New file.
9597 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
9598 * gcc/config/t-netbsd: New file.
9599 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
9600 (tmake_file) Add t-netbsd.
9601 (extra_objs) Add netbsd.o.
9603 2017-09-26 Janus Weil <janus@gcc.gnu.org>
9607 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
9609 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9611 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
9612 sign extension from a vector register to a GPR by doing a 32-bit
9613 direct move and then an EXTSW.
9614 (extendsi<mode>2 splitter): Likewise.
9615 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
9616 right or vector extract after doing XSCVDPSPN. Use
9617 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
9619 (movdi_from_sf_zero_ext): Likewise.
9620 (reload_gpr_from_vsxsf): Likewise.
9621 (p8_mfvsrd_4_disf): Delete, no longer used.
9622 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
9623 then needing to move the SFmode to a GPR to use the XSCVDPSP
9624 instruction instead of FRSP and XSCVDPSPN.
9625 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
9626 it is adjacent to the other XSCVSPDP insns.
9627 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
9628 SFmode to be in traditional Altivec registers.
9629 (vsx_xscvdpspn): Eliminate useless alternative constraint.
9630 (vsx_xscvspdpn): Likewise.
9631 (vsx_xscvspdpn_scalar): Likewise.
9633 2017-09-26 Martin Jambor <mjambor@suse.cz>
9635 * tree-sra.c (compare_access_positions): Put integral types first,
9636 stabilize sorting of integral types, remove conditions putting
9637 non-full-precision integers last.
9638 (sort_and_splice_var_accesses): Disable scalarization if a
9639 non-integert would be represented by a non-full-precision integer.
9641 2017-09-26 Joseph Myers <joseph@codesourcery.com>
9643 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
9644 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
9645 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
9646 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
9647 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
9648 conditionals inside the function instead of around it. Call
9649 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
9650 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
9652 2017-09-26 Richard Biener <rguenther@suse.de>
9654 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
9656 (scop_detection::build_scop_breadth): ... this. Removed.
9657 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
9658 (scop_detection::harmful_stmt_in_bb): Likewise.
9659 (scop_detection::graphite_can_represent_stmt): Likewise.
9660 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
9661 (scop_detection::can_represent_loop): Remove recursion, fold in ...
9662 (scop_detection::can_represent_loop_1): ... this. Removed.
9663 (scop_detection::harmful_loop_in_region): Simplify after inlining
9664 the above and remove more quadraticness.
9665 (build_scops): Adjust.
9666 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
9669 2017-09-26 Jakub Jelinek <jakub@redhat.com>
9672 * config/i386/i386.c (ix86_print_operand_address_as): Only test
9673 REGNO (base) == SP_REG if base is a REG.
9676 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
9677 if it is different SSA_NAME.
9678 (optimize_range_tests_cmp_bitwise): New function.
9679 (optimize_range_tests): Call it.
9681 2017-09-26 Richard Biener <rguenther@suse.de>
9683 PR tree-optimization/82321
9684 * graphite.c (canonicalize_loop_closed_ssa): Properly check
9685 for the def being inside the loop.
9687 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9689 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
9691 * config/s390/s390-builtins.def: Fix constraint on op4.
9693 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9695 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
9696 independent expanders.
9697 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
9698 ("vec_ordered", "vec_unordered"): New expanders.
9700 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9702 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
9705 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9707 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
9708 vec_unpacks_lo_v16qi.
9709 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
9711 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9713 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
9714 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
9715 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
9717 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9719 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
9721 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
9723 ("*vec_slb<mode>"): New insn pattern.
9724 ("vec_shr_<mode>"): New expander.
9725 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
9726 and force the shift count operand to V16QImode.
9727 ("vec_srb<mode>"): Set shift count mode to V16QI.
9729 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9731 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
9732 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
9733 ("vec_widen_smult_hi_<mode>"): New expander definitions.
9735 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
9738 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
9740 2017-09-26 Richard Biener <rguenther@suse.de>
9742 PR tree-optimization/82320
9743 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
9746 2017-09-25 Jeff Law <law@redhat.com>
9748 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
9749 prototype for new argument.
9750 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
9751 mostly extracted from rs6000_emit_allocate_stack.
9752 (rs6000_emit_probe_stack_range_stack_clash): New function.
9753 (rs6000_emit_allocate_stack): Call
9754 rs6000_emit_probe_stack_range_stack_clash as needed.
9755 (rs6000_emit_probe_stack_range): Add additional argument
9756 to call to gen_probe_stack_range{si,di}.
9757 (output_probe_stack_range): New.
9758 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
9759 (output_probe_stack_range_stack_clash): New.
9760 (rs6000_emit_prologue): Emit notes into dump file as requested.
9761 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
9762 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
9763 Add additional operand and pass it to output_probe_stack_range.
9765 2017-09-25 Bin Cheng <bin.cheng@arm.com>
9767 PR tree-optimization/82163
9768 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
9769 (checking_verify_loop_closed_ssa): New parameter.
9770 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
9771 (check_loop_closed_ssa_stmt): Delete.
9772 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
9773 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
9774 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
9777 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
9779 * brig-builtins.def: Treat HSAIL barrier builtins as
9780 setjmp/longjump style functions.
9782 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9784 * target.def (constant_alignment): New hook.
9785 * defaults.h (CONSTANT_ALIGNMENT): Delete.
9786 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
9787 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
9788 * doc/tm.texi: Regenerate.
9789 * targhooks.h (default_constant_alignment): Declare.
9790 (constant_alignment_word_strings): Likewise.
9791 * targhooks.c (default_constant_alignment): New function.
9792 (constant_alignment_word_strings): Likewise.
9793 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
9794 instead of CONSTANT_ALIGNMENT.
9795 * varasm.c (align_variable, get_variable_align, build_constant_desc)
9796 (force_const_mem): Likewise.
9797 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
9798 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
9799 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
9800 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9801 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
9803 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
9804 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9805 constant_alignment_word_strings.
9806 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
9807 (CONSTANT_ALIGNMENT): Likewise.
9808 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9809 (arm_constant_alignment): New function.
9810 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
9811 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9812 constant_alignment_word_strings.
9813 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
9814 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9815 constant_alignment_word_strings.
9816 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
9817 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9818 (cris_constant_alignment): New function.
9819 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
9820 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9821 (epiphany_constant_alignment): New function.
9822 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
9823 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9824 constant_alignment_word_strings.
9825 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
9826 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9827 constant_alignment_word_strings.
9828 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
9829 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9830 constant_alignment_word_strings.
9831 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
9832 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
9833 * config/i386/i386.c (ix86_constant_alignment): Make static.
9834 Use the same interface as the target hook.
9835 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9836 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
9837 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9838 constant_alignment_word_strings.
9839 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
9840 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
9841 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9842 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
9843 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9844 constant_alignment_word_strings.
9845 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
9846 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9847 constant_alignment_word_strings.
9848 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
9849 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9850 constant_alignment_word_strings.
9851 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
9852 * config/microblaze/microblaze.c (microblaze_constant_alignment):
9854 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9855 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
9856 * config/mips/mips.c (mips_constant_alignment): New function.
9857 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9858 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
9859 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
9860 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9861 (mmix_constant_alignment): Make static. Use the same interface
9863 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
9864 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9865 constant_alignment_word_strings.
9866 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
9867 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9868 constant_alignment_word_strings.
9869 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
9870 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9871 constant_alignment_word_strings.
9872 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
9873 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9874 (rs6000_constant_alignment): New function.
9875 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
9876 * config/riscv/riscv.c (riscv_constant_alignment): New function.
9877 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9878 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
9879 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9880 (rs6000_constant_alignment): New function.
9881 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
9882 * config/s390/s390.c (s390_constant_alignment): New function.
9883 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9884 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
9885 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9886 constant_alignment_word_strings.
9887 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
9888 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9889 (sparc_constant_alignment): New function.
9890 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
9891 * config/spu/spu.c (spu_constant_alignment): New function.
9892 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9893 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
9894 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9895 constant_alignment_word_strings.
9896 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
9897 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9898 constant_alignment_word_strings.
9899 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
9900 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9901 constant_alignment_word_strings.
9902 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
9903 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9904 (visium_constant_alignment): New function.
9905 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
9906 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9907 (xtensa_constant_alignment): New function.
9908 * system.h (CONSTANT_ALIGNMENT): Poison.
9910 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
9912 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9913 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
9914 (rs6000_builtin_valid_without_lhs): New helper function.
9915 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9916 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
9918 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9920 * target.h (vec_perm_indices): Use unsigned short rather than
9922 (auto_vec_perm_indices): Likewise.
9923 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
9924 Use unsigned int rather than unsigned char.
9925 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
9927 2017-09-25 Richard Biener <rguenther@suse.de>
9929 * cfgloop.h (sort_sibling_loops): Declare.
9930 * cfgloop.c (sort_sibling_loops_cmp): New helper.
9931 (sort_sibling_loops): New function sorting the sibling loop list
9933 * graphite.c (graphite_transform_loops): Sort sibling loops.
9935 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
9937 * target.def (vec_perm_const_ok): Change sel parameter to
9939 * optabs-query.c (can_vec_perm_p): Update accordingly.
9940 * doc/tm.texi: Regenerate.
9941 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
9942 auto_vec_perm_indices and remove separate nelt field.
9943 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
9944 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
9945 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
9946 (aarch64_expand_vec_perm_const): Update accordingly.
9947 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
9948 to vec_perm_indices.
9949 * config/arm/arm.c (expand_vec_perm_d): Change perm to
9950 auto_vec_perm_indices and remove separate nelt field.
9951 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
9952 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
9953 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
9955 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
9956 to vec_perm_indices.
9957 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
9958 sel to vec_perm_indices.
9959 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
9960 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
9961 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
9963 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
9966 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
9969 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
9970 on the FUNCTION_DECL function context if it has a DIE that is a
9973 2017-09-25 Richard Biener <rguenther@suse.de>
9975 PR tree-optimization/82285
9976 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
9979 2017-09-25 Tom de Vries <tom@codesourcery.com>
9983 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
9986 2017-09-25 Richard Biener <rguenther@suse.de>
9988 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
9989 ISL errors other than isl_error_quota happen. Dump if the
9990 schedule is the same.
9991 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
9992 errors instead of aborting inside ISL.
9994 2017-09-25 Iain Sandoe <iain@codesourcery.com>
9997 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
9998 of libgcc_eh for m64.
9999 * config/i386/darwin64.h: Likewise.
10001 2017-09-25 Richard Biener <rguenther@suse.de>
10003 PR middle-end/82144
10004 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10005 attribute for incomplete types nor twice for complete ones.
10007 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10010 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10011 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10014 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10017 * config/i386/i386.opt (mprefer-avx256): Use
10018 ix86_target_flags variable.
10019 * config/i386/i386.c (ix86_target_string): Move
10020 -mprefer-avx256 to flag2_opts.
10022 2017-09-22 Jakub Jelinek <jakub@redhat.com>
10024 PR middle-end/35691
10025 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
10026 and x != -1 | y != -1 into (x & y) != -1.
10028 2017-09-22 Steve Ellcey <sellcey@cavium.com>
10030 * config.gcc: Add new case statement to set
10031 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
10032 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
10033 s390x-*-linux* case statements. Added aarch64 to the list of
10034 supported architectures.
10036 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10038 PR tree-optimization/82289
10039 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
10040 STMT_VINFO_RELEVANT_P.
10042 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10043 Alan Hayward <alan.hayward@arm.com>
10044 David Sherwood <david.sherwood@arm.com>
10046 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
10047 for VR_RANGE only; don't allow VR_ANTI_RANGE.
10048 (extract_range_from_binary_expr_1): Don't call
10049 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
10051 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10052 Alan Hayward <alan.hayward@arm.com>
10053 David Sherwood <david.sherwood@arm.com>
10055 * target.def (preferred_vector_alignment): New hook.
10056 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
10058 * doc/tm.texi: Regenerate.
10059 * targhooks.h (default_preferred_vector_alignment): Declare.
10060 * targhooks.c (default_preferred_vector_alignment): New function.
10061 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
10063 (DR_TARGET_ALIGNMENT): New macro.
10064 (aligned_access_p): Update commentary.
10065 (vect_known_alignment_in_bytes): New function.
10066 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
10068 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
10069 Calculate the misalignment based on the target alignment rather than
10071 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
10072 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
10073 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
10074 the target alignment, rather than masking the element misalignment
10075 with the number of elements in a vector. Also use the target
10076 alignment when calculating the maximum number of peels.
10077 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
10078 instead of TYPE_ALIGN_UNIT.
10079 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
10080 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
10081 (vect_create_addr_base_for_vector_ref): Update call accordingly.
10082 (vect_create_data_ref_ptr): Likewise.
10083 (vect_setup_realignment): Realign by ANDing with
10084 -DR_TARGET_MISALIGNMENT.
10085 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
10086 the number of peels based on DR_TARGET_ALIGNMENT.
10087 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
10088 with the guaranteed alignment boundary when deciding whether
10090 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
10091 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
10092 (ensure_base_align): Remove stmt_info parameter. Get the
10093 target base alignment from DR_TARGET_ALIGNMENT.
10094 (vectorizable_store): Update call accordingly. Interpret
10095 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
10097 (vectorizable_load): Likewise.
10099 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10100 Alan Hayward <alan.hayward@arm.com>
10101 David Sherwood <david.sherwood@arm.com>
10103 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
10104 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
10105 (vect_enhance_data_refs_alignment): Likewise.
10107 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10109 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10110 error. Only quit immediately if parsing is complete.
10111 (BEGIN): Initialize fatal_err and parse_done.
10112 (begin fpu, end fpu): Check number of arguments.
10113 (begin arch, end arch): Likewise.
10114 (begin cpu, end cpu): Likewise.
10115 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10116 (optalias): Likewise.
10118 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10120 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10121 * config/arm/arm-isa.h: Delete. Move definitions to ...
10122 * arm-cpus.in: ... here. Use new feature and fgroup values.
10123 * config/arm/arm.c (arm_option_override): Use lower case for feature
10125 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10126 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10127 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10129 (print_isa_bits_for): New function.
10130 (gen_isa): New function.
10131 (gen_comm_data): Use print_isa_bits_for.
10132 (define feature): New keyword.
10133 (define fgroup): New keyword.
10134 * config/arm/t-arm (TM_H): Remove.
10135 (GTM_H): Add arm-isa.h.
10136 (arm-isa.h): Add rule to generate file.
10137 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10138 case for feature bit names.
10140 2017-09-22 Richard Biener <rguenther@suse.de>
10142 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
10144 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
10145 print of no dependency loops ...
10146 * graphite.c (graphite_transform_loops): ... here.
10147 (canonicalize_loop_closed_ssa_form): Work from inner to outer
10149 (same_close_phi_node, remove_duplicate_close_phi,
10150 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
10151 (canonicalize_loop_closed_ssa): ... here and simplify.
10152 * graphite-optimize-isl.c: Include tree-vectorizer.h.
10153 (optimize_isl): Use dump_printf_loc to tell when we stopped
10154 optimizing because of an ISL timeout.
10156 2017-09-22 Richard Biener <rguenther@suse.de>
10158 PR tree-optimization/82291
10159 * tree-if-conv.c (predicate_mem_writes): Make sure to
10160 remove writes in blocks predicated with false.
10162 2017-09-22 Richard Biener <rguenther@suse.de>
10164 * sese.c: Include cfganal.h.
10165 (if_region_set_false_region): Remove.
10166 (create_if_region_on_edge): Likewise.
10167 (move_sese_in_condition): Re-implement without destroying
10170 2017-09-22 Richard Biener <rguenther@suse.de>
10172 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
10173 Verify both BBs contain loop PHI nodes before dispatching to
10174 copy_loop_phi_args.
10175 (graphite_regenerate_ast_isl): Do not recompute dominators,
10176 do not verify three times. Restructure for clarity.
10177 * graphite-scop-detection.c (same_close_phi_node,
10178 remove_duplicate_close_phi, make_close_phi_nodes_unique,
10179 defined_in_loop_p, canonicalize_loop_closed_ssa,
10180 canonicalize_loop_closed_ssa_form): Simplify, remove excess
10181 checking and SSA rewrite, move to ...
10182 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
10183 (graphite_initialize): Do not pass in ctx, do not reset the
10184 SCEV cache, compute only dominators.
10185 (graphite_transform_loops): Allocate ISL ctx after
10186 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
10187 Maintain post-dominators only around build_scops.
10188 * sese.c (if_region_set_false_region): Make static. Free
10189 and recompute dominators.
10190 (move_sese_in_condition): Assert we don't get called with
10191 post-dominators computed.
10192 * sese.h (if_region_set_false_region): Remove.
10194 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
10196 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
10197 mode attribute for TARGET_AVX512VL.
10199 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
10201 * config/i386/i386.opt (mprefer-avx256): New option.
10202 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
10204 (ix86_preferred_simd_mode): Return 256-bit AVX modes
10205 for TARGET_PREFER_AVX256.
10206 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
10208 2017-09-21 Jeff Law <law@redhat.com>
10210 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
10211 Fix dump output if the only stack space is for pushed registers.
10213 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10215 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
10218 2017-09-21 Martin Sebor <msebor@redhat.com>
10221 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
10222 code (in C++) or code that triggers warnings.
10224 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
10226 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
10228 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
10230 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
10231 * sched-rgn.c: Ditto.
10232 * sel-sched-ir.c: Ditto.
10234 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
10236 * toplev.h (set_random_seed): Adjust return type.
10237 * toplev.c (init_local_tick): Move eager initialization of random_seed
10238 to get_random_seed. Adjust comment.
10239 (init_random_seed): Inline to get_random_seed, delete.
10240 (get_random_seed): Initialize random_seed lazily.
10241 (set_random_seed): Do not return previous value.
10242 (print_switch_value): Do not call get_random_seed.
10244 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
10246 * cgraph.c (delete_function_version): New, broken out from...
10247 (cgraph_node::delete_function_version): ...here. Rename to
10248 cgraph_node::delete_function_version_by_decl. Update all uses.
10249 (cgraph_node::remove): Call delete_function_version.
10251 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10254 * tree-inline.c (expand_call_inline): Emit clobber stmts for
10255 VAR_DECLs to which addressable non-volatile parameters are mapped
10256 and for id->retvar after the return value assignment. Clear
10257 id->retval and id->retbnd after inlining.
10259 2017-09-21 Richard Biener <rguenther@suse.de>
10261 PR tree-optimization/82276
10262 PR tree-optimization/82244
10263 * tree-vrp.c (build_assert_expr_for): Set
10264 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
10266 (remove_range_assertions): Revert earlier change.
10268 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
10271 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
10273 2017-09-21 Richard Biener <rguenther@suse.de>
10275 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10276 Restore valid IL after code generation errors.
10277 * graphite.c (graphite_transform_loops): Diagnose code
10278 generation issues as MSG_MISSED_OPTIMIZATION and continue
10279 with processing SCOPs.
10281 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10282 Alan Hayward <alan.hayward@arm.com>
10283 David Sherwood <david.sherwood@arm.com>
10285 * calls.c (compute_argument_addresses): Use simplify_gen_binary
10286 rather than choosing between plus_constant and gen_rtx_<CODE>.
10287 * expr.c (emit_push_insn): Likewise.
10288 (expand_expr_real_2): Likewise.
10290 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10291 Alan Hayward <alan.hayward@arm.com>
10292 David Sherwood <david.sherwood@arm.com>
10294 * loop-unroll.c (split_iv): Call copy_rtx on the step.
10296 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10297 Alan Hayward <alan.hayward@arm.com>
10298 David Sherwood <david.sherwood@arm.com>
10300 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
10301 calling tree_to_uhwi.
10303 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10304 Alan Hayward <alan.hayward@arm.com>
10305 David Sherwood <david.sherwood@arm.com>
10307 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
10308 INTEGER_CST rather than a negative test for ADDR_EXPR.
10310 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10311 Alan Hayward <alan.hayward@arm.com>
10312 David Sherwood <david.sherwood@arm.com>
10314 * tree-vrp.c (extract_range_from_binary_expr_1): Check
10315 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
10317 2017-09-21 Richard Biener <rguenther@suse.de>
10319 PR tree-optimization/71351
10320 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
10321 graphite_create_new_loop_guard): Remove, fold remaining parts
10323 (translate_isl_ast_node_for): ... here and simplify.
10325 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10328 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
10329 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
10330 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
10331 alternative always use QI mode, for -Os imov (=R,R) alternative
10332 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
10335 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10336 Jeff Law <law@redhat.com>
10338 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
10339 (allocate_stack_space): New function, partially extracted from
10340 s390_emit_prologue.
10341 (s390_emit_prologue): Track offset to most recent stack probe.
10342 Code to allocate space moved into allocate_stack_space.
10343 Dump actions when no stack is allocated.
10344 (s390_prologue_plus_offset): New function.
10345 (s390_emit_stack_probe): Likewise.
10347 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
10349 * common.opt (Wa, Wl, Wp, g, gz=): Add
10351 (gno-column-info): Remove.
10352 (gcolumn-info): Drop RejectNegative.
10353 (gno-): New prefix.
10354 (gno-record-gcc-switches): Remove.
10355 (grecord-gcc-switches): Drop RejectNegative.
10356 (gno-split-dwarf): Remove.
10357 (gsplit-dwarf): Drop RejectNegative.
10358 (gno-strict-dwarf): Remove.
10359 (gstrict-dwarf): Drop RejectNegative.
10360 * config/darwin.opt (gfull, gused): Add RejectNegative.
10361 * dwarf2out.c (gen_producer_string): Drop
10362 gno-record-gcc-switches handler.
10363 * optc-gen.awk: Add g to prefixes with negative forms.
10364 * opts-common.c (remapping_prefix_p): New.
10365 (find_opt): Check it.
10366 (generate_canonical_option): Test g prefix.
10367 (option_map): Add -gno- mapping.
10368 (add_misspelling_candidates): Check remapping_prefix_p.
10370 2017-09-20 Jeff Law <law@redhat.com>
10372 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
10373 thinko in stack clash protection support.
10375 * explow.c (compute_stack_clash_protection_loop_data): Use
10376 CONST_INT_P instead of explicit test. Verify object is a
10377 CONST_INT_P before looking at INTVAL.
10378 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
10379 instead of explicit test.
10381 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
10384 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
10385 address instead of to r1 and r11.
10387 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
10389 * config.gcc: Support "knm".
10390 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
10391 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10393 * config/i386/i386.c (m_KNM): Define.
10394 (processor_target_table): Add "knm".
10396 (ix86_option_override_internal): Add "knm".
10397 (ix86_issue_rate): Add PROCESSOR_KNM.
10398 (ix86_adjust_cost): Ditto.
10399 (ia32_multipass_dfa_lookahead): Ditto.
10400 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
10401 (fold_builtin_cpu): Add M_INTEL_KNM.
10402 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
10403 (processor_type): Add PROCESSOR_KNM.
10404 * config/i386/x86-tune.def: Add m_KNM.
10405 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
10407 2017-09-20 Richard Biener <rguenther@suse.de>
10409 PR tree-optimization/80213
10410 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
10411 are allowed in empty BBs as well.
10412 (canonicalize_loop_closed_ssa): Also look for other complex
10414 (scop_detection::get_sese): Include the loop-closed PHI block
10416 (scop_detection::merge_sese): Remove code adding extra blocks.
10417 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
10418 (build_scops): Assert the final returned scop is invalid.
10420 2017-09-20 Richard Biener <rguenther@suse.de>
10422 PR tree-optimization/82264
10423 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
10425 (vn_phi_lookup): Likewise.
10426 (vn_phi_insert): Likewise.
10428 2017-09-20 Jakub Jelinek <jakub@redhat.com>
10430 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
10431 that fits into uhwi or shwi, add DW_AT_const_value regardless
10432 of early_dwarf without going through RTL, using add_AT_unsigned
10435 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
10436 (DEBUG_LTO_ABBREV_SECTION): Likewise.
10437 (DEBUG_LTO_MACINFO_SECTION): Likewise.
10438 (DEBUG_MACRO_SECTION): Likewise.
10439 (DEBUG_LTO_MACRO_SECTION): Likewise.
10440 (DEBUG_STR_DWO_SECTION): Likewise.
10441 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
10442 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
10443 (DEBUG_LTO_DWO_LINE_SECTION): Define.
10444 (DEBUG_LTO_LINE_STR_SECTION): Define.
10445 (init_sections_and_labels): Initialize debug_line_str_section
10446 variable. Initialize debug_loc_section for -gdwarf-5 to
10447 DEBUG_LOCLISTS_SECTION. Formatting fixes.
10449 2017-09-20 Richard Biener <rguenther@suse.de>
10451 * graphite-sese-to-poly.c (extract_affine): Properly handle
10452 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
10454 2017-09-20 Richard Biener <rguenther@suse.de>
10456 PR tree-optimization/81373
10457 * graphite-scop-detection.c (build_cross_bb_scalars_def):
10458 Force SESE live-out defs to be handled even if they are
10461 2017-09-19 Jeff Law <law@redhat.com>
10463 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
10464 nothing for stack adjustments with REG_STACK_CHECK.
10465 * sched-deps.c (parse_add_or_inc): Reject insns with
10466 REG_STACK_CHECK from dependency breaking.
10467 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
10468 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
10469 * reg-notes.def (STACK_CHECK): New note.
10471 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
10472 (ix86_expand_prologue): Dump stack clash info as needed.
10473 Call ix86_adjust_stack_and_probe_stack_clash as needed.
10475 * function.c (dump_stack_clash_frame_info): New function.
10476 * function.h (dump_stack_clash_frame_info): Prototype.
10477 (enum stack_clash_probes): New enum.
10479 * config/alpha/alpha.c (alpha_expand_prologue): Also check
10480 flag_stack_clash_protection.
10481 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
10482 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
10483 (arm_frame_pointer_required): Likewise.
10484 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
10485 (ia64_expand_prologue): Likewise.
10486 * config/mips/mips.c (mips_expand_prologue): Likewise.
10487 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
10488 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10489 (sparc_flat_expand_prologue): Likewise.
10490 * config/spu/spu.c (spu_expand_prologue): Likewise.
10492 * explow.c: Include "params.h".
10493 (anti_adjust_stack_and_probe_stack_clash): New function.
10494 (get_stack_check_protect): Likewise.
10495 (compute_stack_clash_protection_loop_data): Likewise.
10496 (emit_stack_clash_protection_loop_start): Likewise.
10497 (emit_stack_clash_protection_loop_end): Likewise.
10498 (allocate_dynamic_stack_space): Use get_stack_check_protect.
10499 Use anti_adjust_stack_and_probe_stack_clash.
10500 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
10501 (emit_stack_clash_protection_loop_start): Likewise.
10502 (emit_stack_clash_protection_loop_end): Likewise.
10503 * rtl.h (get_stack_check_protect): Prototype.
10504 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
10505 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
10506 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
10508 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
10510 * doc/tm.texi: Rebuilt.
10511 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
10512 get_stack_check_protect.
10513 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
10514 * config/arm/arm.c (arm_expand_prologue): Likewise.
10515 (arm_frame_pointer_required): Likewise.
10516 * config/i386/i386.c (ix86_expand_prologue): Likewise.
10517 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
10518 * config/mips/mips.c (mips_expand_prologue): Likewise.
10519 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
10520 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
10521 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10522 (sparc_flat_expand_prologue): Likewise.
10524 * common.opt (-fstack-clash-protection): New option.
10525 * flag-types.h (enum stack_check_type): Note difference between
10526 -fstack-check= and -fstack-clash-protection.
10527 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
10528 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
10529 * toplev.c (process_options): Issue warnings/errors for cases
10530 not handled with -fstack-clash-protection.
10531 * doc/invoke.texi (-fstack-clash-protection): Document new option.
10532 (-fstack-check): Note additional problem with -fstack-check=generic.
10533 Note that -fstack-check is primarily for Ada and refer users
10534 to -fstack-clash-protection for stack-clash-protection.
10535 Document new params for stack clash protection.
10537 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
10539 * config/i386/i386.c (ix86_split_long_move): Do not handle
10540 address used for LEA in a special way.
10542 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
10544 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
10546 2017-09-19 Martin Sebor <msebor@redhat.com>
10549 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
10550 of incompatible types.
10552 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
10554 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10555 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
10556 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10557 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
10559 2017-09-19 Richard Biener <rguenther@suse.de>
10561 PR tree-optimization/82244
10562 * tree-vrp.c (remove_range_assertions): Do not propagate
10563 a constant to abnormals but replace the assert with a copy.
10565 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
10567 PR rtl-optimization/57878
10568 PR rtl-optimization/68988
10569 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
10570 avoidance test involving non_reload_pseudos. Move frequency test
10571 below the general fragmentation avoidance test.
10573 2017-09-19 Richard Biener <rguenther@suse.de>
10575 PR tree-optimization/69728
10576 * graphite-sese-to-poly.c (schedule_error): New global.
10577 (add_loop_schedule): Handle empty domain by failing the
10579 (build_original_schedule): Handle schedule_error.
10581 2017-09-19 Richard Biener <rguenther@suse.de>
10583 * graphite-scop-detection.c (scop_detection::can_represent_loop):
10584 Do not iterate to sibling loops but only to siblings of inner
10587 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
10590 * config/m68k/m68k.md (moveq feeding equality comparison): Check
10591 that the registers are different.
10593 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
10595 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
10596 to processor_model and "amdfam17h" to arch_names_table.
10597 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
10599 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10602 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
10604 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10605 Alan Hayward <alan.hayward@arm.com>
10606 David Sherwood <david.sherwood@arm.com>
10608 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
10610 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
10612 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
10613 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
10614 vect_schedule_slp_instance.
10615 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
10616 Update call to vect_slp_analyze_node_operations. Simplify return
10618 (vect_slp_analyze_bb_1): Update call accordingly.
10619 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
10620 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
10621 (vect_schedule_slp): Update call accordingly.
10623 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10624 Alan Hayward <alan.hayward@arm.com>
10625 David Sherwood <david.sherwood@arm.com>
10627 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
10628 with types that aren't in fact scalar.
10630 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10632 * tree-vect-slp.c (vect_record_max_nunits): New function,
10634 (vect_build_slp_tree_1): ...here.
10635 (vect_build_slp_tree_2): Call it for phis too.
10637 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10639 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
10640 to vect_get_vec_def_for_operand when getting the mask operand.
10642 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10643 Alan Hayward <alan.hayward@arm.com>
10644 David Sherwood <david.sherwood@arm.com>
10646 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
10649 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10650 Alan Hayward <alan.hayward@arm.com>
10651 David Sherwood <david.sherwood@arm.com>
10653 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
10654 calculation for vector booleans.
10656 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10657 Alan Hayward <alan.hayward@arm.com>
10658 David Sherwood <david.sherwood@arm.com>
10660 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
10662 (vect_transform_stmt): ...here.
10663 (vect_analyze_stmt): Use it instead of calling
10664 vectorizable_live_operation directly.
10666 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
10668 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
10669 non-SIMT targets in acc vector loops.
10671 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
10673 * configure.ac: Add arc and check if assembler supports gdwarf2.
10674 * configure: Regenerate.
10676 2017-09-18 Richard Biener <rguenther@suse.de>
10678 PR tree-optimization/82220
10679 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
10680 epilogue niters from the min_profitable_iters compute.
10682 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10685 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
10686 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
10687 (ix86_init_pic_reg): Revert 2017-09-01 changes.
10689 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
10692 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
10693 switching to a new text section.
10695 2017-09-18 Richard Biener <rguenther@suse.de>
10697 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
10699 (build_alias_set): Reject aliases with no access function.
10701 2017-09-18 Richard Biener <rguenther@suse.de>
10703 PR tree-optimization/79622
10704 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
10706 (build_cross_bb_scalars_use): Likewise.
10708 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
10710 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
10712 2017-09-18 Alan Modra <amodra@gmail.com>
10715 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
10716 stack_pointer_rtx for count 0. Update comments. Break up
10717 large rtl expression.
10719 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
10721 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
10722 Increase to 20 bytes.
10723 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
10724 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
10725 or avx version of the stub.
10727 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
10730 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
10731 compute the minimum stack alignment. Also update preferred stack
10732 boundary for leaf functions.
10734 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
10736 PR tree-optimization/82228
10737 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
10740 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10742 * common/config/nds32/nds32-common.c
10743 (nds32_option_optimization_table): Refine formatting.
10744 (nds32_option_optimization_table): Use -fsched-pressure and
10745 -fomit-frame-pointer for specific optimization level.
10747 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10749 * config/nds32/nds32.c: Refine formatting and comments.
10750 * config/nds32/nds32.h: Likewise.
10751 * config/nds32/nds32.md: Likewise.
10752 * config/nds32/nds32-cost.c: Likewise.
10753 * config/nds32/nds32-isr.c: Likewise.
10754 * config/nds32/nds32-md-auxiliary.c: Likewise.
10755 * config/nds32/nds32-multiple.md: Likewise.
10756 * config/nds32/nds32-predicates.c: Likewise.
10758 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
10759 Jakub Jelinek <jakub@redhat.com>
10761 Add support for -std=c++2a.
10762 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
10764 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
10766 2017-09-15 Steve Ellcey <sellcey@cavium.com>
10769 * doc/extend.texi (Common Function Attributes): Add
10770 references to ARM, AArch64, and S/390 specific attributes.
10771 (Function Specific Option Pragmas): Add AArch64 and S/390
10772 to list of back ends that support the target pragma.
10774 2017-09-15 Nathan Sidwell <nathan@acm.org>
10776 * doc/standards.texi: Fix C++17 description. Update URLs for
10779 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
10781 * common.opt (Wcast-align=strict): New warning option.
10782 * doc/invoke.texi: Document -Wcast-align=strict.
10784 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
10786 * cgraph.h (cgraph_thunk_info): Add comments.
10787 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
10788 assert for VIRTUAL_* arguments stricter.
10790 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
10792 PR tree-optimization/71026
10793 * match.pd: Move RDIV patterns from fold-const.c
10794 * fold-const.c (distribute_real_division): Removed.
10795 (fold_binary_loc): Remove calls to distribute_real_divison.
10797 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10799 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
10800 c++1z and gnu++1z as deprecated. Change other references to
10801 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
10802 Change -Wc++1z-compat to -Wc++17-compat.
10803 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
10804 * dwarf2out.c (highest_c_language): Handle C++17.
10805 (gen_compile_unit_die): Likewise.
10807 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10809 PR rtl-optimization/82192
10810 * combine.c (make_extraction): Don't look through non-paradoxical
10811 SUBREGs or TRUNCATE if pos + len is or might be bigger than
10814 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10815 Alan Hayward <alan.hayward@arm.com>
10816 David Sherwood <david.sherwood@arm.com>
10818 * target.def (function_arg_offset): New hook.
10819 * targhooks.h (default_function_arg_offset): Declare.
10820 * targhooks.c (default_function_arg_offset): New function.
10821 * function.c (locate_and_pad_parm): Use
10822 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
10823 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
10824 (TARGET_FUNCTION_ARG_OFFSET): ...this.
10825 * doc/tm.texi: Regenerate.
10826 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
10827 * config/spu/spu.c (spu_function_arg_offset): New function.
10828 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
10829 * system.h (FUNCTION_ARG_OFFSET): Poison.
10831 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10832 Alan Hayard <alan.hayward@arm.com>
10833 David Sherwood <david.sherwood@arm.com>
10835 * target.def (truly_noop_truncation): New hook.
10836 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
10837 than TRULY_NOOP_TRUNCATION.
10838 * hooks.h (hook_bool_uint_uint_true): Declare.
10839 * hooks.c (hook_bool_uint_uint_true): New function.
10840 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
10841 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
10842 * doc/tm.texi: Regenerate.
10843 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
10844 rather than TRULY_NOOP_TRUNCATION in comments.
10845 (simplify_comparison): Likewise.
10846 (record_truncated_value): Likewise.
10847 * expmed.c (extract_bit_field_1): Likewise.
10848 (extract_split_bit_field): Likewise.
10849 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
10850 instead of TRULY_NOOP_TRUNCATION.
10851 * function.c (assign_parm_setup_block): Likewise.
10852 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
10853 * rtlhooks.c: Include target.h.
10854 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
10855 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
10856 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
10857 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
10858 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
10859 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
10860 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
10861 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
10862 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
10863 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
10864 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
10865 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
10866 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
10867 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
10868 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
10869 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
10870 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
10871 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
10872 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
10873 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
10874 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
10875 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
10876 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
10877 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
10878 * config/mips/mips.c (mips_truly_noop_truncation): New function.
10879 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10880 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10881 rather than TRULY_NOOP_TRUNCATION in comments.
10882 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
10883 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
10884 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
10885 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
10886 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
10887 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
10888 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
10889 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
10890 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
10891 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
10892 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
10893 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10894 rather than TRULY_NOOP_TRUNCATION in comments.
10895 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
10896 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
10897 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
10898 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
10899 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
10900 TRULY_NOOP_TRUNCATION condition.
10901 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
10902 (TRULY_NOOP_TRUNCATION): Delete.
10903 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
10904 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
10905 * config/spu/spu.c (spu_truly_noop_truncation): New function.
10906 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10907 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
10908 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
10909 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
10910 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10911 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10912 rather than TRULY_NOOP_TRUNCATION in comments.
10913 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
10914 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
10915 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
10916 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
10917 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
10918 * system.h (TRULY_NOOP_TRUNCATION): Poison.
10920 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
10923 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
10924 (*cmp_ior): Likewise.
10925 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
10926 (*ior_scc_scc_cmp): Likewise.
10927 (*and_scc_scc): Likewise.
10928 (*and_scc_scc_cmp): Likewise.
10930 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10931 Alan Hayard <alan.hayward@arm.com>
10932 David Sherwood <david.sherwood@arm.com>
10934 * target.def (can_change_mode_class): New hook.
10935 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
10936 (hard_regno_nregs): Likewise.
10937 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
10938 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
10939 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
10940 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
10941 (LOAD_EXTEND_OP): Update accordingly.
10942 * doc/tm.texi: Regenerate.
10943 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
10944 CANNOT_CHANGE_MODE_CLASS.
10945 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
10946 (REG_CAN_CHANGE_MODE_P): ...this new macro.
10947 * combine.c (simplify_set): Update accordingly.
10948 * emit-rtl.c (validate_subreg): Likewise.
10949 * recog.c (general_operand): Likewise.
10950 * regcprop.c (mode_change_ok): Likewise.
10951 * reload1.c (choose_reload_regs): Likewise.
10952 (inherit_piecemeal_p): Likewise.
10953 * rtlanal.c (simplify_subreg_regno): Likewise.
10954 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
10955 instead of CANNOT_CHANGE_MODE_CLASS.
10956 (reload_cse_simplify_operands): Likewise.
10957 * reload.c (push_reload): Use targetm.can_change_mode_class
10958 instead of CANNOT_CHANGE_MODE_CLASS.
10959 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
10960 REG_CANNOT_CHANGE_MODE_P.
10961 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10962 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
10963 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10964 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10965 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10966 (arm_can_change_mode_class): New function.
10967 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10968 than CANNOT_CHANGE_MODE_CLASS in comments.
10969 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10970 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
10971 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
10972 (ix86_can_change_mode_class): ...this new function, inverting the
10973 sense of the return value.
10974 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10975 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10976 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10977 (ia64_can_change_mode_class): New function.
10978 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10979 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
10980 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
10981 (m32c_can_change_mode_class): ...this new function, inverting the
10982 sense of the return value.
10983 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10984 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10985 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
10986 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
10987 (mips_can_change_mode_class): ...this new function, inverting the
10988 sense of the return value.
10989 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10990 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10991 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10992 (msp430_can_change_mode_class): New function.
10993 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10994 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
10995 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10996 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10997 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10998 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
10999 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11000 (pa_cannot_change_mode_class): Replace with...
11001 (pa_can_change_mode_class): ...this new function, inverting the
11002 sense of the return value.
11003 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11004 than CANNOT_CHANGE_MODE_CLASS in comments.
11005 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11006 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11007 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11008 (pdp11_cannot_change_mode_class): Replace with...
11009 (pdp11_can_change_mode_class): ...this new function, inverting the
11010 sense of the return value.
11011 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11012 * config/powerpcspe/powerpcspe-protos.h
11013 (rs6000_cannot_change_mode_class_ptr): Delete.
11014 * config/powerpcspe/powerpcspe.c
11015 (rs6000_cannot_change_mode_class_ptr): Delete.
11016 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11017 (rs6000_option_override_internal): Assign to
11018 targetm.can_change_mode_class instead of
11019 rs6000_cannot_change_mode_class_ptr.
11020 (rs6000_cannot_change_mode_class): Replace with...
11021 (rs6000_can_change_mode_class): ...this new function, inverting the
11022 sense of the return value.
11023 (rs6000_debug_cannot_change_mode_class): Replace with...
11024 (rs6000_debug_can_change_mode_class): ...this new function.
11025 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11026 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
11027 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11028 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11029 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
11031 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
11032 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11033 (rs6000_option_override_internal): Assign to
11034 targetm.can_change_mode_class instead of
11035 rs6000_cannot_change_mode_class_ptr.
11036 (rs6000_cannot_change_mode_class): Replace with...
11037 (rs6000_can_change_mode_class): ...this new function, inverting the
11038 sense of the return value.
11039 (rs6000_debug_cannot_change_mode_class): Replace with...
11040 (rs6000_debug_can_change_mode_class): ...this new function.
11041 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11042 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
11043 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
11044 (s390_can_change_mode_class): ...this new function, inverting the
11045 sense of the return value.
11046 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11047 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11048 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
11049 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11050 (sh_cannot_change_mode_class): Replace with...
11051 (sh_can_change_mode_class): ...this new function, inverting the
11052 sense of the return value.
11053 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11054 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11055 (sparc_can_change_mode_class): New function.
11056 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11057 * config/spu/spu.c (spu_can_change_mode_class): New function.
11058 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11059 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11060 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11061 (visium_can_change_mode_class): New function.
11062 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
11064 2017-09-15 Richard Biener <rguenther@suse.de>
11066 PR tree-optimization/82217
11067 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
11068 but not undefined case.
11070 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11073 * postreload.c (reload_cse_simplify_operands): Skip
11074 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
11076 2017-09-15 Richard Biener <rguenther@suse.de>
11078 PR tree-optimization/68823
11079 * graphite-scop-detection.c (build_alias_set): If we have a
11080 possible dependence check whether we can handle them by just
11081 looking at the DRs DR_ACCESS_FNs.
11082 (build_scops): If build_alias_set fails, fail the SCOP.
11084 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11086 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
11087 to support float128 built-in functions that require the ISA 3.0
11089 (BU_FLOAT128_3_HW): Likewise.
11090 (SQRTF128): Add support for the IEEE 128-bit square root and fma
11091 built-in functions.
11092 (FMAF128): Likewise.
11094 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
11095 support for built-in functions that need the ISA 3.0 IEEE 128-bit
11096 floating point instructions.
11097 (rs6000_invalid_builtin): Likewise.
11098 (rs6000_builtin_mask_names): Likewise.
11099 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
11100 (RS6000_BTM_FLOAT128_HW): Likewise.
11101 (RS6000_BTM_COMMON): Likewise.
11102 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
11104 * doc/extend.texi (RS/6000 built-in functions): Document the
11105 IEEE 128-bit floating point square root and fused multiply-add
11106 built-in functions.
11108 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
11110 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
11111 reg (r2) isn't in the set of registers defined in the prologue.
11113 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11114 Alan Hayward <alan.hayward@arm.com>
11115 David Sherwood <david.sherwood@arm.com>
11117 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
11118 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
11119 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
11120 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
11121 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
11123 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
11124 max_vectorization_factor.
11125 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
11127 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11128 Alan Hayward <alan.hayward@arm.com>
11129 David Sherwood <david.sherwood@arm.com>
11131 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
11132 (vect_worthwhile_without_simd_p): Declare.
11133 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
11134 (vectorizable_reduction): Use it.
11135 * tree-vect-stmts.c (vectorizable_shift): Likewise.
11136 (vectorizable_operation): Likewise.
11138 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11139 Alan Hayward <alan.hayward@arm.com>
11140 David Sherwood <david.sherwood@arm.com>
11142 * tree-vectorizer.h (vect_get_num_copies): New function.
11143 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
11144 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11145 (vectorizable_induction): Likewise.
11146 (vectorizable_live_operation): Likewise.
11147 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11148 (vectorizable_bswap): Likewise.
11149 (vectorizable_call): Likewise.
11150 (vectorizable_conversion): Likewise.
11151 (vectorizable_assignment): Likewise.
11152 (vectorizable_shift): Likewise.
11153 (vectorizable_operation): Likewise.
11154 (vectorizable_store): Likewise.
11155 (vectorizable_load): Likewise.
11156 (vectorizable_condition): Likewise.
11157 (vectorizable_comparison): Likewise.
11158 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
11160 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11161 Alan Hayward <alan.hayward@arm.com>
11162 David Sherwood <david.sherwood@arm.com>
11164 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
11165 of vect_init_vector.
11167 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11168 Alan Hayward <alan.hayward@arm.com>
11169 David Sherwood <david.sherwood@arm.com>
11171 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
11172 an inline wrapper that provides a location.
11173 (gimple_build_vector): Likewise.
11174 * gimple-fold.c (gimple_build_vector_from_val): New function.
11175 (gimple_build_vector): Likewise.
11176 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
11177 functions to build the initial value. Always return a gimple value.
11178 (get_initial_defs_for_reduction): Likewise. Only compute
11180 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
11181 vect_init_vector on the results from get_initial_def(s)_for_reduction.
11182 (vectorizable_induction): Use gimple_build_vector rather than
11185 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11186 Alan Hayward <alan.hayward@arm.com>
11187 David Sherwood <david.sherwood@arm.com>
11189 * target.h (vec_perm_indices): New typedef.
11190 (auto_vec_perm_indices): Likewise.
11191 * optabs-query.h: Include target.h
11192 (can_vec_perm_p): Take a vec_perm_indices *.
11193 * optabs-query.c (can_vec_perm_p): Likewise.
11194 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
11195 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11196 * tree-vect-generic.c (lower_vec_perm): Likewise.
11197 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
11198 (vect_grouped_load_supported): Likewise.
11199 (vect_shift_permute_load_chain): Likewise.
11200 (vect_permute_store_chain): Use auto_vec_perm_indices.
11201 (vect_permute_load_chain): Likewise.
11202 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
11203 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
11204 Update uses of can_vec_perm_p.
11205 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
11206 mode with a number of elements. Take a vec_perm_indices *.
11207 (vect_create_epilog_for_reduction): Update accordingly.
11208 Use auto_vec_perm_indices.
11209 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
11210 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
11211 (vect_transform_slp_perm_load): Likewise.
11212 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
11213 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
11214 (vect_gen_perm_mask_checked): Likewise.
11215 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
11216 (vect_gen_perm_mask_checked): Likewise.
11217 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
11218 (vectorizable_store): Likewise.
11219 (vectorizable_load): Likewise.
11220 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
11221 (vectorizable_bswap): Likewise.
11223 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11224 Alan Hayward <alan.hayward@arm.com>
11225 David Sherwood <david.sherwood@arm.com>
11227 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
11228 * tree.c (build_vector): Likewise.
11229 (build_vector_from_ctor): Update accordingly.
11230 (build_vector_from_val): Likewise.
11231 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
11232 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11233 * tree-vect-generic.c (add_rshift): Likewise.
11234 (expand_vector_divmod): Likewise.
11235 (optimize_vector_constructor): Likewise.
11236 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
11237 (vect_transform_slp_perm_load): Likewise.
11238 (vect_schedule_slp_instance): Likewise.
11239 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
11240 (vectorizable_call): Likewise.
11241 (vect_gen_perm_mask_any): Likewise. Add elements in order.
11242 * expmed.c (make_tree): Likewise.
11243 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
11244 a vector passed to build_vector.
11245 (fold_convert_const): Likewise.
11246 (exact_inverse): Likewise.
11247 (fold_ternary_loc): Likewise.
11248 (fold_relational_const): Likewise.
11249 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
11250 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
11251 (const_unop): Likewise. Store the reduction accumulator in a
11252 variable rather than an array.
11253 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
11254 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
11255 the new vector, rather than constructing it after the input arrays.
11256 (native_interpret_vector): Use auto_vec<tree> when building
11257 a vector passed to build_vector. Add elements in order.
11258 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
11259 auto_vec<tree> when building a vector passed to build_vector.
11260 (vect_create_epilog_for_reduction): Likewise.
11261 (vectorizable_induction): Likewise.
11262 (get_initial_def_for_reduction): Likewise. Fix indentation of
11264 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
11266 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
11267 passed to build_vector.
11269 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11270 Alan Hayward <alan.hayward@arm.com>
11271 David Sherwood <david.sherwood@arm.com>
11273 * tree-core.h (tree_base::u): Add an "nelts" field.
11274 (tree_vector): Use VECTOR_CST_NELTS as the length.
11275 * tree.c (tree_size): Likewise.
11276 (make_vector): Initialize VECTOR_CST_NELTS.
11277 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
11278 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
11279 TYPE_VECTOR_SUBPARTS.
11280 * expr.c (const_vector_mask_from_tree): Consistently use "units"
11281 as the number of units, setting it from VECTOR_CST_NELTS.
11282 (const_vector_from_tree): Likewise.
11283 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
11284 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
11285 (fold_negate_expr_1): Likewise.
11286 (fold_convert_const): Likewise.
11287 (const_binop): Likewise. Differentiate the number of output and
11289 (const_unop): Likewise.
11290 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
11291 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
11292 in cases that did the opposite.
11294 2017-09-14 Richard Biener <rguenther@suse.de>
11296 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
11299 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
11301 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
11303 2017-09-14 Jakub Jelinek <jakub@redhat.com>
11306 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
11307 if and where to split a bb, except for splitting before debug insn
11308 sequences followed by non-label real insn. Delete debug insns
11309 in between basic blocks.
11311 * combine.c (make_compound_operation_int): Formatting fixes.
11313 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
11314 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11315 * config/netbsd.h (LINK_EH_SPEC): Likewise.
11316 * config/sol2.h (LINK_EH_SPEC): Likewise.
11317 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11318 * config/s390/linux.h (LINK_SPEC): Likewise.
11319 * config/freebsd.h (LINK_EH_SPEC): Likewise.
11320 * config/openbsd.h (LINK_EH_SPEC): Likewise.
11321 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11322 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
11323 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
11324 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11325 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
11326 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
11328 2017-09-13 Jakub Jelinek <jakub@redhat.com>
11330 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
11332 (ENDFILE_LINUX_SPEC): Likewise.
11333 (LINK_EH_SPEC): Likewise.
11334 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
11335 (LINK_OS_LINUX_SPEC32): Likewise.
11336 (LINK_OS_LINUX_SPEC64): Likewise.
11337 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
11338 (LINK_OS_LINUX_SPEC): Likewise.
11340 2017-09-13 Martin Liska <mliska@suse.cz>
11342 PR middle-end/82154
11343 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
11344 CASE_HIGH is NULL_TREE.
11346 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11347 Alan Hayward <alan.hayward@arm.com>
11348 David Sherwood <david.sherwood@arm.com>
11350 * target.def (secondary_memory_needed): New hook.
11351 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
11352 instead of SECONDARY_MEMORY_NEEDED.
11353 (secondary_memory_needed_mode): Likewise.
11354 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
11355 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
11356 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
11357 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
11358 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
11359 * doc/tm.texi: Regenerate.
11360 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
11361 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
11362 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11363 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
11364 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
11365 * config/i386/i386.c (inline_secondary_memory_needed): Put the
11366 mode argument first and change the reg_class arguments to reg_class_t.
11367 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
11368 Make static. Update the call to inline_secondary_memory_needed.
11369 (ix86_register_move_cost): Update the call to
11370 inline_secondary_memory_needed.
11371 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11372 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
11374 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
11375 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11377 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
11378 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
11379 * config/mips/mips.c (mips_secondary_memory_needed): Make static
11380 and match hook interface. Add comment from mips.h.
11381 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11382 * config/mmix/mmix.md (truncdfsf2): Refer to
11383 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11385 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
11386 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
11387 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11388 (pa_secondary_memory_needed): New function.
11389 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
11390 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
11391 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11392 (pdp11_secondary_memory_needed): Make static and match hook interface.
11393 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
11394 * config/powerpcspe/powerpcspe-protos.h
11395 (rs6000_secondary_memory_needed_ptr): Delete.
11396 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
11398 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11399 (rs6000_option_override_internal): Assign to
11400 targetm.secondary_memory_needed rather than
11401 rs6000_secondary_memory_needed_ptr.
11402 (rs6000_secondary_memory_needed): Match hook interface.
11403 (rs6000_debug_secondary_memory_needed): Likewise.
11404 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
11405 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
11406 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
11407 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11408 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
11409 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
11411 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
11412 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11413 (rs6000_option_override_internal): Assign to
11414 targetm.secondary_memory_needed rather than
11415 rs6000_secondary_memory_needed_ptr.
11416 (rs6000_secondary_memory_needed): Match hook interface.
11417 (rs6000_debug_secondary_memory_needed): Likewise.
11418 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
11419 * config/s390/s390.c (s390_secondary_memory_needed): New function.
11420 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11421 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
11422 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11423 (sparc_secondary_memory_needed): New function.
11424 * lra-constraints.c (check_and_process_move): Refer to
11425 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11427 (curr_insn_transform): Likewise.
11428 (process_alt_operands): Use targetm.secondary_memory_needed
11429 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11430 (check_secondary_memory_needed_p): Likewise.
11431 (choose_split_class): Likewise.
11432 * reload.c: Unconditionally include code that was previously
11433 conditional on SECONDARY_MEMORY_NEEDED.
11434 (push_secondary_reload): Use targetm.secondary_memory_needed
11435 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11436 (push_reload): Likewise.
11437 * reload1.c: Unconditionally include code that was previously
11438 conditional on SECONDARY_MEMORY_NEEDED.
11439 (choose_reload_regs): Use targetm.secondary_memory_needed
11440 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11441 (gen_reload): Likewise.
11442 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
11444 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11445 Alan Hayward <alan.hayward@arm.com>
11446 David Sherwood <david.sherwood@arm.com>
11448 * target.def (secondary_memory_needed_mode): New hook:
11449 * targhooks.c (default_secondary_memory_needed_mode): Declare.
11450 * targhooks.h (default_secondary_memory_needed_mode): New function.
11451 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
11452 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
11453 * doc/tm.texi: Regenerate.
11454 * lra-constraints.c (check_and_process_move): Use
11455 targetm.secondary_memory_needed_mode instead of
11456 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
11457 (curr_insn_transform): Likewise.
11458 * reload.c (get_secondary_mem): Likewise.
11459 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11460 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
11462 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11463 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11464 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
11465 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11466 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
11468 * config/powerpcspe/powerpcspe-protos.h
11469 (rs6000_secondary_memory_needed_mode): Delete.
11470 * config/powerpcspe/powerpcspe.c
11471 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11472 (rs6000_secondary_memory_needed_mode): Make static.
11473 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11474 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
11476 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11478 (rs6000_secondary_memory_needed_mode): Make static.
11479 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11480 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
11481 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11482 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11483 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11485 (sparc_secondary_memory_needed_mode): New function.
11486 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
11488 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
11490 * config/aarch64/constraints.md (Umq): New constraint.
11491 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
11493 (mov<mode>): Update condition.
11495 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11497 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
11498 when bitposition is the same.
11500 2017-09-13 Richard Biener <rguenther@suse.de>
11502 * dwarf2out.c (output_die_symbol): Remove.
11503 (output_die): Do not output a DIEs symbol.
11505 2017-09-13 Richard Biener <rguenther@suse.de>
11507 PR middle-end/82128
11508 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
11509 default-def to avoid breaking iterator update with the weird
11510 interaction with cgraph_update_edges_for_call_stmt_node.
11512 2017-09-13 Richard Biener <rguenther@suse.de>
11514 * tree-cfg.c (verify_gimple_assign_binary): Add verification
11515 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
11516 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
11517 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
11519 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
11521 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
11522 Disable pc relative literal load irrespective of
11523 TARGET_FIX_ERR_A53_84341 for default.
11525 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
11527 * config/sparc/sparc.c (output_return): Output the source location of
11528 the insn in the delay slot, if any.
11529 (output_sibcall): Likewise.
11531 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
11534 * common.opt (-static-pie): New alias.
11535 (shared): Negate static-pie.
11536 (-no-pie): Update help text.
11538 (static-pie): New option.
11539 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
11540 -static-pie support.
11541 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
11542 (LINK_EH_SPEC): Likewise.
11543 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11544 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11545 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11546 * gcc.c (LINK_COMMAND_SPEC): Likewise.
11547 (init_gcc_specs): Likewise.
11548 (init_spec): Likewise.
11549 (display_help): Update help message for -pie.
11550 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
11553 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
11555 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
11556 (movdi_aarch64): Likewise.
11557 (movti_aarch64): Likewise.
11559 2017-09-12 Simon Wright <simon@pushface.org>
11562 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
11563 calculation of the minor version, always output as 0.
11565 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11568 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11569 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
11570 on it early, rather than manual conversion late. For
11571 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
11572 instead of performing manual conversion.
11574 2017-09-12 Carl Love <cel@us.ibm.com>
11576 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
11577 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
11579 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11580 VMULOSW): Add definitions.
11581 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11582 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11583 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11584 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11585 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11587 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11589 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
11591 (movti_aarch64): Likewise.
11592 (movdf_aarch64): Likewise.
11593 (movtf_aarch64): Likewise.
11594 (load_pairdi): Likewise.
11595 (store_pairdi): Likewise.
11596 (load_pairdf): Likewise.
11597 (store_pairdf): Likewise.
11598 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
11599 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
11600 (ldr_got_small_<mode>): Likewise.
11601 (ldr_got_small_28k_<mode>): Likewise.
11602 (ldr_got_tiny): Likewise.
11603 * config/aarch64/iterators.md (ldst_sz): New.
11604 (ldpstp_sz): Likewise.
11605 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
11607 (thunderx_load): Split load_8 to load_16.
11608 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
11610 (thunderx2t99_storepair_basic): Split store_8 to store_16.
11611 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
11612 (xgene1_store_pair): Split store_8 to store_16.
11613 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
11614 (falkor_st_0_st_sd): Split store_8 to store_16.
11616 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11618 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
11619 and store1/2/3/4 to store_4/8/12/16.
11620 * config/aarch64/aarch64.md: Update for rename.
11621 * config/arm/arm.md: Likewise.: Likewise.
11622 * config/arm/arm.c: Likewise.
11623 * config/arm/thumb1.md: Likewise.
11624 * config/arm/thumb2.md: Likewise.
11625 * config/arm/vfp.md: Likewise.
11626 * config/arm/arm-generic.md: Likewise.
11627 * config/arm/arm1020e.md: Likewise.
11628 * config/arm/arm1026ejs.md: Likewise.
11629 * config/arm/arm1136jfs.md: Likewise.
11630 * config/arm/arm926ejs.md: Likewise.
11631 * config/arm/cortex-a15.md: Likewise.
11632 * config/arm/cortex-a17.md: Likewise.
11633 * config/arm/cortex-a5.md: Likewise.
11634 * config/arm/cortex-a53.md: Likewise.
11635 * config/arm/cortex-a57.md: Likewise.
11636 * config/arm/cortex-a7.md: Likewise.
11637 * config/arm/cortex-a8.md: Likewise.
11638 * config/arm/cortex-a9.md: Likewise.
11639 * config/arm/cortex-m4.md: Likewise.
11640 * config/arm/cortex-m7.md: Likewise.
11641 * config/arm/cortex-r4.md: Likewise.
11642 * config/arm/exynos-m1.md: Likewise.
11643 * config/arm/fa526.md: Likewise.
11644 * config/arm/fa606te.md: Likewise.
11645 * config/arm/fa626te.md: Likewise.
11646 * config/arm/fa726te.md: Likewise.
11647 * config/arm/fmp626.md: Likewise.
11648 * config/arm/iwmmxt.md: Likewise.
11649 * config/arm/ldmstm.md: Likewise.
11650 * config/arm/marvell-pj4.md: Likewise.
11651 * config/arm/xgene1.md: Likewise.
11652 * config/aarch64/thunderx.md: Likewise.
11653 * config/aarch64/thunderx2t99.md: Likewise.
11654 * config/aarch64/falkor.md: Likewise.
11656 2017-09-12 Martin Liska <mliska@suse.cz>
11658 * attribs.c (private_lookup_attribute): New function.
11659 * attribs.h (private_lookup_attribute): Declared here.
11660 (lookup_attribute): Called from this place.
11662 2017-09-12 Richard Biener <rguenther@suse.de>
11664 PR tree-optimization/82157
11665 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
11666 stmts with side-effects.
11668 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11669 Alan Hayward <alan.hayward@arm.com>
11670 David Sherwood <david.sherwood@arm.com>
11672 * target.def (hard_regno_nregs): New hook.
11673 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
11674 * targhooks.h (default_hard_regno_nregs): Declare.
11675 * targhooks.c (default_hard_regno_nregs): New function.
11676 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
11677 (TARGET_HARD_REGNO_NREGS): ...this hook.
11678 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
11679 (CLASS_MAX_NREGS): Likewise.
11680 * doc/tm.texi: Regenerate.
11681 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
11682 instead of HARD_REGNO_NREGS.
11683 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
11684 HARD_REGNO_NREGS in the comment.
11685 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
11686 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
11687 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
11688 Return an unsigned int.
11689 (TARGET_HARD_REGNO_NREGS): Redefine.
11690 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
11691 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
11692 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11693 (arc_hard_regno_nregs): New function.
11694 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
11695 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
11696 (arm_hard_regno_nregs): New function.
11697 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
11698 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
11699 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
11700 (TARGET_HARD_REGNO_NREGS): Redefine.
11701 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
11702 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
11703 (HARD_REGNO_NREGS): Delete.
11704 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
11705 (cr16_hard_regno_nregs): New function.
11706 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
11707 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
11708 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
11709 (cris_hard_regno_nregs): New function.
11710 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
11711 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
11712 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11713 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
11714 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
11715 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
11716 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
11717 (frv_hard_regno_nregs): Make static. Take and return an
11719 (frv_class_max_nregs): Remove outdated copy of documentation.
11720 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
11721 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
11722 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
11723 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
11724 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
11725 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
11726 (TARGET_HARD_REGNO_NREGS): Redefine.
11727 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
11728 (CLASS_MAX_NREGS): Update comment.
11729 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
11730 (ia64_hard_regno_nregs): New function.
11731 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
11732 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
11733 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
11734 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
11735 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
11737 (m32c_hard_regno_nregs): Likewise. Make static.
11738 (TARGET_HARD_REGNO_NREGS): Redefine.
11739 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
11740 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
11741 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
11742 (m68k_hard_regno_nregs): New function.
11743 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
11744 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
11745 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
11746 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
11747 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
11748 Take and return an unsigned int.
11749 (TARGET_HARD_REGNO_NREGS): Redefine.
11750 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
11751 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11752 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
11753 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
11754 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
11755 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
11756 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
11757 (msp430_hard_regno_nregs): Make static. Take and return an
11759 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
11760 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
11761 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
11762 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
11763 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
11764 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
11765 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
11766 (TARGET_HARD_REGNO_NREGS): Redefine.
11767 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
11768 (PA_HARD_REGNO_NREGS): ...this.
11769 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
11770 (PA_HARD_REGNO_NREGS): ...this.
11771 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11772 (pa_hard_regno_nregs): New function.
11773 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
11774 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
11775 (pdp11_hard_regno_nregs): New function.
11776 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
11777 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
11778 (rs6000_hard_regno_nregs_hook): New function.
11779 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
11780 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
11781 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
11782 Take and return an unsigned int. Move earlier in file.
11783 (TARGET_HARD_REGNO_NREGS): Redefine.
11784 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
11785 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
11786 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
11787 (rl78_hard_regno_nregs): Make static. Take and return an
11789 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
11790 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
11791 (rs6000_hard_regno_nregs_hook): New function.
11792 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
11793 * config/rx/rx.c (rx_hard_regno_nregs): New function.
11794 (TARGET_HARD_REGNO_NREGS): Redefine.
11795 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
11796 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
11797 instead of HARD_REGNO_NREGS.
11798 (s390_hard_regno_nregs): New function.
11799 (s390_hard_regno_mode_ok): Add comment from s390.h.
11800 (TARGET_HARD_REGNO_NREGS): Redefine.
11801 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
11802 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
11803 (sh_hard_regno_nregs): New function.
11804 (sh_pass_in_reg_p): Use it.
11805 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
11806 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11807 (sparc_hard_regno_nregs): New function.
11808 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
11809 * config/spu/spu.c (spu_hard_regno_nregs): New function.
11810 (spu_function_arg_advance): Use it, supplying a valid register number.
11811 (TARGET_HARD_REGNO_NREGS): Redefine.
11812 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
11813 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
11814 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
11815 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
11816 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
11817 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
11818 (CLASS_MAX_NREGS): Remove copy of old documentation.
11819 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
11820 (visium_hard_regno_nregs): New function.
11821 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
11822 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
11823 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11824 xtensa_hard_regno_nregs): New function.
11825 * system.h (HARD_REGNO_NREGS): Poison.
11827 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11829 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
11830 hard_regno_nregs instead of HARD_REGNO_NREGS.
11831 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
11832 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
11833 (c6x_expand_epilogue): Likewise.
11834 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
11835 (frv_read_iacc_argument): Likewise.
11836 * config/sh/sh.c: Include regs.h.
11837 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11838 (regs_used): Likewise.
11839 (output_stack_adjust): Likewise.
11840 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
11841 * expmed.c: Include regs.h.
11842 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11843 * ree.c: Include regs.h.
11844 (combine_reaching_defs): Use hard_regno_nregs instead of
11846 (add_removable_extension): Likewise.
11848 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11850 * regs.h (hard_regno_nregs): Turn into a function.
11851 (end_hard_regno): Update accordingly.
11852 * caller-save.c (setup_save_areas): Likewise.
11853 (save_call_clobbered_regs): Likewise.
11854 (replace_reg_with_saved_mem): Likewise.
11855 (insert_restore): Likewise.
11856 (insert_save): Likewise.
11857 * combine.c (can_change_dest_mode): Likewise.
11858 (move_deaths): Likewise.
11859 (distribute_notes): Likewise.
11860 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
11861 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
11862 (rs6000_split_multireg_move): Likewise.
11863 (rs6000_register_move_cost): Likewise.
11864 (rs6000_memory_move_cost): Likewise.
11865 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
11866 (rs6000_split_multireg_move): Likewise.
11867 (rs6000_register_move_cost): Likewise.
11868 (rs6000_memory_move_cost): Likewise.
11869 * cselib.c (cselib_reset_table): Likewise.
11870 (cselib_lookup_1): Likewise.
11871 * emit-rtl.c (set_mode_and_regno): Likewise.
11872 * function.c (aggregate_value_p): Likewise.
11873 * ira-color.c (setup_profitable_hard_regs): Likewise.
11874 (check_hard_reg_p): Likewise.
11875 (calculate_saved_nregs): Likewise.
11876 (assign_hard_reg): Likewise.
11877 (improve_allocation): Likewise.
11878 (calculate_spill_cost): Likewise.
11879 * ira-emit.c (modify_move_list): Likewise.
11880 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
11881 (ira_hard_reg_in_set_p): Likewise.
11882 * ira.c (setup_reg_mode_hard_regset): Likewise.
11883 (clarify_prohibited_class_mode_regs): Likewise.
11884 (check_allocation): Likewise.
11885 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11886 (lra_setup_reg_renumber): Likewise.
11887 (setup_try_hard_regno_pseudos): Likewise.
11888 (spill_for): Likewise.
11889 (assign_hard_regno): Likewise.
11890 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
11891 * lra-constraints.c (in_class_p): Likewise.
11892 (lra_constraint_offset): Likewise.
11893 (simplify_operand_subreg): Likewise.
11894 (lra_constraints): Likewise.
11895 (split_reg): Likewise.
11896 (split_if_necessary): Likewise.
11897 (invariant_p): Likewise.
11898 (inherit_in_ebb): Likewise.
11899 * lra-lives.c (process_bb_lives): Likewise.
11900 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
11901 (get_hard_regs): Likewise.
11902 (do_remat): Likewise.
11903 * lra-spills.c (assign_spill_hard_regs): Likewise.
11904 * mode-switching.c (create_pre_exit): Likewise.
11905 * postreload.c (reload_combine_recognize_pattern): Likewise.
11906 * recog.c (peep2_find_free_register): Likewise.
11907 * regcprop.c (kill_value_regno): Likewise.
11908 (set_value_regno): Likewise.
11909 (copy_value): Likewise.
11910 (maybe_mode_change): Likewise.
11911 (find_oldest_value_reg): Likewise.
11912 (copyprop_hardreg_forward_1): Likewise.
11913 * regrename.c (check_new_reg_p): Likewise.
11914 (regrename_do_replace): Likewise.
11915 * reload.c (push_reload): Likewise.
11916 (combine_reloads): Likewise.
11917 (find_dummy_reload): Likewise.
11918 (operands_match_p): Likewise.
11919 (find_reloads): Likewise.
11920 (find_equiv_reg): Likewise.
11921 (reload_adjust_reg_for_mode): Likewise.
11922 * reload1.c (count_pseudo): Likewise.
11923 (count_spilled_pseudo): Likewise.
11924 (find_reg): Likewise.
11925 (clear_reload_reg_in_use): Likewise.
11926 (free_for_value_p): Likewise.
11927 (allocate_reload_reg): Likewise.
11928 (choose_reload_regs): Likewise.
11929 (reload_adjust_reg_for_temp): Likewise.
11930 (emit_reload_insns): Likewise.
11931 (delete_output_reload): Likewise.
11932 * rtlanal.c (subreg_get_info): Likewise.
11933 * sched-deps.c (sched_analyze_reg): Likewise.
11934 * sel-sched.c (init_regs_for_mode): Likewise.
11935 (mark_unavailable_hard_regs): Likewise.
11936 (choose_best_reg_1): Likewise.
11937 (verify_target_availability): Likewise.
11938 * valtrack.c (dead_debug_insert_temp): Likewise.
11939 * var-tracking.c (track_loc_p): Likewise.
11940 (emit_note_insn_var_location): Likewise.
11941 * varasm.c (make_decl_rtl): Likewise.
11942 * reginfo.c (choose_hard_reg_mode): Likewise.
11943 (init_reg_modes_target): Refer directly to
11944 this_target_regs->x_hard_regno_nregs.
11946 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11948 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
11949 instead of hard_regno_nregs.
11951 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11953 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
11954 end_hard_regno instead of hard_regno_nregs.
11955 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
11956 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
11957 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
11958 * ira-color.c (improve_allocation): Likewise.
11959 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11960 * lra-lives.c (mark_regno_live): Likewise.
11961 (mark_regno_dead): Likewise.
11962 * lra-remat.c (operand_to_remat): Likewise.
11963 * lra.c (collect_non_operand_hard_regs): Likewise.
11964 * postreload.c (reload_combine_note_store): Likewise.
11965 (move2add_valid_value_p): Likewise.
11966 * reload.c (regno_clobbered_p): Likewise.
11968 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11970 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
11972 * config/v850/v850.c (v850_reorg): Likewise.
11973 * reload.c (refers_to_regno_for_reload_p): Likewise.
11974 (find_equiv_reg): Likewise.
11975 * reload1.c (reload_reg_reaches_end_p): Likewise.
11977 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11979 * caller-save.c (add_used_regs): Use REG_NREGS instead of
11981 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
11982 * config/arm/arm.c (output_move_neon): Likewise.
11983 (arm_attr_length_move_neon): Likewise.
11984 (neon_split_vcombine): Likewise.
11985 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
11986 (c6x_mark_reg_written): Likewise.
11987 (c6x_dwarf_register_span): Likewise.
11988 * config/i386/i386.c (ix86_save_reg): Likewise.
11989 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
11990 (rws_access_reg): Likewise.
11991 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
11992 * mode-switching.c (create_pre_exit): Likewise.
11993 * ree.c (combine_reaching_defs): Likewise.
11994 (add_removable_extension): Likewise.
11995 * regcprop.c (find_oldest_value_reg): Likewise.
11996 (copyprop_hardreg_forward_1): Likewise.
11997 * reload.c (reload_inner_reg_of_subreg): Likewise.
11998 (push_reload): Likewise.
11999 (combine_reloads): Likewise.
12000 (find_dummy_reload): Likewise.
12001 (reload_adjust_reg_for_mode): Likewise.
12002 * reload1.c (find_reload_regs): Likewise.
12003 (forget_old_reloads_1): Likewise.
12004 (reload_reg_free_for_value_p): Likewise.
12005 (reload_adjust_reg_for_temp): Likewise.
12006 (emit_reload_insns): Likewise.
12007 (delete_output_reload): Likewise.
12008 * sel-sched.c (choose_best_reg_1): Likewise.
12009 (choose_best_pseudo_reg): Likewise.
12011 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12012 Alan Hayward <alan.hayward@arm.com>
12013 David Sherwood <david.sherwood@arm.com>
12015 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12016 * target.def (slow_unaligned_access): New hook.
12017 * targhooks.h (default_slow_unaligned_access): Declare.
12018 * targhooks.c (default_slow_unaligned_access): New function.
12019 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
12020 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
12021 * doc/tm.texi: Regenerate.
12022 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
12023 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
12024 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
12026 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
12027 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
12029 (rs6000_slow_unaligned_access): New function.
12030 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12031 (expand_block_compare): Likewise.
12032 (expand_strn_compare): Likewise.
12033 (rs6000_rtx_costs): Likewise.
12034 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
12035 (riscv_slow_unaligned_access): Likewise.
12036 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
12037 (riscv_slow_unaligned_access_p): ...this and make static.
12038 (riscv_option_override): Update accordingly.
12039 (riscv_slow_unaligned_access): New function.
12040 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12041 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
12042 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12043 (rs6000_slow_unaligned_access): New function.
12044 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12045 (rs6000_rtx_costs): Likewise.
12046 * config/rs6000/rs6000-string.c (expand_block_compare)
12047 (expand_strn_compare): Use targetm.slow_unaligned_access instead
12048 of SLOW_UNALIGNED_ACCESS.
12049 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
12050 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
12051 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
12052 of SLOW_UNALIGNED_ACCESS.
12053 * expmed.c (simple_mem_bitfield_p): Likewise.
12054 * expr.c (alignment_for_piecewise_move): Likewise.
12055 (emit_group_load_1): Likewise.
12056 (emit_group_store): Likewise.
12057 (copy_blkmode_from_reg): Likewise.
12058 (emit_push_insn): Likewise.
12059 (expand_assignment): Likewise.
12060 (store_field): Likewise.
12061 (expand_expr_real_1): Likewise.
12062 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12063 * lra-constraints.c (simplify_operand_subreg): Likewise.
12064 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12065 * gimple-ssa-store-merging.c: Likewise in block comment at start
12067 * tree-ssa-strlen.c: Include target.h.
12068 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
12069 of SLOW_UNALIGNED_ACCESS.
12070 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
12072 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12074 PR rtl-optimization/82185
12075 * expmed.c (emit_store_flag_int): Only test tem if it has been
12078 2017-09-12 Richard Biener <rguenther@suse.de>
12080 PR middle-end/82149
12081 * match.pd ((FTYPE) N CMP CST): Fix typo.
12083 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
12085 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
12087 (mips_near_type_p): Add 'short_call' attribute as a synonym
12089 * doc/extend.texi (short_call): Document new function attribute.
12091 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12094 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
12095 assertion check that in the condition.
12096 (get_atomic_generic_size): Likewise. Before testing if parameter
12097 has pointer type, if it has array type, call for C++
12098 default_conversion to perform array-to-pointer conversion.
12100 2017-09-12 Richard Biener <rguenther@suse.de>
12102 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
12103 for operations we cannot scalarize.
12105 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
12107 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
12108 vectors heap vectors. Clean up comments.
12109 Make visited_bbs a reference.
12110 (profitable_jump_thread_path): Make GC
12111 vectors heap vectors. Clean up comments.
12113 (convert_and_register_jump_thread_path): Make GC vectors heap
12115 (check_subpath_and_update_thread_path): Same. Clean up comments.
12116 Make visited_bbs a reference.
12117 (handle_phi): Abstract common code to to
12118 register_jump_thread_path_if_profitable.
12119 Rename VAR_BB to DEF_BB.
12121 Make GC vectors heap vectors.
12122 Make visited_bbs a reference.
12123 (handle_assignment): Same.
12124 (register_jump_thread_path_if_profitable): New.
12125 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
12127 Make GC vectors heap vectors. Clean up comments.
12128 Make visited_bbs a reference.
12129 (find_jump_threads_backwards): Make visited_bbs live in the stack.
12130 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
12133 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
12136 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
12137 words of E_DImode object are reachable by xtensa_uimm8x4 access.
12139 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
12141 Revert r251800 and r251799.
12143 2017-09-11 Martin Jambor <mjambor@suse.cz>
12146 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
12147 arguments in advance.
12148 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
12149 use it to find predecessor edges.
12150 (naive_outof_ssa): Collect vector of predecessors.
12152 2017-09-08 Jason Merrill <jason@redhat.com>
12154 PR c++/70029 - ICE with ref-qualifier and -flto
12155 * langhooks.h (struct lang_hooks_for_types): Add
12156 copy_lang_qualifiers.
12157 * attribs.c (build_type_attribute_qual_variant): Use it.
12158 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
12160 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
12161 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
12163 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
12166 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
12167 (*mulsi3_sp64): New instruction.
12168 (mulsi3): New expander.
12170 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
12172 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
12174 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12176 * sancov.c: Include memmodel.h.
12178 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
12181 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
12184 2017-09-07 Carl Love <cel@us.ibm.com>
12186 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
12187 the sldi instruction.
12189 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
12191 * sancov.c: Include tm_p.h.
12193 2017-09-07 Jakub Jelinek <jakub@redhat.com>
12196 * output.h (switch_to_other_text_partition): New declaration.
12197 * varasm.c (switch_to_other_text_partition): New function.
12198 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
12199 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
12200 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
12201 to the other text partition before emitting LCL label and switch back
12202 after emitting the word after it.
12204 2017-09-07 Richard Biener <rguenther@suse.de>
12206 * passes.def (pass_split_crit_edges): Remove instance before PRE.
12207 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
12208 critical edges here, after loop init.
12209 (pass_data_pre): Remove PROP_no_crit_edges flags.
12210 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
12211 for valueization of call args to avoid leaking VN_TOP.
12212 (visit_use): Assert we do not visit default defs.
12213 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
12214 Use error_mark_node to more easily detect leaking VN_TOP.
12215 All default-defs are varying, not VN_TOP. Mark them visited.
12216 (run_scc_vn): Make code match comment.
12218 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12220 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
12221 OPTION_MASK_FLOAT128_KEYWORD.
12222 (POWERPC_MASKS): Likewise.
12223 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
12224 support for the -mfloat128-type option, and make -mfloat128
12225 default on PowerPC Linux systems. Define or undefine
12226 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
12227 Define __float128 to be __ieee128 if IEEE 128-bit support is
12228 enabled, or undefine it.
12229 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
12230 Delete defining __FLOAT128_TYPE__.
12231 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
12232 -mfloat128-type option and make -mfloat128 default on PowerPC
12234 (TARGET_FLOAT128_TYPE): Likewise.
12235 (-mfloat128-type): Likewise.
12236 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12237 Delete the -mfloat128-type option and make -mfloat128 default on
12238 PowerPC Linux systems. Always use __ieee128 to be the keyword for
12239 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
12240 128-bit floating point is enabled. Change tests from using
12241 -mfloat128-type to -mfloat128.
12242 (rs6000_mangle_type): Use the correct mangling for the __float128
12243 type even if normal long double is restricted to 64-bits.
12244 (floatn_mode): Enable the _Float128 type by default on VSX Linux
12246 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
12247 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
12248 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
12250 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
12251 documentation for -mfloat128.
12253 2017-09-06 Olivier Hainque <hainque@adacore.com>
12255 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
12257 2017-09-06 Wish Wu <wishwu007@gmail.com>
12258 Jakub Jelinek <jakub@redhat.com>
12260 * asan.c (initialize_sanitizer_builtins): Add
12261 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
12262 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
12263 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
12264 BT_FN_VOID_UINT64_PTR variables.
12265 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
12266 (BT_FN_VOID_UINT16_UINT16): Likewise.
12267 (BT_FN_VOID_UINT32_UINT32): Likewise.
12268 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
12269 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
12270 (BT_FN_VOID_UINT64_PTR): Likewise.
12271 * common.opt (flag_sanitize_coverage): New variable.
12272 (fsanitize-coverage=trace-pc): Remove.
12273 (fsanitize-coverage=): Add.
12274 * flag-types.h (enum sanitize_coverage_code): New enum.
12275 * fold-const.c (fold_range_test): Disable non-short-circuit
12276 optimization if flag_sanitize_coverage.
12277 (fold_truth_andor): Likewise.
12278 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12279 * opts.c (COVERAGE_SANITIZER_OPT): Define.
12280 (coverage_sanitizer_opts): New array.
12281 (get_closest_sanitizer_option): Add OPTS argument, handle also
12282 OPT_fsanitize_coverage_.
12283 (parse_sanitizer_options): Adjusted to also handle
12284 OPT_fsanitize_coverage_.
12285 (common_handle_option): Add OPT_fsanitize_coverage_.
12286 * sancov.c (instrument_comparison, instrument_switch): New function.
12287 (sancov_pass): Add trace-cmp support.
12288 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
12289 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
12290 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
12291 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
12292 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
12293 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
12294 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
12295 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
12296 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
12297 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
12299 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12301 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12302 error. Only quit immediately if parsing is complete.
12303 (BEGIN): Initialize fatal_err and parse_done.
12304 (begin fpu, end fpu): Check number of arguments.
12305 (begin arch, end arch): Likewise.
12306 (begin cpu, end cpu): Likewise.
12307 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12308 (optalias): Likewise.
12310 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12312 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12313 * config/arm/arm-isa.h: Delete. Move definitions to ...
12314 * arm-cpus.in: ... here. Use new feature and fgroup values.
12315 * config/arm/arm.c (arm_option_override): Use lower case for feature
12317 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12318 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12319 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12321 (print_isa_bits_for): New function.
12322 (gen_isa): New function.
12323 (gen_comm_data): Use print_isa_bits_for.
12324 (define feature): New keyword.
12325 (define fgroup): New keyword.
12326 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
12327 (arm-isa.h): Add rule to generate file.
12328 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12329 case for feature bit names.
12331 2017-09-06 Richard Biener <rguenther@suse.de>
12333 * tree-ssa-pre.c (NECESSARY): Remove.
12334 (create_expression_by_pieces): Do not touch pass-local flags.
12335 (insert_into_preds_of_block): Likewise.
12336 (do_pre_regular_insertion): Likewise.
12337 (eliminate_insert): Likewise.
12338 (eliminate_dom_walker::before_dom_children): Likewise.
12339 (fini_eliminate): Do not look at inserted_exprs.
12340 (mark_operand_necessary): Remove.
12341 (remove_dead_inserted_code): Replace with simple work-list
12342 algorithm based on inserted_exprs and SSA uses.
12343 (pass_pre::execute): Re-order fini_eliminate and
12344 remove_dead_inserted_code.
12346 2017-09-06 Olivier Hainque <hainque@adacore.com>
12348 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
12349 for VxWorks 7. Adjust surrounding comments.
12351 2017-09-06 Richard Biener <rguenther@suse.de>
12353 * gimple-ssa-strength-reduction.c
12354 (find_candidates_dom_walker::before_dom_children): Also allow
12357 2017-09-06 Richard Biener <rguenther@suse.de>
12359 PR tree-optimization/82108
12360 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
12361 for gap in the non-permutation SLP case.
12363 2017-09-06 Martin Jambor <mjambor@suse.cz>
12365 PR tree-optimization/82078
12366 * tree-sra.c (sort_and_splice_var_accesses): Move call to
12367 add_access_to_work_queue...
12368 (build_accesses_from_assign): ...here.
12369 (propagate_all_subaccesses): Make sure racc is the group
12370 representative, if there is one.
12372 2017-09-06 Jakub Jelinek <jakub@redhat.com>
12374 PR middle-end/82095
12375 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
12378 2017-09-06 Richard Biener <rguenther@suse.de>
12380 * gimple-ssa-strength-reduction.c
12381 (find_candidates_dom_walker::before_doom_children): Use a
12382 type and not a mode check.
12384 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
12387 * config/arm/predicates.md (arm_general_adddi_operand): Create new
12389 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
12391 2017-09-05 Jeff Law <law@redhat.com>
12393 PR tree-optimization/64910
12394 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
12395 cases where we have 3 or more operands.
12397 2017-09-05 Jakub Jelinek <jakub@redhat.com>
12399 PR middle-end/81768
12400 * omp-low.c (lower_omp_for): Recompute tree invariant if
12401 gimple_omp_for_initial/final is ADDR_EXPR.
12403 PR middle-end/81768
12404 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
12405 into gimple val before gimplification fo the COND_EXPR.
12407 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
12409 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
12410 REGION_COPY argument.
12411 (thread_through_all_blocks): Remove unused argument to
12412 duplicate_thread_path.
12414 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12415 Alan Hayward <alan.hayward@arm.com>
12416 David Sherwood <david.sherwood@arm.com>
12418 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
12419 Take a scalar_mode rather than a machine_mode.
12420 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12421 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
12422 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12423 (aarch64_gen_adjusted_ldpstp): Likewise.
12424 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
12426 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12427 Alan Hayward <alan.hayward@arm.com>
12428 David Sherwood <david.sherwood@arm.com>
12430 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
12431 Take a scalar_int_mode instead of a machine_mode.
12432 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12433 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12434 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12435 (aarch64_simd_attr_length_rglist): Delete.
12436 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
12437 a scalar_int_mode instead of a machine_mode.
12438 (aarch64_add_offset): Likewise.
12439 (aarch64_internal_mov_immediate): Likewise
12440 (aarch64_add_constant_internal): Likewise.
12441 (aarch64_add_constant): Likewise.
12442 (aarch64_movw_imm): Likewise.
12443 (aarch64_rtx_arith_op_extract_p): Likewise.
12444 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12445 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12446 Remove assert that the mode isn't a vector.
12447 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12448 (aarch64_expand_mov_immediate): Update calls after above changes.
12449 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
12450 (aarch64_and_bitmask_imm): Check for scalar integer modes.
12451 (aarch64_move_imm): Likewise.
12452 (aarch64_can_const_movi_rtx_p): Likewise.
12453 (aarch64_strip_extend): Likewise.
12454 (aarch64_extr_rtx_p): Likewise.
12455 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
12456 a CONST_INT when the mode parameter is VOIDmode.
12457 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
12459 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12461 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
12462 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12463 (bitwise_type_for_mode): Update accordingly.
12465 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12467 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
12468 * stor-layout.c (mode_for_size_tree): Likewise.
12469 (mode_for_array): Update accordingly.
12470 (layout_decl): Likewise.
12471 (compute_record_mode): Likewise. Only set the mode once.
12473 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12475 * target.def (get_mask_mode): Change return type to opt_mode.
12477 * doc/tm.texi: Regenerate.
12478 * targhooks.h (default_get_mask_mode): Return an opt_mode.
12479 * targhooks.c (default_get_mask_mode): Likewise.
12480 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12481 * optabs-query.c (can_vec_mask_load_store_p): Update use of
12482 targetm.get_mask_mode.
12483 * tree.c (build_truth_vector_type): Likewise.
12485 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12487 * machmode.h (mode_for_vector): Return an opt_mode.
12488 * stor-layout.c (mode_for_vector): Likewise.
12489 (mode_for_int_vector): Update accordingly.
12490 (layout_type): Likewise.
12491 * config/i386/i386.c (emit_memmov): Likewise.
12492 (ix86_expand_set_or_movmem): Likewise.
12493 (ix86_expand_vector_init): Likewise.
12494 (ix86_get_mask_mode): Likewise.
12495 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
12497 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
12498 * expmed.c (extract_bit_field_1): Likewise.
12499 * expr.c (expand_expr_real_2): Likewise.
12500 * optabs-query.c (can_vec_perm_p): Likewise.
12501 (can_vec_mask_load_store_p): Likewise.
12502 * optabs.c (expand_vec_perm): Likewise.
12503 * targhooks.c (default_get_mask_mode): Likewise.
12504 * tree-vect-stmts.c (vectorizable_store): Likewise.
12505 (vectorizable_load): Likewise.
12506 (get_vectype_for_scalar_type_and_size): Likewise.
12508 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12510 * machmode.h (mode_for_int_vector): New function.
12511 * stor-layout.c (mode_for_int_vector): Likewise.
12512 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
12513 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
12514 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12515 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12516 (s390_expand_vcond): Likewise.
12518 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12520 * machmode.h (opt_machine_mode): New type.
12521 (opt_mode<T>): Allow construction from anything that can be
12523 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
12524 (mode_for_size): Return an opt_machine_mode.
12525 * stor-layout.c (mode_for_size): Likewise.
12526 (mode_for_size_tree): Update call accordingly.
12527 (bitwise_mode_for_mode): Likewise.
12528 (make_fract_type): Likewise.
12529 (make_accum_type): Likewise.
12530 * caller-save.c (replace_reg_with_saved_mem): Update call
12532 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12533 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12534 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12535 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12536 * expmed.c (extract_bit_field_1): Likewise.
12537 * reload.c (get_secondary_mem): Likewise.
12538 * varasm.c (assemble_integer): Likewise.
12539 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
12542 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12544 * machmode.h (decimal_float_mode_for_size): New function.
12545 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
12546 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
12547 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
12548 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
12549 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
12550 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
12552 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12554 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
12555 (get_builtin_sync_mode): Likewise.
12556 (expand_ifn_atomic_compare_exchange): Likewise.
12557 (expand_builtin_atomic_clear): Likewise.
12558 (expand_builtin_atomic_test_and_set): Likewise.
12559 (fold_builtin_atomic_always_lock_free): Likewise.
12560 * calls.c (compute_argument_addresses): Likewise.
12561 (emit_library_call_value_1): Likewise.
12562 (store_one_arg): Likewise.
12563 * combine.c (combine_instructions): Likewise.
12564 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
12565 * config/arm/arm.c (arm_function_value): Likewise.
12566 (aapcs_allocate_return_reg): Likewise.
12567 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
12568 * config/i386/i386.c (construct_container): Likewise.
12569 (ix86_gimplify_va_arg): Likewise.
12570 (ix86_expand_sse_cmp): Likewise.
12571 (emit_memmov): Likewise.
12572 (emit_memset): Likewise.
12573 (expand_small_movmem_or_setmem): Likewise.
12574 (ix86_expand_pextr): Likewise.
12575 (ix86_expand_pinsr): Likewise.
12576 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
12577 * config/microblaze/microblaze.c (microblaze_block_move_straight):
12579 * config/mips/mips.c (mips_function_value_1) Likewise.
12580 (mips_block_move_straight): Likewise.
12581 (mips_expand_ins_as_unaligned_store): Likewise.
12582 * config/powerpcspe/powerpcspe.c
12583 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12584 (rs6000_darwin64_record_arg_flush): Likewise.
12585 * config/rs6000/rs6000.c
12586 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12587 (rs6000_darwin64_record_arg_flush): Likewise.
12588 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
12589 (sparc_function_value_1): Likewise.
12590 * config/spu/spu.c (adjust_operand): Likewise.
12591 (spu_emit_branch_or_set): Likewise.
12592 (arith_immediate_p): Likewise.
12593 * emit-rtl.c (gen_lowpart_common): Likewise.
12594 * expr.c (expand_expr_real_1): Likewise.
12595 * function.c (assign_parm_setup_block): Likewise.
12596 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
12597 * reload1.c (alter_reg): Likewise.
12598 * stor-layout.c (mode_for_vector): Likewise.
12599 (layout_type): Likewise.
12601 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12603 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
12604 (spu_convert_move): Likewise.
12605 * lower-subreg.c (resolve_simple_move): Likewise.
12607 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12610 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
12611 define_insn to a define_expand.
12612 (altivec_vsum2sws_direct): New define_insn.
12613 (altivec_vsumsws): Convert from a define_insn to a define_expand.
12615 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12617 * config/arm/arm.c (arm_option_params_internal): Improve setting of
12620 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
12625 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
12626 realign stack if stack alignment needed is less than incoming
12629 2017-09-05 Marek Polacek <polacek@redhat.com>
12632 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
12635 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12637 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
12639 2017-09-05 Richard Biener <rguenther@suse.de>
12641 PR tree-optimization/82084
12642 * fold-const.c (can_native_encode_string_p): Handle wide characters.
12644 2017-09-05 Richard Biener <rguenther@suse.de>
12646 PR tree-optimization/82102
12647 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
12649 2017-09-05 Martin Liska <mliska@suse.cz>
12651 PR tree-optimization/82032
12652 * tree-cfg.c (generate_range_test): New function.
12653 * tree-cfg.h (generate_range_test): Declared here.
12654 * tree-cfgcleanup.c (convert_single_case_switch): New function.
12655 (cleanup_control_expr_graph): Use it.
12656 * tree-switch-conversion.c (try_switch_expansion): Remove
12658 (emit_case_nodes): Use generate_range_test.
12660 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12663 * config/i386/i386.md (*<btsc><mode>_mask): Add
12664 TARGET_USE_BT to insn constraint.
12665 (*btr<mode>_mask): Ditto.
12667 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
12669 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
12670 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
12672 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
12675 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
12676 TARGET_NEON and TARGET_IWMMXT.
12677 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
12678 TARGET_NEON and TARGET_IWMMXT.
12679 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
12681 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12683 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
12684 (ix86_rewrite_tls_address): Ditto.
12685 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
12686 (ix86_rewrite_tls_address_1): Ditto.
12687 (ix86_rewrite_tls_address): Ditto.
12688 * config/i386/predicates.md (tls_address_pattern): New predicate.
12689 * config/i386/i386.md (TLS address splitter): New splitter.
12691 2017-09-04 Richard Biener <rguenther@suse.de>
12693 PR tree-optimization/82084
12694 * fold-const.h (can_native_encode_string_p): Declare.
12695 * fold-const.c (can_native_encode_string_p): Factor out from ...
12696 (native_encode_string): ... here.
12697 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
12698 vectorizing stores from constants we later cannot handle.
12700 2017-09-04 Marek Polacek <polacek@redhat.com>
12703 * doc/invoke.texi: Update -Wtautological-compare documentation.
12705 2017-09-04 Jeff Law <law@redhat.com>
12707 PR tree-optimization/64910
12708 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
12709 swap the first and last operand if the last is a constant.
12711 2017-09-04 Marek Polacek <polacek@redhat.com>
12714 * convert.c (do_narrow): When sanitizing signed integer overflows,
12715 bail out for signed types.
12716 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
12718 2017-09-04 Richard Biener <rguenther@suse.de>
12720 PR tree-optimization/82060
12721 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12722 Move devirtualization after stmt folding and before EH/AB/noreturn
12723 cleanup to get the stmt refs canonicalized. Use a bool instead
12724 of gimple_modified_p since that doesn't work for NOPs. Schedule
12725 NOPs generated by folding for removal.
12727 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12728 Alan Hayward <alan.hayward@arm.com>
12729 David Sherwood <david.sherwood@arm.com>
12731 * coretypes.h (pad_direction): New enum.
12732 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
12733 (FUNCTION_ARG_PADDING): Likewise.
12734 * target.def (function_arg_padding): New hook.
12735 * targhooks.h (default_function_arg_padding): Declare.
12736 * targhooks.c (default_function_arg_padding): New function.
12737 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
12738 (TARGET_FUNCTION_ARG_PADDING): ...this.
12739 * doc/tm.texi: Regenerate.
12740 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
12741 instead of direction.
12742 (compute_argument_addresses): Likewise.
12743 (load_register_parameters): Likewise.
12744 (emit_library_call_value_1): Likewise.
12745 (store_one_arg): Use targetm.calls.function_arg_padding instead
12746 of FUNCTION_ARG_PADDING.
12747 (must_pass_in_stack_var_size_or_pad): Likewise.
12748 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
12749 (emit_group_store): Likewise.
12750 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
12751 instead of FUNCTION_ARG_PADDING.
12752 (emit_push_insn): Likewise, and propagate enum change throughout
12754 * function.h (direction): Delete.
12755 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
12757 * function.c (assign_parm_find_stack_rtl): Likewise.
12758 (assign_parm_setup_block_p): Likewise.
12759 (assign_parm_setup_block): Likewise.
12760 (gimplify_parameters): Likewise.
12761 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
12762 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
12764 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
12765 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12766 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
12767 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
12768 (aarch64_function_arg_padding): ...this new function.
12769 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
12770 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12771 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
12772 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12773 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
12774 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12775 (arm_pad_arg_upward): Replace with...
12776 (arm_function_arg_padding): ...this new function.
12777 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
12779 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
12780 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
12781 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12782 (ia64_hpux_function_arg_padding): Replace with...
12783 (ia64_function_arg_padding): ...this new function. Use pad_direction
12784 instead of direction. Check for TARGET_HPUX.
12785 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
12786 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12787 (iq2000_function_arg_padding): New function.
12788 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
12789 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
12790 (mips_function_arg_padding): ...this new function.
12791 (mips_pad_reg_upward): Update accordingly.
12792 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12793 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
12794 targetm.calls.function_arg_padding.
12795 (FUNCTION_ARG_PADDING): Delete.
12796 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12797 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
12798 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12799 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
12800 (nios2_block_reg_padding): Return pad_direction instead of direction.
12801 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
12802 instead of direction.
12803 (nios2_function_arg_padding): Likewise. Make static.
12804 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12805 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
12806 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
12807 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
12808 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12809 (pa_function_arg_padding): Make static. Return pad_direction instead
12811 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
12812 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12813 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
12814 instead of direction. Use targetm.calls.function_arg_padding.
12815 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
12816 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12817 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
12818 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
12819 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
12821 (function_arg_padding): Rename to...
12822 (rs6000_function_arg_padding): ...this. Make static. Return
12823 pad_direction instead of direction.
12824 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12825 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
12826 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12827 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
12828 instead of direction. Use targetm.calls.function_arg_padding.
12829 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
12830 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12831 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
12832 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
12833 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12834 (function_arg_padding): Rename to...
12835 (rs6000_function_arg_padding): ...this. Make static. Return
12836 pad_direction instead of direction.
12837 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12838 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
12839 * config/s390/s390.c (s390_function_arg_padding): New function.
12840 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12841 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
12842 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
12843 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12844 (function_arg_padding): Rename to...
12845 (sparc_function_arg_padding): ...this. Make static. Return
12846 pad_direction instead of direction.
12847 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
12848 * config/spu/spu.c (spu_function_arg_padding): New function.
12849 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12850 * system.h (FUNCTION_ARG_PADDING): Poison.
12852 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12853 Alan Hayward <alan.hayward@arm.com>
12854 David Sherwood <david.sherwood@arm.com>
12856 * target.def (modes_tieable_p): New hook.
12857 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
12858 (TARGET_MODES_TIEABLE_P): ...this.
12859 * doc/tm.texi.in: Regenerate.
12860 * hooks.h (hook_bool_mode_mode_true): Declare.
12861 * hooks.c (hook_bool_mode_mode_true): New function.
12862 * combine.c (subst): Use targetm.modes_tieable_p instead of
12864 * dse.c (find_shift_sequence): Likewise.
12865 * expmed.c (extract_low_bits): Likewise.
12866 * lower-subreg.c: Include target.h.
12867 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
12869 * rtlanal.c (rtx_cost): Likewise.
12870 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
12871 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
12872 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
12873 (TARGET_MODES_TIEABLE_P): Redefine.
12874 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
12875 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
12876 (TARGET_MODES_TIEABLE_P): Redefine.
12877 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
12878 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
12879 (arc_modes_tieable_p): New function.
12880 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
12881 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
12882 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
12883 (arm_modes_tieable_p): Make static.
12884 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
12885 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
12886 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
12887 (TARGET_MODES_TIEABLE_P): Redefine.
12888 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
12889 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
12890 (TARGET_MODES_TIEABLE_P): Redefine.
12891 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
12892 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
12893 (cr16_modes_tieable_p): New function.
12894 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
12895 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
12896 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
12897 (TRULY_NOOP_TRUNCATION): Update comment.
12898 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
12899 (TRULY_NOOP_TRUNCATION): Update comment.
12900 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
12901 (frv_modes_tieable_p): New function.
12902 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
12903 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
12904 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
12905 (TARGET_MODES_TIEABLE_P): Redefine.
12906 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
12907 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
12908 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
12909 (TARGET_MODES_TIEABLE_P): Redefine.
12910 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
12911 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
12912 (ia64_modes_tieable_p): New function.
12913 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
12914 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
12915 (iq2000_modes_tieable_p): New function.
12916 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
12917 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
12918 (lm32_modes_tieable_p): New function.
12919 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
12920 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
12921 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
12922 (TARGET_MODES_TIEABLE_P): Redefine.
12923 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
12924 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
12925 (m32r_modes_tieable_p): New function.
12926 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
12927 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
12928 (m68k_modes_tieable_p): New function.
12929 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
12930 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
12931 (mcore_modes_tieable_p): New function.
12932 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
12933 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
12935 (TARGET_MODES_TIEABLE_P): Redefine.
12936 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
12937 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
12938 * config/mips/mips.c (mips_modes_tieable_p): Make static.
12939 (TARGET_MODES_TIEABLE_P): Redefine.
12940 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
12941 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
12942 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
12943 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
12944 (mn10300_modes_tieable_p): ...this and make static.
12945 (TARGET_MODES_TIEABLE_P): Redefine.
12946 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
12947 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
12948 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
12949 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
12950 (msp430_modes_tieable_p): Make static.
12951 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
12952 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
12953 (TARGET_MODES_TIEABLE_P): Redefine.
12954 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
12955 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
12956 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
12957 (TARGET_MODES_TIEABLE_P): Redefine.
12958 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
12959 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
12960 * config/pa/pa.c (pa_modes_tieable_p): Make static.
12961 (TARGET_MODES_TIEABLE_P): Redefine.
12962 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
12963 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
12964 (pdp11_modes_tieable_p): New function.
12965 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
12966 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
12967 (rs6000_modes_tieable_p): New function.
12968 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12969 * config/powerpcspe/powerpcspe.md: Update comment.
12970 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
12971 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
12972 (TARGET_MODES_TIEABLE_P): Redefine.
12973 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
12974 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
12975 (rl78_modes_tieable_p): New function.
12976 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
12977 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
12978 (rs6000_modes_tieable_p): New function.
12979 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12980 * config/rs6000/rs6000.md: Update comment.
12981 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
12982 * config/rx/rx.c (rx_modes_tieable_p): New function.
12983 (TARGET_MODES_TIEABLE_P): Redefine.
12984 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
12985 * config/s390/s390.c (s390_modes_tieable_p): New function.
12986 (TARGET_MODES_TIEABLE_P): Redefine.
12987 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
12988 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
12989 (sh_modes_tieable_p): New function.
12990 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
12991 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
12992 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
12993 (sparc_modes_tieable_p): Make static.
12994 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
12995 * config/spu/spu.c (spu_modes_tieable_p): New function.
12996 (TARGET_MODES_TIEABLE_P): Redefine.
12997 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
12998 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
12999 (TARGET_MODES_TIEABLE_P): Redefine.
13000 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13001 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13002 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13003 * config/v850/v850.c (v850_modes_tieable_p): New function.
13004 (TARGET_MODES_TIEABLE_P): Redefine.
13005 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13006 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13007 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13008 (visium_modes_tieable_p): New function.
13009 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13010 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13011 (xtensa_modes_tieable_p): New function.
13012 * system.h (MODES_TIEABLE_P): Poison.
13014 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13015 Alan Hayward <alan.hayward@arm.com>
13016 David Sherwood <david.sherwood@arm.com>
13018 * target.def (hard_regno_mode_ok): New hook.
13019 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
13020 (TARGET_HARD_REGNO_MODE_OK): ...this.
13021 * doc/tm.texi.in: Regenerate.
13022 * hooks.h (hook_bool_uint_mode_true): Declare.
13023 * hooks.c (hook_bool_uint_mode_true): New function.
13024 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
13025 HARD_REGNO_MODE_OK.
13026 * genpreds.c (write_insn_preds_c): Add an include of target.h.
13027 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
13028 instead of HARD_REGNO_MODE_OK.
13029 * caller-save.c: Include target.h.
13030 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
13031 HARD_REGNO_MODE_OK.
13032 * combine.c (can_combine_p): Likewise.
13033 (combinable_i3pat): Likewise.
13034 (can_change_dest_mode): Likewise.
13035 * expr.c (init_expr_target): Likewise.
13036 (convert_move): Likewise.
13037 (convert_modes): Likewise.
13038 * ira.c (setup_prohibited_class_mode_regs): Likewise.
13039 (setup_prohibited_mode_move_regs): Likewise.
13040 * ira.h (target_ira): Likewise.
13041 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13042 * lra-constraints.c (process_alt_operands): Likewise.
13043 (split_reg): Likewise.
13044 * recog.c (peep2_find_free_register): Likewise.
13045 * ree.c (combine_reaching_defs): Likewise.
13046 * regcprop.c (maybe_mode_change): Likewise.
13047 * reginfo.c (init_reg_sets_1): Likewise.
13048 (choose_hard_reg_mode): Likewise.
13049 (simplifiable_subregs): Likewise.
13050 * regrename.c (check_new_reg_p): Likewise.
13051 * reload.c (find_valid_class): Likewise.
13052 (find_valid_class_1): Likewise.
13053 (reload_inner_reg_of_subreg): Likewise.
13054 (push_reload): Likewise.
13055 (combine_reloads): Likewise.
13056 (find_dummy_reload): Likewise.
13057 (find_reloads): Likewise.
13058 * reload1.c (find_reg): Likewise.
13059 (set_reload_reg): Likewise.
13060 (allocate_reload_reg): Likewise.
13061 (choose_reload_regs): Likewise.
13062 (reload_adjust_reg_for_temp): Likewise.
13063 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
13064 (simplify_subreg_regno): Likewise.
13065 * sel-sched.c (init_regs_for_mode): Likewise.
13066 * varasm.c (make_decl_rtl): Likewise.
13067 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
13068 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
13069 HARD_REGNO_MODE_OK.
13070 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
13071 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
13072 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13073 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
13074 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
13075 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13076 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
13077 (arc_mode_class): Delete.
13078 (HARD_REGNO_MODE_OK): Delete.
13079 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13080 (arc_hard_regno_mode_ok): Rename old array to...
13081 (arc_hard_regno_mode_ok_modes): ...this.
13082 (arc_conditional_register_usage): Update accordingly.
13083 (arc_mode_class): Make static.
13084 (arc_hard_regno_mode_ok): New function.
13085 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
13086 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
13087 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13088 (arm_hard_regno_mode_ok): Make static.
13089 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
13090 HARD_REGNO_MODE_OK.
13091 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
13092 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
13093 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
13095 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13096 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
13097 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
13098 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
13099 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
13100 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13101 * config/bfin/predicates.md (valid_reg_operand): Use
13102 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
13103 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
13104 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
13105 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13106 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
13107 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
13108 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13109 (cr16_hard_regno_mode_ok): Make static and return a bool.
13110 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
13111 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13112 (cris_hard_regno_mode_ok): New function.
13113 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
13114 (epiphany_mode_class): Delete.
13115 (HARD_REGNO_MODE_OK): Delete.
13116 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
13117 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13118 (hard_regno_mode_ok): Rename to...
13119 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
13120 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
13121 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
13122 HARD_REGNO_MODE_OK.
13123 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
13124 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
13125 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13126 (frv_hard_regno_mode_ok): Make static and return a bool.
13127 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
13128 HARD_REGNO_MODE_OK.
13129 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
13130 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
13131 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
13132 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
13134 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13135 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
13136 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
13137 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
13139 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13140 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
13141 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13142 (ia64_hard_regno_mode_ok): New function.
13143 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
13144 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13145 (iq2000_hard_regno_mode_ok): New function.
13146 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
13147 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13148 (lm32_hard_regno_mode_ok): New function.
13149 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
13150 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
13151 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
13152 instead of HARD_REGNO_MODE_OK.
13153 (m32c_hard_regno_ok): Rename to...
13154 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
13155 (m32c_cannot_change_mode_class): Update accordingly.
13156 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13157 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
13158 (m32r_mode_class): Delete.
13159 (HARD_REGNO_MODE_OK): Delete.
13160 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13161 (m32r_hard_regno_mode_ok): Rename to...
13162 (m32r_hard_regno_modes): ...this.
13163 (m32r_mode_class): Make static.
13164 (m32r_hard_regno_mode_ok): New function.
13165 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
13166 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
13167 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13168 (m68k_hard_regno_mode_ok): Make static.
13169 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
13170 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13171 (mcore_hard_regno_mode_ok): New function.
13172 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
13173 (HARD_REGNO_MODE_OK): Delete.
13174 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
13176 (microblaze_hard_regno_mode_ok_p): ...this and make static.
13177 (microblaze_hard_regno_mode_ok): New function.
13178 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13179 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
13180 (mips_hard_regno_mode_ok): Delete.
13181 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
13182 (mips_hard_regno_mode_ok_p): ...this and make static.
13183 (mips_hard_regno_mode_ok_p): Rename to...
13184 (mips_hard_regno_mode_ok_uncached): ...this.
13185 (mips_hard_regno_mode_ok): New function.
13186 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
13187 of HARD_REGNO_MODE_OK.
13188 (mips_option_override): Update after above name changes.
13189 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13190 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
13191 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
13192 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
13193 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
13194 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13195 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
13196 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
13197 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
13198 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13199 (msp430_hard_regno_mode_ok): Make static and return a bool.
13200 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
13201 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
13202 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
13204 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13205 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
13206 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
13207 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
13208 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
13209 (PA_HARD_REGNO_MODE_OK): ...this
13210 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
13211 (PA_HARD_REGNO_MODE_OK): ...this.
13212 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13213 (pa_hard_regno_mode_ok): New function.
13214 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
13215 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13216 (pdp11_hard_regno_mode_ok): New function.
13217 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
13218 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
13220 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
13222 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13223 (rs6000_hard_regno_mode_ok): Rename to...
13224 (rs6000_hard_regno_mode_ok_uncached): ...this.
13225 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13226 (rs6000_hard_regno_mode_ok): New function.
13227 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
13228 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
13229 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
13230 (riscv_hard_regno_mode_ok): ...this and make static.
13231 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13232 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
13233 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
13234 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13235 (rl78_hard_regno_mode_ok): Make static and return bool.
13236 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
13237 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
13239 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
13240 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13241 (rs6000_hard_regno_mode_ok): Rename to...
13242 (rs6000_hard_regno_mode_ok_uncached): ...this.
13243 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13244 (rs6000_hard_regno_mode_ok): New function.
13245 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
13246 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
13247 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13248 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
13249 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
13250 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
13251 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13252 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
13253 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
13254 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13255 (sh_hard_regno_mode_ok): Make static.
13256 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
13257 instead of HARD_REGNO_MODE_OK.
13258 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
13259 (sparc_mode_class): Delete.
13260 (HARD_REGNO_MODE_OK): Delete.
13261 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13262 (hard_regno_mode_classes): Make static.
13263 (sparc_mode_class): Likewise.
13264 (sparc_hard_regno_mode_ok): New function.
13265 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
13266 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
13267 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
13269 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13270 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
13271 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
13272 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
13273 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
13274 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13275 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
13276 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
13277 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13278 (visium_hard_regno_mode_ok): New function.
13279 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
13280 instead of HARD_REGNO_MODE_OK.
13281 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
13282 (HARD_REGNO_MODE_OK): Delete.
13283 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
13284 (xtensa_hard_regno_mode_ok_p): ...this and make static.
13285 (xtensa_option_override): Update accordingly.
13286 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13287 (xtensa_hard_regno_mode_ok): New function.
13288 * system.h (HARD_REGNO_MODE_OK): Poison.
13290 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13291 Alan Hayward <alan.hayward@arm.com>
13292 David Sherwood <david.sherwood@arm.com>
13294 * target.def (hard_regno_call_part_clobbered): New hook.
13295 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
13296 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
13297 * doc/tm.texi: Regenerate.
13298 * hooks.h (hook_bool_uint_mode_false): Declare.
13299 * hooks.c (hook_bool_uint_mode_false): New function.
13300 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13301 * cselib.c (cselib_process_insn): Use
13302 targetm.hard_regno_call_part_clobbered instead of
13303 HARD_REGNO_CALL_PART_CLOBBERED.
13304 * ira-conflicts.c (ira_build_conflicts): Likewise.
13305 * ira-costs.c (ira_tune_allocno_costs): Likewise.
13306 * lra-constraints.c (need_for_call_save_p): Likewise.
13307 * lra-lives.c: Include target.h.
13308 (check_pseudos_live_through_calls): Use
13309 targetm.hard_regno_call_part_clobbered instead of
13310 HARD_REGNO_CALL_PART_CLOBBERED.
13311 * regcprop.c: Include target.h.
13312 (copyprop_hardreg_forward_1): Use
13313 targetm.hard_regno_call_part_clobbered instead of
13314 HARD_REGNO_CALL_PART_CLOBBERED.
13315 * reginfo.c (choose_hard_reg_mode): Likewise.
13316 * regrename.c (check_new_reg_p): Likewise.
13317 * reload.c (find_equiv_reg): Likewise.
13318 * reload1.c (emit_reload_insns): Likewise.
13319 * sched-deps.c (deps_analyze_insn): Likewise.
13320 * sel-sched.c (init_regs_for_mode): Likewise.
13321 (mark_unavailable_hard_regs): Likewise.
13322 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
13323 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13324 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
13326 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13327 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13328 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
13330 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
13332 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13333 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13334 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
13336 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13337 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13338 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
13340 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13341 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
13343 * config/powerpcspe/powerpcspe.c
13344 (rs6000_hard_regno_call_part_clobbered): New function.
13345 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13346 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13347 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
13349 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13350 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13351 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
13353 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13354 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13355 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
13357 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13358 Alan Hayward <alan.hayward@arm.com>
13359 David Sherwood <david.sherwood@arm.com>
13361 * rtl.h (subreg_memory_offset): Declare.
13362 * emit-rtl.c (subreg_memory_offset): New function.
13363 * expmed.c (store_bit_field_1): Use it.
13364 * expr.c (undefined_operand_subword_p): Likewise.
13365 * simplify-rtx.c (simplify_subreg): Likewise.
13367 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
13369 PR rtl-optimization/57448
13372 * optabs.c (expand_atomic_load): Place compiler memory barriers if
13373 using atomic_load pattern.
13374 (expand_atomic_store): Likewise.
13376 2017-09-04 Jakub Jelinek <jakub@redhat.com>
13379 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
13380 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
13381 handling. Use replace_call_with_value with NULL instead of
13382 gsi_replace, unlink_stmt_vdef and release_defs.
13384 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
13387 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
13389 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13392 * rtl.h (emit_library_call_value_1): Declare.
13393 (emit_library_call): Replace declaration with a series of overloads.
13394 Remove the parameter count argument.
13395 (emit_library_call_value): Likewise.
13396 * calls.c (emit_library_call_value_1): Make global. Replace varargs
13397 with an "rtx_mode_t *".
13398 (emit_library_call_value): Delete.
13399 (emit_library_call): Likewise.
13400 * asan.c (asan_emit_stack_protection): Update calls accordingly.
13401 (asan_emit_allocas_unpoison): Likewise.
13402 * builtins.c (expand_builtin_powi): Likewise.
13403 (expand_asan_emit_allocas_unpoison): Likewise.
13404 * cfgexpand.c (expand_main_function): Likewise.
13405 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
13406 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
13407 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
13408 * config/arm/arm.c (arm_trampoline_init): Likewise.
13409 (arm_call_tls_get_addr): Likewise.
13410 (arm_expand_divmod_libfunc): Likewise.
13411 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
13412 (smulsi3_highpart): Likewise.
13413 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
13414 (c6x_expand_compare): Likewise.
13415 (c6x_expand_movmem): Likewise.
13416 * config/frv/frv.c (frv_trampoline_init): Likewise.
13417 * config/i386/i386.c (ix86_trampoline_init): Likewise.
13418 (ix86_expand_divmod_libfunc): Likewise.
13419 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
13420 (ia64_expand_compare): Likewise.
13421 (ia64_profile_hook): Likewise.
13422 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
13423 (nonlocal_goto): Likewise.
13424 (restore_stack_nonlocal): Likewise.
13425 * config/m32r/m32r.c (block_move_call): Likewise.
13426 (m32r_trampoline_init): Likewise.
13427 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
13428 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
13429 (m68k_call_m68k_read_tp): Likewise.
13430 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
13431 (microblaze_expand_divide): Likewise.
13432 * config/mips/mips.h (mips_args): Likewise.
13433 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
13434 (MIPS_ICACHE_SYNC): Likewise.
13435 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
13436 (nios2_trampoline_init): Likewise.
13437 * config/pa/pa.c (hppa_tls_call): Likewise.
13438 (pa_trampoline_init): Likewise.
13439 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
13440 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
13441 (expand_strn_compare): Likewise.
13442 (rs6000_generate_compare): Likewise.
13443 (rs6000_expand_float128_convert): Likewise.
13444 (output_profile_hook): Likewise.
13445 (rs6000_trampoline_init): Likewise.
13446 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
13447 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
13448 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
13449 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
13450 (rs6000_generate_compare): Likewise.
13451 (rs6000_expand_float128_convert): Likewise.
13452 (output_profile_hook): Likewise.
13453 (rs6000_trampoline_init): Likewise.
13454 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
13455 * config/sh/sh.c (sh_trampoline_init): Likewise.
13456 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
13457 (sparc_emit_float_lib_cmp): Likewise.
13458 (sparc32_initialize_trampoline): Likewise.
13459 (sparc64_initialize_trampoline): Likewise.
13460 (sparc_profile_hook): Likewise.
13461 * config/spu/spu.c (ea_load_store): Likewise.
13462 * config/spu/spu.md (floatunssidf2): Likewise.
13463 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
13464 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
13465 * config/visium/visium.c (expand_block_move_4): Likewise.
13466 (expand_block_move_2): Likewise.
13467 (expand_block_move_1): Likewise.
13468 (expand_block_set_4): Likewise.
13469 (expand_block_set_2): Likewise.
13470 (expand_block_set_1): Likewise.
13471 (visium_trampoline_init): Likewise.
13472 (visium_profile_hook): Likewise.
13473 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
13474 (xtensa_setup_frame_addresses): Likewise.
13475 (xtensa_trampoline_init): Likewise.
13476 * except.c (sjlj_emit_function_enter): Likewise.
13477 (sjlj_emit_function_exit): Likewise.
13478 * explow.c (allocate_dynamic_stack_space): Likewise.
13479 (probe_stack_range): Likewise.
13480 * expr.c (convert_mode_scalar): Likewise.
13481 * optabs.c (expand_binop): Likewise.
13482 (expand_twoval_binop_libfunc): Likewise.
13483 (expand_unop): Likewise.
13484 (prepare_cmp_insn): Likewise.
13485 (prepare_float_lib_cmp): Likewise.
13486 (expand_float): Likewise.
13487 (expand_fix): Likewise.
13488 (expand_fixed_convert): Likewise.
13489 (maybe_emit_sync_lock_test_and_set): Likewise.
13490 (expand_atomic_compare_and_swap): Likewise.
13491 (expand_mem_thread_fence): Likewise.
13492 (expand_atomic_fetch_op): Likewise.
13494 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
13496 * doc/generic.texi (OpenACC): Adjust URL.
13497 * doc/invoke.texi (C Dialect Options): Ditto.
13499 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
13501 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
13502 predicate for operand 1. Add (m,<S>) constraint.
13503 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
13504 Prevent memory operand 1 with register operand 2.
13506 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
13508 PR rtl-optimization/82024
13509 * combine.c (try_combine): If the combination result is a PARALLEL,
13510 and we only need to retain the SET in there that would be placed
13511 at I2, check that we can place that at I3 instead, before doing so.
13513 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13516 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
13518 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
13519 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
13522 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
13523 Jeff Law <law@redhat.com>
13525 * varasm.c (bss_initializer_p): Do not put constants into .bss
13526 (categorize_decl_for_section): Handle bss_initializer_p returning
13527 false when DECL_INITIAL is NULL.
13529 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13532 * config/s390/s390.c (s390_can_inline_p): New function.
13534 2017-09-01 Jeff Law <law@redhat.com>
13536 PR tree-optimization/82052
13537 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13538 Always initialize the returned slot after a hash table miss
13539 when INSERT is true.
13541 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
13543 * config/s390/s390.md (mem_signal_fence): Remove.
13544 * doc/md.texi (mem_signal_fence): Remove.
13545 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
13547 * target-insns.def (mem_signal_fence): Remove.
13549 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13552 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
13555 * asan.c (create_odr_indicator): Strip name encoding from assembler
13556 name before appending it after __odr_asan_.
13558 2017-09-01 Martin Liska <mliska@suse.cz>
13560 PR tree-optimization/82059
13561 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
13562 frequency only when an edge is redirected.
13564 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13566 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
13567 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
13568 (arc_conditional_register_usage): Remove ARC600 lp_count
13570 (arc_file_start): Emit Tag_ARC_CPU_variation.
13571 (arc_can_use_doloop_p): New conditions to use ZOLs.
13572 (hwloop_fail): New function.
13573 (hwloop_optimize): Likewise.
13574 (hwloop_pattern_reg): Likewise.
13575 (arc_doloop_hooks): New struct, to be used with reorg_loops.
13576 (arc_reorg_loops): New function, calls reorg_loops.
13577 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
13578 (arc600_corereg_hazard): Remove ZOL checking, case handled by
13580 (arc_loop_hazard): Remove function, functionality moved into
13582 (arc_hazard): Remove arc_loop_hazard call.
13583 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
13584 into hwloop_optimize.
13585 (arc_label_align): Remove ZOL handling.
13586 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
13587 * config/arc/arc.md (doloop_begin): Remove pattern.
13588 (doloop_begin_i): Likewise.
13589 (doloop_end_i): Likewise.
13590 (doloop_fallback): Likewise.
13591 (doloop_fallback_m): Likewise.
13592 (doloop_end): Reimplement expand.
13593 (arc_lp): New pattern for LP instruction.
13594 (loop_end): New pattern.
13595 (loop_fail): Likewise.
13596 (decrement_and_branch_until_zero): Likewise.
13597 * config/arc/arc.opt (mlpc-width): New option.
13598 * doc/invoke.texi (mlpc-width): Document option.
13600 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13602 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
13603 (arc_ccfsm_advance): Fix checking for delay slots.
13604 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
13606 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13608 * config/arc/arc.md (movqi_insn): Add stores to save constant long
13610 (movhi_insn): Update store instruction constraint which are saving
13611 6-bit short immediates.
13612 (movsi_insn): Consider also short scaled load operations.
13613 (zero_extendhisi2_i): Use Usd constraint instead of T.
13614 (extendhisi2_i): Add q constraint.
13615 (arc_clzsi2): Add type and length attributes.
13616 (arc_ctzsi2): Likewise.
13617 * config/arc/constraints.md (Usc): Update constraint, the
13618 assembler can parse two relocations for a single instruction.
13620 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13622 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
13623 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
13625 2017-08-31 Olivier Hainque <hainque@adacore.com>
13627 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
13628 match as powerpc-wrs-vxworks*.
13630 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
13632 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
13633 register constraint for by-element operand.
13634 (aarch64_mls_elt_merge<mode>): Likewise.
13636 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13638 * config/arc/arc.c (arc_can_follow_jump): Check for short
13641 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13643 * config.gcc: Use g.opt for arc.
13644 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
13645 functionality moved to ...
13646 (legitimate_scaled_address_p): New function, ...here.
13647 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
13648 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
13649 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
13651 (arc_override_options): Handle G option.
13652 (arc_output_pic_addr_const): Correct function definition.
13653 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
13654 (arc_decl_anon_ns_mem_p): Delete.
13655 (arc_in_small_data_p): Overhaul this function to take into
13656 consideration the value given via G option.
13657 (arc_rewrite_small_data_1): Renamed and corrected old
13658 arc_rewrite_small_data function.
13659 (arc_rewrite_small_data): New function.
13660 (small_data_pattern): Don't use pic_offset_table_rtx.
13661 * config/arc/arc.h (CC1_SPEC): Recognize G option.
13662 * config/arc/simdext.md (movmisalignv2hi): Use
13663 prepare_move_operands function.
13665 (movmisalign*): Likewise.
13666 * doc/invoke.texi (ARC options): Document -G option.
13668 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13670 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
13672 * config/arc/arc.c (arc_print_operand): Output scalled address for
13673 sdata whenever is possible.
13674 (arc_in_small_data_p): Allow sdata for 64bit datum when double
13675 load/stores are available.
13676 (compact_sda_memory_operand): Check for the alignment required by
13677 code density instructions.
13678 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
13680 * config/arc/constraints.md (Usd): Update constraint.
13681 (Us0): New constraint.
13682 (Usc): Update constraint.
13684 2017-08-31 Richard Biener <rguenther@suse.de>
13686 PR middle-end/82054
13687 * dwarf2out.c (dwarf2out_early_global_decl): Process each
13688 function only once.
13690 2017-08-31 Tamar Christina <tamar.christina@arm.com>
13692 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
13693 Resize type_signature.
13695 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13696 Alan Hayward <alan.hayward@arm.com>
13697 David Sherwood <david.sherwood@arm.com>
13699 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
13700 subregs whose inner modes can be stored in GPRs.
13701 (aarch64_classify_index): Likewise.
13703 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13704 Alan Hayward <alan.hayward@arm.com>
13705 David Sherwood <david.sherwood@arm.com>
13707 * config/aarch64/iterators.md (V_cmp_result): Rename to...
13708 (V_INT_EQUIV): ...this.
13709 (v_cmp_result): Rename to...
13710 (v_int_equiv): ...this.
13711 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
13712 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
13713 (copysign<mode>3): Likewise.
13714 (aarch64_simd_bsl<mode>_internal): Likewise.
13715 (aarch64_simd_bsl<mode>): Likewise.
13716 (vec_cmp<mode><mode>): Likewise.
13717 (vcond<mode><mode>): Likewise.
13718 (vcond<v_cmp_mixed><mode>): Likewise.
13719 (vcondu<mode><v_cmp_mixed>): Likewise.
13720 (aarch64_cm<optab><mode>): Likewise.
13721 (aarch64_cmtst<mode>): Likewise.
13722 (aarch64_fac<optab><mode>): Likewise.
13723 (vec_perm_const<mode>): Likewise.
13724 (vcond_mask_<mode><v_cmp_result>): Rename to...
13725 (vcond_mask_<mode><v_int_equiv>): ...this.
13726 (vec_cmp<mode><v_cmp_result>): Rename to...
13727 (vec_cmp<mode><v_int_equiv>): ...this.
13729 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13730 Alan Hayward <alan.hayward@arm.com>
13731 David Sherwood <david.sherwood@arm.com>
13733 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
13735 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
13736 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
13737 (UNSPEC_LD4_DREG): New unspecs.
13738 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
13739 (aarch64_ld2<mode>_dreg_be): Replace with...
13740 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
13742 (aarch64_ld3<mode>_dreg_le)
13743 (aarch64_ld3<mode>_dreg_be): Replace with...
13744 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
13746 (aarch64_ld4<mode>_dreg_le)
13747 (aarch64_ld4<mode>_dreg_be): Replace with...
13748 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
13751 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13753 PR tree-optimization/81987
13754 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
13755 insert an initializer in a location not dominated by the stride
13758 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13760 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
13761 on the entire header of the finally block in the fallthru case.
13763 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13765 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
13767 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
13769 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
13770 rs6000_emit_move_from_cr and call renamed function.
13771 (rs6000_emit_prologue): Call renamed functions.
13772 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
13773 movesi_from_cr, remove volatile CRs.
13775 2017-08-30 Jon Beniston <jon@beniston.com>
13776 Richard Biener <rguenther@suse.de>
13778 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
13779 of VECTOR_MODE_P check.
13780 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
13781 element vector types.
13783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13785 * df.h (df_read_modify_subreg_p): Remove in favor of...
13786 * rtl.h (read_modify_subreg_p): ...this new function. Take a
13787 const_rtx instead of an rtx.
13788 * cprop.c (local_cprop_find_used_regs): Update accordingly.
13789 * df-problems.c (df_word_lr_mark_ref): Likewise.
13790 * ira-lives.c (mark_pseudo_reg_live): Likewise.
13791 (mark_pseudo_reg_dead): Likewise.
13792 (mark_ref_dead): Likewise.
13793 * reginfo.c (init_subregs_of_mode): Likewise.
13794 * sched-deps.c (sched_analyze_1): Likewise.
13795 * df-scan.c (df_def_record_1): Likewise.
13796 (df_uses_record): Likewise.
13797 (df_read_modify_subreg_p): Remove in favor of...
13798 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
13799 const_rtx instead of an rtx.
13801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13802 Alan Hayward <alan.hayward@arm.com>
13803 David Sherwood <david.sherwood@arm.com>
13805 * rtl.h (partial_subreg_p): New function.
13806 * caller-save.c (save_call_clobbered_regs): Use it.
13807 * calls.c (expand_call): Likewise.
13808 * combine.c (combinable_i3pat): Likewise.
13809 (simplify_set): Likewise.
13810 (make_extraction): Likewise.
13811 (make_compound_operation_int): Likewise.
13812 (gen_lowpart_or_truncate): Likewise.
13813 (force_to_mode): Likewise.
13814 (make_field_assignment): Likewise.
13815 (reg_truncated_to_mode): Likewise.
13816 (record_truncated_value): Likewise.
13817 (move_deaths): Likewise.
13818 * cse.c (record_jump_cond): Likewise.
13819 (cse_insn): Likewise.
13820 * cselib.c (cselib_lookup_1): Likewise.
13821 * expmed.c (extract_bit_field_using_extv): Likewise.
13822 * function.c (assign_parm_setup_reg): Likewise.
13823 * ifcvt.c (noce_convert_multiple_sets): Likewise.
13824 * ira-build.c (create_insn_allocnos): Likewise.
13825 * lra-coalesce.c (merge_pseudos): Likewise.
13826 * lra-constraints.c (match_reload): Likewise.
13827 (simplify_operand_subreg): Likewise.
13828 (curr_insn_transform): Likewise.
13829 * lra-lives.c (process_bb_lives): Likewise.
13830 * lra.c (new_insn_reg): Likewise.
13831 (lra_substitute_pseudo): Likewise.
13832 * regcprop.c (mode_change_ok): Likewise.
13833 (maybe_mode_change): Likewise.
13834 (copyprop_hardreg_forward_1): Likewise.
13835 * reload.c (push_reload): Likewise.
13836 (find_reloads): Likewise.
13837 (find_reloads_subreg_address): Likewise.
13838 * reload1.c (alter_reg): Likewise.
13839 (eliminate_regs_1): Likewise.
13840 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13842 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
13844 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
13845 back to if statements, including unpack.
13847 2017-08-30 Martin Liska <mliska@suse.cz>
13849 PR inline-asm/82001
13850 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
13852 (func_checker::compare_asm_inputs_outputs): ... this function.
13853 (func_checker::compare_gimple_asm): Use the function to compare
13854 also ASM constrains.
13855 * ipa-icf-gimple.h: Rename the function.
13857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13858 Alan Hayward <alan.hayward@arm.com>
13859 David Sherwood <david.sherwood@arm.com>
13861 * coretypes.h (complex_mode): New type.
13862 * gdbhooks.py (build_pretty_printer): Handle it.
13863 * machmode.h (complex_mode): New class.
13864 (complex_mode::includes_p): New function.
13865 (is_complex_int_mode): Likewise.
13866 (is_complex_float_mode): Likewise.
13867 * genmodes.c (get_mode_class): Handle complex mode classes.
13868 * function.c (expand_function_end): Use is_complex_int_mode.
13870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13871 Alan Hayward <alan.hayward@arm.com>
13872 David Sherwood <david.sherwood@arm.com>
13874 * coretypes.h (scalar_mode_pod): New typedef.
13875 * gdbhooks.py (build_pretty_printer): Handle it.
13876 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
13877 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
13878 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
13879 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
13880 as_a <scalar_mode>.
13882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13883 Alan Hayward <alan.hayward@arm.com>
13884 David Sherwood <david.sherwood@arm.com>
13886 * machmode.h (mode_for_vector): Take a scalar_mode instead
13888 * stor-layout.c (mode_for_vector): Likewise.
13889 * explow.c (promote_mode): Use as_a <scalar_mode>.
13890 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
13892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13893 Alan Hayward <alan.hayward@arm.com>
13894 David Sherwood <david.sherwood@arm.com>
13896 * target.def (preferred_simd_mode): Take a scalar_mode
13897 instead of a machine_mode.
13898 * targhooks.h (default_preferred_simd_mode): Likewise.
13899 * targhooks.c (default_preferred_simd_mode): Likewise.
13900 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
13901 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
13902 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
13903 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
13904 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
13905 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
13906 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
13907 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
13909 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
13910 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
13911 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
13912 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
13913 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
13914 * doc/tm.texi: Regenerate.
13915 * optabs-query.c (can_vec_mask_load_store_p): Return false for
13918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13919 Alan Hayward <alan.hayward@arm.com>
13920 David Sherwood <david.sherwood@arm.com>
13922 * target.def (scalar_mode_supported_p): Take a scalar_mode
13923 instead of a machine_mode.
13924 * targhooks.h (default_scalar_mode_supported_p): Likewise.
13925 * targhooks.c (default_scalar_mode_supported_p): Likewise.
13926 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
13927 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13928 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
13929 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
13930 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
13931 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
13932 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
13933 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
13934 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
13935 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
13936 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
13937 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
13939 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
13940 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
13941 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13942 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13943 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
13945 * doc/tm.texi: Regenerate.
13947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13948 Alan Hayward <alan.hayward@arm.com>
13949 David Sherwood <david.sherwood@arm.com>
13951 * coretypes.h (opt_scalar_mode): New typedef.
13952 * gdbhooks.py (build_pretty_printers): Handle it.
13953 * machmode.h (mode_iterator::get_2xwider): Add overload for
13955 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
13957 * expr.c (convert_mode_scalar): Likewise.
13958 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13959 * optabs.c (expand_float): Likewise.
13960 (expand_fix): Likewise.
13961 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13964 Alan Hayward <alan.hayward@arm.com>
13965 David Sherwood <david.sherwood@arm.com>
13967 * optabs.c (expand_float): Explicitly check for scalars before
13968 using a branching expansion.
13969 (expand_fix): Likewise.
13971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13972 Alan Hayward <alan.hayward@arm.com>
13973 David Sherwood <david.sherwood@arm.com>
13975 * expr.c (convert_mode): Split scalar handling out into...
13976 (convert_mode_scalar): ...this new function. Treat the modes
13979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13980 Alan Hayward <alan.hayward@arm.com>
13981 David Sherwood <david.sherwood@arm.com>
13983 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
13985 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
13987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13988 Alan Hayward <alan.hayward@arm.com>
13989 David Sherwood <david.sherwood@arm.com>
13991 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
13992 rather than a machine_mode.
13993 (fixed_from_string): Likewise.
13994 (fixed_convert): Likewise.
13995 (fixed_convert_from_int): Likewise.
13996 (fixed_convert_from_real): Likewise.
13997 (real_convert_from_fixed): Likewise.
13998 * fixed-value.c (fixed_from_double_int): Likewise.
13999 (fixed_from_string): Likewise.
14000 (fixed_convert): Likewise.
14001 (fixed_convert_from_int): Likewise.
14002 (fixed_convert_from_real): Likewise.
14003 (real_convert_from_fixed): Likewise.
14004 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14007 Alan Hayward <alan.hayward@arm.com>
14008 David Sherwood <david.sherwood@arm.com>
14010 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14011 of separate mode class checks. Do not allow vector modes here.
14012 (immed_wide_int_const): Use as_a <scalar_mode>.
14013 * explow.c (trunc_int_for_mode): Likewise.
14014 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14015 (wi::shwi): Likewise.
14016 (wi::min_value): Likewise.
14017 (wi::max_value): Likewise.
14018 * dwarf2out.c (loc_descriptor): Likewise.
14019 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
14020 for CONST_WIDE_INT.
14022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14023 Alan Hayward <alan.hayward@arm.com>
14024 David Sherwood <david.sherwood@arm.com>
14026 * tree.h (SCALAR_TYPE_MODE): New macro.
14027 * expr.c (expand_expr_addr_expr_1): Use it.
14028 (expand_expr_real_2): Likewise.
14029 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
14030 (fold_convert_const_fixed_from_int): Likewise.
14031 (fold_convert_const_fixed_from_real): Likewise.
14032 (native_encode_fixed): Likewise
14033 (native_encode_complex): Likewise
14034 (native_encode_vector): Likewise.
14035 (native_interpret_fixed): Likewise.
14036 (native_interpret_real): Likewise.
14037 (native_interpret_complex): Likewise.
14038 (native_interpret_vector): Likewise.
14039 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
14040 (simd_clone_adjust_argument_types): Likewise.
14041 (simd_clone_init_simd_arrays): Likewise.
14042 (simd_clone_adjust): Likewise.
14043 * stor-layout.c (layout_type): Likewise.
14044 * tree.c (build_minus_one_cst): Likewise.
14045 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14046 * tree-inline.c (estimate_move_cost): Likewise.
14047 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
14048 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
14049 (vectorizable_reduction): Likewise.
14050 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
14051 (vect_recog_mixed_size_cond_pattern): Likewise.
14052 (check_bool_pattern): Likewise.
14053 (adjust_bool_pattern): Likewise.
14054 (search_type_for_mask_1): Likewise.
14055 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
14056 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14057 (vectorizable_load): Likewise.
14058 (vectorizable_store): Likewise.
14059 * ubsan.c (ubsan_encode_value): Likewise.
14060 * varasm.c (output_constant): Likewise.
14062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14063 Alan Hayward <alan.hayward@arm.com>
14064 David Sherwood <david.sherwood@arm.com>
14066 * coretypes.h (scalar_mode): New class.
14067 * machmode.h (scalar_mode): Likewise.
14068 (scalar_mode::includes_p): New function.
14069 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
14070 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
14071 * genmodes.c (get_mode_class): Handle remaining scalar modes.
14072 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
14073 * expmed.c (store_bit_field_1): Likewise.
14074 (extract_bit_field_1): Likewise.
14075 * expr.c (write_complex_part): Likewise.
14076 (read_complex_part): Likewise.
14077 (emit_move_complex_push): Likewise.
14078 (expand_expr_real_2): Likewise.
14079 * function.c (assign_parm_setup_reg): Likewise.
14080 (assign_parms_unsplit_complex): Likewise.
14081 * optabs.c (expand_binop): Likewise.
14082 * rtlanal.c (subreg_get_info): Likewise.
14083 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14084 * varasm.c (output_constant_pool_2): Likewise.
14086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14087 Alan Hayward <alan.hayward@arm.com>
14088 David Sherwood <david.sherwood@arm.com>
14090 * expmed.c (extract_high_half): Use scalar_int_mode and remove
14092 (expmed_mult_highpart_optab): Likewise.
14093 (expmed_mult_highpart): Likewise.
14095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14096 Alan Hayward <alan.hayward@arm.com>
14097 David Sherwood <david.sherwood@arm.com>
14099 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
14100 instead of a machine_mode.
14101 (builtin_memset_read_str): Likewise.
14102 * builtins.c (c_readstr): Likewise.
14103 (builtin_memcpy_read_str): Likewise.
14104 (builtin_strncpy_read_str): Likewise.
14105 (builtin_memset_read_str): Likewise.
14106 (builtin_memset_gen_str): Likewise.
14107 (expand_builtin_signbit): Use scalar_int_mode for local variables.
14108 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
14109 instead of a machine_mode.
14110 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
14112 (make_extraction): Likewise.
14113 (try_widen_shift_mode): Take and return scalar_int_modes instead
14115 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
14116 a scalar_int_mode instead of a machine_mode.
14117 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
14118 (avr_addr_space_pointer_mode): Likewise.
14119 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
14120 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
14121 (msp430_unwind_word_mode): Likewise.
14122 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
14123 (spu_addr_space_pointer_mode): Likewise.
14124 (spu_addr_space_address_mode): Likewise.
14125 (spu_libgcc_cmp_return_mode): Likewise.
14126 (spu_libgcc_shift_count_mode): Likewise.
14127 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
14128 (rl78_addr_space_pointer_mode): Likewise.
14129 (fl78_unwind_word_mode): Likewise.
14130 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
14132 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
14133 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
14134 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
14135 (mips_valid_pointer_mode): Likewise.
14136 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
14137 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
14138 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
14140 (ft32_addr_space_address_mode): Likewise.
14141 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
14142 scalar_int_mode instead of a machine_mode.
14143 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
14145 (m32c_addr_space_address_mode): Likewise.
14146 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
14147 (rs6000_eh_return_filter_mode): Likewise.
14148 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
14149 (rs6000_eh_return_filter_mode): Likewise.
14150 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
14151 (s390_libgcc_shift_count_mode): Likewise.
14152 (s390_unwind_word_mode): Likewise.
14153 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
14155 * target.def (mode_rep_extended): Likewise.
14156 (valid_pointer_mode): Likewise.
14157 (addr_space.valid_pointer_mode): Likewise.
14158 (eh_return_filter_mode): Return a scalar_int_mode rather than
14160 (libgcc_cmp_return_mode): Likewise.
14161 (libgcc_shift_count_mode): Likewise.
14162 (unwind_word_mode): Likewise.
14163 (addr_space.pointer_mode): Likewise.
14164 (addr_space.address_mode): Likewise.
14165 * doc/tm.texi: Regenerate.
14166 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
14168 (do_jump): Use scalar_int_mode for local variables.
14169 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
14170 rather than a machine_mode.
14171 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
14172 (scompare_loc_descriptor_wide): Likewise.
14173 (scompare_loc_descriptor_narrow): Likewise.
14174 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
14176 * except.c (sjlj_emit_dispatch_table): Likewise.
14177 (expand_builtin_eh_copy_values): Likewise.
14178 * explow.c (convert_memory_address_addr_space_1): Likewise.
14179 Take a scalar_int_mode rather than a machine_mode.
14180 (convert_memory_address_addr_space): Take a scalar_int_mode rather
14181 than a machine_mode.
14182 (memory_address_addr_space): Use scalar_int_mode for local variables.
14183 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
14184 rather than a machine_mode.
14185 * expmed.c (mask_rtx): Likewise.
14186 (init_expmed_one_conv): Likewise.
14187 (expand_mult_highpart_adjust): Likewise.
14188 (extract_high_half): Likewise.
14189 (expmed_mult_highpart_optab): Likewise.
14190 (expmed_mult_highpart): Likewise.
14191 (expand_smod_pow2): Likewise.
14192 (expand_sdiv_pow2): Likewise.
14193 (emit_store_flag_int): Likewise.
14194 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
14196 (extract_low_bits): Likewise.
14197 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
14199 * expr.c (pieces_addr::adjust): Likewise.
14200 (can_store_by_pieces): Likewise.
14201 (store_by_pieces): Likewise.
14202 (clear_by_pieces_1): Likewise.
14203 (expand_expr_addr_expr_1): Likewise.
14204 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
14205 (expand_expr_real_1): Likewise.
14206 (try_casesi): Likewise.
14207 * final.c (shorten_branches): Likewise.
14208 * fold-const.c (fold_convert_const_int_from_fixed): Change the
14209 type of "mode" to machine_mode.
14210 * internal-fn.c (expand_arith_overflow_result_store): Take a
14211 scalar_int_mode rather than a machine_mode.
14212 (expand_mul_overflow): Use scalar_int_mode for local variables.
14213 * loop-doloop.c (doloop_modify): Likewise.
14214 (doloop_optimize): Likewise.
14215 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
14216 than a machine_mode.
14217 (expand_doubleword_shift_condmove): Likewise.
14218 (expand_doubleword_shift): Likewise.
14219 (expand_doubleword_clz): Likewise.
14220 (expand_doubleword_popcount): Likewise.
14221 (expand_doubleword_parity): Likewise.
14222 (expand_absneg_bit): Use scalar_int_mode for local variables.
14223 (prepare_float_lib_cmp): Likewise.
14224 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
14225 rather than a machine_mode.
14226 (convert_memory_address_addr_space): Likewise.
14227 (get_mode_bounds): Likewise.
14228 (get_address_mode): Return a scalar_int_mode rather than a
14230 * rtlanal.c (get_address_mode): Likewise.
14231 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
14232 than a machine_mode.
14233 * targhooks.c (default_mode_rep_extended): Likewise.
14234 (default_valid_pointer_mode): Likewise.
14235 (default_addr_space_valid_pointer_mode): Likewise.
14236 (default_eh_return_filter_mode): Return a scalar_int_mode rather
14237 than a machine_mode.
14238 (default_libgcc_cmp_return_mode): Likewise.
14239 (default_libgcc_shift_count_mode): Likewise.
14240 (default_unwind_word_mode): Likewise.
14241 (default_addr_space_pointer_mode): Likewise.
14242 (default_addr_space_address_mode): Likewise.
14243 * targhooks.h (default_eh_return_filter_mode): Likewise.
14244 (default_libgcc_cmp_return_mode): Likewise.
14245 (default_libgcc_shift_count_mode): Likewise.
14246 (default_unwind_word_mode): Likewise.
14247 (default_addr_space_pointer_mode): Likewise.
14248 (default_addr_space_address_mode): Likewise.
14249 (default_mode_rep_extended): Take a scalar_int_mode rather than
14251 (default_valid_pointer_mode): Likewise.
14252 (default_addr_space_valid_pointer_mode): Likewise.
14253 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
14255 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
14256 rather than a machine_mode.
14257 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
14258 for local variables.
14259 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14260 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
14261 than a machine_mode.
14263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14264 Alan Hayward <alan.hayward@arm.com>
14265 David Sherwood <david.sherwood@arm.com>
14267 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
14268 the mode argument to scalar_int_mode.
14269 (do_jump_by_parts_zero_rtx): Likewise.
14270 (do_jump_by_parts_equality_rtx): Likewise.
14271 (do_jump_by_parts_greater): Take a mode argument.
14272 (do_jump_by_parts_equality): Likewise.
14273 (do_jump_1): Update calls accordingly.
14275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14276 Alan Hayward <alan.hayward@arm.com>
14277 David Sherwood <david.sherwood@arm.com>
14279 * is-a.h (safe_dyn_cast): New function.
14280 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
14281 (jump_table_for_label): Likewise.
14282 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
14283 instead of an rtx_insn *.
14284 (shorten_branches): Use dyn_cast instead of LABEL_P and
14285 JUMP_TABLE_DATA_P. Use jump_table_for_label and
14286 rtx_jump_table_data::get_data_mode.
14287 (final_scan_insn): Likewise.
14289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14290 Alan Hayward <alan.hayward@arm.com>
14291 David Sherwood <david.sherwood@arm.com>
14293 * combine.c (try_combine): Use is_a <scalar_int_mode> when
14294 trying to combine a full-register integer set with a subreg
14297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14298 Alan Hayward <alan.hayward@arm.com>
14299 David Sherwood <david.sherwood@arm.com>
14301 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
14302 that is always either address_mode or pointer_mode.
14304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14305 Alan Hayward <alan.hayward@arm.com>
14306 David Sherwood <david.sherwood@arm.com>
14308 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
14309 when the two are known to be equal.
14311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14312 Alan Hayward <alan.hayward@arm.com>
14313 David Sherwood <david.sherwood@arm.com>
14315 * simplify-rtx.c (simplify_const_unary_operation): Use
14316 is_a <scalar_int_mode> instead of checking for a nonzero
14317 precision. Forcibly convert op_mode to a scalar_int_mode
14318 in that case. More clearly differentiate the operand and
14319 result modes and use the former when deciding what the value
14320 of a count-bits operation should be. Use is_int_mode instead
14321 of checking for a MODE_INT. Remove redundant check for whether
14322 this mode has a zero precision.
14324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14325 Alan Hayward <alan.hayward@arm.com>
14326 David Sherwood <david.sherwood@arm.com>
14328 * optabs.c (widen_leading): Change the type of the mode argument
14329 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
14330 (widen_bswap): Likewise.
14331 (expand_parity): Likewise.
14332 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
14333 (expand_ffs): Likewise.
14334 (epand_unop): Check for scalar integer modes before calling the
14337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14338 Alan Hayward <alan.hayward@arm.com>
14339 David Sherwood <david.sherwood@arm.com>
14341 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
14343 (expand_expr_real_1): Update call accordingly.
14345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14346 Alan Hayward <alan.hayward@arm.com>
14347 David Sherwood <david.sherwood@arm.com>
14349 * expmed.c (store_bit_field_using_insv): Add op0_mode and
14350 value_mode arguments. Use scalar_int_mode internally.
14351 (store_bit_field_1): Rename the new integer mode from imode
14352 to op0_mode and use it instead of GET_MODE (op0). Update calls
14353 to store_split_bit_field, store_bit_field_using_insv and
14354 store_fixed_bit_field.
14355 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
14356 Use scalar_int_mode internally. Use a bit count rather than a mode
14357 when calculating the largest bit size for get_best_mode.
14358 Update calls to store_split_bit_field and store_fixed_bit_field_1.
14359 (store_fixed_bit_field_1): Add mode and value_mode arguments.
14360 Remove assertion that OP0 has a scalar integer mode.
14361 (store_split_bit_field): Add op0_mode and value_mode arguments.
14362 Update calls to extract_fixed_bit_field.
14363 (extract_bit_field_using_extv): Add an op0_mode argument.
14364 Use scalar_int_mode internally.
14365 (extract_bit_field_1): Rename the new integer mode from imode to
14366 op0_mode and use it instead of GET_MODE (op0). Update calls to
14367 extract_split_bit_field, extract_bit_field_using_extv and
14368 extract_fixed_bit_field.
14369 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
14370 to extract_split_bit_field and extract_fixed_bit_field_1.
14371 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
14372 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
14373 on the target mode.
14374 (extract_split_bit_field): Add an op0_mode argument. Update call
14375 to extract_fixed_bit_field.
14377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14378 Alan Hayward <alan.hayward@arm.com>
14379 David Sherwood <david.sherwood@arm.com>
14381 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
14382 * explow.c (hard_function_value): Likewise.
14383 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
14384 convert_to_mode call outside the loop.
14385 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
14386 for the mode iterator. Require the mode specified by max_pieces
14388 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
14390 (copy_blkmode_to_reg): Likewise.
14391 (set_storage_via_setmem): Likewise.
14392 * optabs.c (prepare_cmp_insn): Likewise.
14393 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14394 * stor-layout.c (finish_bitfield_representative): Likewise.
14396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14397 Alan Hayward <alan.hayward@arm.com>
14398 David Sherwood <david.sherwood@arm.com>
14400 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
14401 * expr.c (convert_move): Use them.
14402 (convert_modes): Likewise.
14403 (store_expr_with_bounds): Likewise.
14405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14406 Alan Hayward <alan.hayward@arm.com>
14407 David Sherwood <david.sherwood@arm.com>
14409 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
14410 parameter for the mode of "x". Remove the "known_x", "known_mode"
14411 and "known_ret" arguments. Change the type of the mode argument
14412 to scalar_int_mode.
14413 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
14414 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
14415 (reg_num_sign_bit_copies_for_combine): Likewise.
14416 * rtlanal.c (nonzero_bits1): Likewise.
14417 (num_sign_bit_copies1): Likewise.
14418 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
14419 (reg_num_sign_bit_copies_general): Likewise.
14420 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
14421 (reg_nonzero_bits_general): Likewise.
14423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14424 Alan Hayward <alan.hayward@arm.com>
14425 David Sherwood <david.sherwood@arm.com>
14427 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
14428 than in subroutines. Return 1 for non-integer modes.
14429 (cached_num_sign_bit_copies): Change the type of the mode parameter
14430 to scalar_int_mode.
14431 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
14432 classes. Handle CONST_INT_P first and then check whether X also
14433 has a scalar integer mode. Check the same thing for inner registers
14434 of a SUBREG and for values that are being extended or truncated.
14436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14437 Alan Hayward <alan.hayward@arm.com>
14438 David Sherwood <david.sherwood@arm.com>
14440 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
14441 in subroutines. Return the mode mask for non-integer modes.
14442 (cached_nonzero_bits): Change the type of the mode parameter
14443 to scalar_int_mode.
14444 (nonzero_bits1): Likewise. Remove early exit for other mode
14445 classes. Handle CONST_INT_P first and then check whether X
14446 also has a scalar integer mode.
14448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14449 Alan Hayward <alan.hayward@arm.com>
14450 David Sherwood <david.sherwood@arm.com>
14452 * expr.c (widest_int_mode_for_size): Make the comment match the code.
14453 Return a scalar_int_mode and assert that the size is greater than
14455 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
14456 (op_by_pieces_d::op_by_pieces_d): Likewise.
14457 (op_by_pieces_d::run): Likewise.
14458 (can_store_by_pieces): Likewise.
14460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14461 Alan Hayward <alan.hayward@arm.com>
14462 David Sherwood <david.sherwood@arm.com>
14464 * combine.c (extract_left_shift): Add a mode argument and update
14466 (make_compound_operation_int): Change the type of the mode parameter
14467 to scalar_int_mode and update the call to extract_left_shift.
14469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14470 Alan Hayward <alan.hayward@arm.com>
14471 David Sherwood <david.sherwood@arm.com>
14473 * combine.c (simplify_and_const_int): Change the type of the mode
14474 parameter to scalar_int_mode.
14475 (simplify_and_const_int_1): Likewise. Update recursive call.
14477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14478 Alan Hayward <alan.hayward@arm.com>
14479 David Sherwood <david.sherwood@arm.com>
14481 * combine.c (simplify_compare_const): Check that the mode is a
14482 scalar_int_mode (rather than VOIDmode) before testing its
14484 (simplify_comparison): Move COMPARISON_P handling out of the
14485 loop and restrict the latter part of the loop to scalar_int_modes.
14486 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
14487 and when considering SUBREG_REGs. Use is_int_mode instead of
14488 checking GET_MODE_CLASS against MODE_INT.
14490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14491 Alan Hayward <alan.hayward@arm.com>
14492 David Sherwood <david.sherwood@arm.com>
14494 * combine.c (try_widen_shift_mode): Move check for equal modes to...
14495 (simplify_shift_const_1): ...here. Use scalar_int_mode for
14496 shift_unit_mode and for modes involved in scalar shifts.
14498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14499 Alan Hayward <alan.hayward@arm.com>
14500 David Sherwood <david.sherwood@arm.com>
14502 * combine.c (force_int_to_mode): New function, split out from...
14503 (force_to_mode): ...here. Keep xmode up-to-date and use it
14504 instead of GET_MODE (x).
14506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14507 Alan Hayward <alan.hayward@arm.com>
14508 David Sherwood <david.sherwood@arm.com>
14510 * optabs-query.h (extraction_insn::struct_mode): Change type to
14511 opt_scalar_int_mode and update comment.
14512 (extraction_insn::field_mode): Change type to scalar_int_mode.
14513 (extraction_insn::pos_mode): Likewise.
14514 * combine.c (make_extraction): Update accordingly.
14515 * optabs-query.c (get_traditional_extraction_insn): Likewise.
14516 (get_optab_extraction_insn): Likewise.
14517 * recog.c (simplify_while_replacing): Likewise.
14518 * expmed.c (narrow_bit_field_mem): Change the type of the mode
14519 parameter to opt_scalar_int_mode.
14521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14522 Alan Hayward <alan.hayward@arm.com>
14523 David Sherwood <david.sherwood@arm.com>
14525 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
14526 to a scalar_int_mode instead of a machine_mode.
14527 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
14528 (get_best_mode): Return a boolean and use a pointer argument to store
14529 the selected mode. Replace the limit mode parameter with a bit limit.
14530 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
14531 for the values returned by bit_field_mode_iterator::next_mode.
14532 (store_bit_field): Update call to get_best_mode.
14533 (store_fixed_bit_field): Likewise.
14534 (extract_fixed_bit_field): Likewise.
14535 * expr.c (optimize_bitfield_assignment_op): Likewise.
14536 * fold-const.c (optimize_bit_field_compare): Likewise.
14537 (fold_truth_andor_1): Likewise.
14538 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
14539 Update for new type of m_mode.
14540 (get_best_mode): As above.
14542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14543 Alan Hayward <alan.hayward@arm.com>
14544 David Sherwood <david.sherwood@arm.com>
14546 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
14547 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
14548 (store_bit_field): Check is_a <scalar_int_mode> before calling
14549 strict_volatile_bitfield_p.
14550 (extract_bit_field): Likewise.
14552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14553 Alan Hayward <alan.hayward@arm.com>
14554 David Sherwood <david.sherwood@arm.com>
14556 * target.def (cstore_mode): Return a scalar_int_mode.
14557 * doc/tm.texi: Regenerate.
14558 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
14559 * targhooks.h (default_cstore_mode): Likewise.
14560 * targhooks.c (default_cstore_mode): Likewise, using a forced
14562 * expmed.c (emit_cstore): Expect the target of the cstore to be
14565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14566 Alan Hayward <alan.hayward@arm.com>
14567 David Sherwood <david.sherwood@arm.com>
14569 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
14571 (niter_desc): Likewise mode.
14572 (iv_analyze): Add a mode parameter.
14574 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
14575 and change its type to scalar_int_mode.
14576 * loop-iv.c: Update commentary at head of file.
14577 (iv_constant): Pass the mode paraeter before the rtx it describes
14578 and change its type to scalar_int_mode. Remove VOIDmode handling.
14579 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
14580 (iv_extend): Likewise.
14581 (shorten_into_mode): Likewise.
14582 (iv_add): Use scalar_int_mode.
14583 (iv_mult): Likewise.
14584 (iv_shift): Likewise.
14585 (canonicalize_iv_subregs): Likewise.
14586 (get_biv_step_1): Pass the outer_mode parameter before the rtx
14587 it describes and change its mode to scalar_int_mode. Also change
14588 the type of the returned inner_mode to scalar_int_mode.
14589 (get_biv_step): Likewise, turning outer_mode from a pointer
14590 into a direct parameter. Update call to get_biv_step_1.
14591 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
14592 iv_constant and get_biv_step.
14593 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
14594 and change its type to scalar_int_mode. Don't initialise iv->mode
14595 to VOIDmode and remove later checks for its still being VOIDmode.
14596 Update calls to iv_analyze_op and iv_analyze_expr. Check
14597 is_a <scalar_int_mode> when changing the mode under consideration.
14598 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
14599 Update call to iv_analyze_expr.
14600 (iv_analyze_op): Add a mode parameter. Reject subregs whose
14601 inner register is not also a scalar_int_mode. Update call to
14603 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
14604 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
14605 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
14606 separate mode class checks. Update calls to iv_analyze. Remove
14607 fix-up of VOIDmodes after iv_analyze_biv.
14608 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
14609 don't have a scalar_int_mode. Update call to biv_p.
14611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14612 Alan Hayward <alan.hayward@arm.com>
14613 David Sherwood <david.sherwood@arm.com>
14615 * cfgexpand.c (convert_debug_memory_address): Use
14616 as_a <scalar_int_mode>.
14617 * combine.c (expand_compound_operation): Likewise.
14618 (make_extraction): Likewise.
14619 (change_zero_ext): Likewise.
14620 (simplify_comparison): Likewise.
14621 * cse.c (cse_insn): Likewise.
14622 * dwarf2out.c (minmax_loc_descriptor): Likewise.
14623 (mem_loc_descriptor): Likewise.
14624 (loc_descriptor): Likewise.
14625 * expmed.c (init_expmed_one_mode): Likewise.
14626 (synth_mult): Likewise.
14627 (emit_store_flag_1): Likewise.
14628 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
14629 of a comparison with size.
14630 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
14631 (reduce_to_bit_field_precision): Likewise.
14632 * function.c (expand_function_end): Likewise.
14633 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
14634 * loop-doloop.c (doloop_modify): Likewise.
14635 * optabs.c (expand_binop): Likewise.
14636 (expand_unop): Likewise.
14637 (expand_copysign_absneg): Likewise.
14638 (prepare_cmp_insn): Likewise.
14639 (maybe_legitimize_operand): Likewise.
14640 * recog.c (const_scalar_int_operand): Likewise.
14641 * rtlanal.c (get_address_mode): Likewise.
14642 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14643 (simplify_cond_clz_ctz): Likewise.
14644 * tree-nested.c (get_nl_goto_field): Likewise.
14645 * tree.c (build_vector_type_for_mode): Likewise.
14646 * var-tracking.c (use_narrower_mode): Likewise.
14648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14649 Alan Hayward <alan.hayward@arm.com>
14650 David Sherwood <david.sherwood@arm.com>
14652 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
14653 * builtins.c (expand_builtin_signbit): Use it.
14654 * cfgexpand.c (expand_debug_expr): Likewise.
14655 * dojump.c (do_jump): Likewise.
14656 (do_compare_and_jump): Likewise.
14657 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
14658 * expmed.c (make_tree): Likewise.
14659 * expr.c (expand_expr_real_2): Likewise.
14660 (expand_expr_real_1): Likewise.
14661 (try_casesi): Likewise.
14662 * fold-const-call.c (fold_const_call_ss): Likewise.
14663 * fold-const.c (unextend): Likewise.
14664 (extract_muldiv_1): Likewise.
14665 (fold_single_bit_test): Likewise.
14666 (native_encode_int): Likewise.
14667 (native_encode_string): Likewise.
14668 (native_interpret_int): Likewise.
14669 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
14670 * internal-fn.c (expand_addsub_overflow): Likewise.
14671 (expand_neg_overflow): Likewise.
14672 (expand_mul_overflow): Likewise.
14673 (expand_arith_overflow): Likewise.
14674 * match.pd: Likewise.
14675 * stor-layout.c (layout_type): Likewise.
14676 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14677 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
14678 * tree-ssanames.c (get_range_info): Likewise.
14679 * tree-switch-conversion.c (array_value_type) Likewise.
14680 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
14681 (vect_recog_divmod_pattern): Likewise.
14682 (vect_recog_mixed_size_cond_pattern): Likewise.
14683 * tree-vrp.c (extract_range_basic): Likewise.
14684 (simplify_float_conversion_using_ranges): Likewise.
14685 * tree.c (int_fits_type_p): Likewise.
14686 * ubsan.c (instrument_bool_enum_load): Likewise.
14687 * varasm.c (mergeable_string_section): Likewise.
14688 (narrowing_initializer_constant_valid_p): Likewise.
14689 (output_constant): Likewise.
14691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14692 Alan Hayward <alan.hayward@arm.com>
14693 David Sherwood <david.sherwood@arm.com>
14695 * machmode.h (NARROWEST_INT_MODE): New macro.
14696 * expr.c (alignment_for_piecewise_move): Use it instead of
14697 GET_CLASS_NARROWEST_MODE (MODE_INT).
14698 (push_block): Likewise.
14699 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
14701 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14704 Alan Hayward <alan.hayward@arm.com>
14705 David Sherwood <david.sherwood@arm.com>
14707 * postreload.c (move2add_valid_value_p): Change the type of the
14708 mode parameter to scalar_int_mode.
14709 (move2add_use_add2_insn): Add a mode parameter and use it instead
14711 (move2add_use_add3_insn): Likewise.
14712 (reload_cse_move2add): Update accordingly.
14714 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14715 Alan Hayward <alan.hayward@arm.com>
14716 David Sherwood <david.sherwood@arm.com>
14718 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
14720 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
14721 * optabs.c (expand_unop): Likewise.
14723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14724 Alan Hayward <alan.hayward@arm.com>
14725 David Sherwood <david.sherwood@arm.com>
14727 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
14728 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
14729 (popcount_loc_descriptor): Likewise.
14730 (bswap_loc_descriptor): Likewise.
14731 (rotate_loc_descriptor): Likewise.
14732 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
14733 calling the functions above.
14735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14736 Alan Hayward <alan.hayward@arm.com>
14737 David Sherwood <david.sherwood@arm.com>
14739 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
14741 (try_combine): Likewise.
14742 (simplify_if_then_else): Likewise.
14743 * cse.c (cse_insn): Likewise.
14744 * dwarf2out.c (mem_loc_descriptor): Likewise.
14745 * emit-rtl.c (gen_lowpart_common): Likewise.
14746 * simplify-rtx.c (simplify_truncation): Likewise.
14747 (simplify_binary_operation_1): Likewise.
14748 (simplify_const_relational_operation): Likewise.
14749 (simplify_ternary_operation): Likewise.
14750 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
14752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14753 Alan Hayward <alan.hayward@arm.com>
14754 David Sherwood <david.sherwood@arm.com>
14756 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
14757 * reload.c (push_reload): Likewise.
14758 (find_reloads): Likewise.
14760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14761 Alan Hayward <alan.hayward@arm.com>
14762 David Sherwood <david.sherwood@arm.com>
14764 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
14765 (make_compound_operation_int): Likewise.
14766 (change_zero_ext): Likewise.
14767 * expr.c (convert_move): Likewise.
14768 (convert_modes): Likewise.
14769 * fwprop.c (forward_propagate_subreg): Likewise.
14770 * loop-iv.c (get_biv_step_1): Likewise.
14771 * optabs.c (widen_operand): Likewise.
14772 * postreload.c (move2add_valid_value_p): Likewise.
14773 * recog.c (simplify_while_replacing): Likewise.
14774 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14775 (simplify_binary_operation_1): Likewise. Remove redundant
14776 mode equality check.
14778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14779 Alan Hayward <alan.hayward@arm.com>
14780 David Sherwood <david.sherwood@arm.com>
14782 * combine.c (combine_simplify_rtx): Add checks for
14783 is_a <scalar_int_mode>.
14784 (simplify_if_then_else): Likewise.
14785 (make_field_assignment): Likewise.
14786 (simplify_comparison): Likewise.
14787 * ifcvt.c (noce_try_bitop): Likewise.
14788 * loop-invariant.c (canonicalize_address_mult): Likewise.
14789 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14792 Alan Hayward <alan.hayward@arm.com>
14793 David Sherwood <david.sherwood@arm.com>
14795 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
14796 is_a <scalar_int_mode> instead of != BLKmode.
14798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14799 Alan Hayward <alan.hayward@arm.com>
14800 David Sherwood <david.sherwood@arm.com>
14802 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
14803 instead of != VOIDmode.
14804 * combine.c (if_then_else_cond): Likewise.
14805 (change_zero_ext): Likewise.
14806 * dwarf2out.c (mem_loc_descriptor): Likewise.
14807 (loc_descriptor): Likewise.
14808 * rtlanal.c (canonicalize_condition): Likewise.
14809 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
14811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14812 Alan Hayward <alan.hayward@arm.com>
14813 David Sherwood <david.sherwood@arm.com>
14815 * simplify-rtx.c (simplify_binary_operation_1): Use
14816 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
14818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14819 Alan Hayward <alan.hayward@arm.com>
14820 David Sherwood <david.sherwood@arm.com>
14822 * wide-int.h (int_traits<unsigned char>) New class.
14823 (int_traits<unsigned short>) Likewise.
14824 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
14825 Use GET_MODE_UNIT_PRECISION and remove redundant test for
14827 * combine.c (set_nonzero_bits_and_sign_copies): Use
14828 is_a <scalar_int_mode>.
14829 (find_split_point): Likewise.
14830 (combine_simplify_rtx): Likewise.
14831 (simplify_logical): Likewise.
14832 (expand_compound_operation): Likewise.
14833 (expand_field_assignment): Likewise.
14834 (make_compound_operation): Likewise.
14835 (extended_count): Likewise.
14836 (change_zero_ext): Likewise.
14837 (simplify_comparison): Likewise.
14838 * dwarf2out.c (scompare_loc_descriptor): Likewise.
14839 (ucompare_loc_descriptor): Likewise.
14840 (minmax_loc_descriptor): Likewise.
14841 (mem_loc_descriptor): Likewise.
14842 (loc_descriptor): Likewise.
14843 * expmed.c (init_expmed_one_mode): Likewise.
14844 * lra-constraints.c (lra_constraint_offset): Likewise.
14845 * optabs.c (prepare_libcall_arg): Likewise.
14846 * postreload.c (move2add_note_store): Likewise.
14847 * reload.c (operands_match_p): Likewise.
14848 * rtl.h (load_extend_op): Likewise.
14849 * rtlhooks.c (gen_lowpart_general): Likewise.
14850 * simplify-rtx.c (simplify_truncation): Likewise.
14851 (simplify_unary_operation_1): Likewise.
14852 (simplify_binary_operation_1): Likewise.
14853 (simplify_const_binary_operation): Likewise.
14854 (simplify_const_relational_operation): Likewise.
14855 (simplify_subreg): Likewise.
14856 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14857 * var-tracking.c (adjust_mems): Likewise.
14858 (prepare_call_arguments): Likewise.
14860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14861 Alan Hayward <alan.hayward@arm.com>
14862 David Sherwood <david.sherwood@arm.com>
14864 * machmode.h (is_int_mode): New fuction.
14865 * combine.c (find_split_point): Use it.
14866 (combine_simplify_rtx): Likewise.
14867 (simplify_if_then_else): Likewise.
14868 (simplify_set): Likewise.
14869 (simplify_shift_const_1): Likewise.
14870 (simplify_comparison): Likewise.
14871 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
14872 * cse.c (notreg_cost): Likewise.
14873 (cse_insn): Likewise.
14874 * cselib.c (cselib_lookup_1): Likewise.
14875 * dojump.c (do_jump_1): Likewise.
14876 (do_compare_rtx_and_jump): Likewise.
14877 * dse.c (get_call_args): Likewise.
14878 * dwarf2out.c (rtl_for_decl_init): Likewise.
14879 (native_encode_initializer): Likewise.
14880 * expmed.c (emit_store_flag_1): Likewise.
14881 (emit_store_flag): Likewise.
14882 * expr.c (convert_modes): Likewise.
14883 (store_field): Likewise.
14884 (expand_expr_real_1): Likewise.
14885 * fold-const.c (fold_read_from_constant_string): Likewise.
14886 * gimple-ssa-sprintf.c (get_format_string): Likewise.
14887 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
14888 * optabs.c (expand_binop): Likewise.
14889 (expand_unop): Likewise.
14890 (expand_abs_nojump): Likewise.
14891 (expand_one_cmpl_abs_nojump): Likewise.
14892 * simplify-rtx.c (mode_signbit_p): Likewise.
14893 (val_signbit_p): Likewise.
14894 (val_signbit_known_set_p): Likewise.
14895 (val_signbit_known_clear_p): Likewise.
14896 (simplify_relational_operation_1): Likewise.
14897 * tree.c (vector_type_mode): Likewise.
14899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14900 Alan Hayward <alan.hayward@arm.com>
14901 David Sherwood <david.sherwood@arm.com>
14903 * machmode.h (smallest_mode_for_size): Fix formatting.
14904 (smallest_int_mode_for_size): New function.
14905 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
14906 instead of smallest_mode_for_size.
14907 * combine.c (make_extraction): Likewise.
14908 * config/arc/arc.c (arc_expand_movmem): Likewise.
14909 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
14910 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14911 * config/s390/s390.c (s390_expand_insv): Likewise.
14912 * config/sparc/sparc.c (assign_int_registers): Likewise.
14913 * config/spu/spu.c (spu_function_value): Likewise.
14914 (spu_function_arg): Likewise.
14915 * coverage.c (get_gcov_type): Likewise.
14916 (get_gcov_unsigned_t): Likewise.
14917 * dse.c (find_shift_sequence): Likewise.
14918 * expmed.c (store_bit_field_1): Likewise.
14919 * expr.c (convert_move): Likewise.
14920 (store_field): Likewise.
14921 * internal-fn.c (expand_arith_overflow): Likewise.
14922 * optabs-query.c (get_best_extraction_insn): Likewise.
14923 * optabs.c (expand_twoval_binop_libfunc): Likewise.
14924 * stor-layout.c (layout_type): Likewise.
14925 (initialize_sizetypes): Likewise.
14926 * targhooks.c (default_get_mask_mode): Likewise.
14927 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
14929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14930 Alan Hayward <alan.hayward@arm.com>
14931 David Sherwood <david.sherwood@arm.com>
14933 * machmode.h (opt_mode::else_blk): New function.
14934 (int_mode_for_mode): Declare.
14935 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
14936 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
14938 * cfgexpand.c (expand_debug_expr): Likewise.
14939 * combine.c (gen_lowpart_or_truncate): Likewise.
14940 (gen_lowpart_for_combine): Likewise.
14941 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
14942 * config/avr/avr.c (avr_to_int_mode): Likewise.
14943 (avr_out_plus_1): Likewise.
14944 (avr_out_plus): Likewise.
14945 (avr_out_round): Likewise.
14946 * config/i386/i386.c (ix86_split_to_parts): Likewise.
14947 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14948 (s390_expand_vcond): Likewise.
14949 * config/spu/spu.c (spu_split_immediate): Likewise.
14950 (spu_expand_mov): Likewise.
14951 * dse.c (get_stored_val): Likewise.
14952 * expmed.c (store_bit_field_1): Likewise.
14953 (convert_extracted_bit_field): Use int_mode_for_mode instead of
14955 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
14956 (extract_low_bits): Likewise.
14957 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
14958 handling rather than repeating the check.
14959 (emit_group_store): Likewise.
14960 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
14961 * optabs.c (expand_absneg_bit): Likewise.
14962 (expand_copysign_absneg): Likewise.
14963 (expand_copysign_bit): Likewise.
14964 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
14965 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
14966 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
14967 * var-tracking.c (prepare_call_arguments): Likewise.
14968 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
14969 int_mode_for_mode instead of mode_for_size.
14970 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14973 Alan Hayward <alan.hayward@arm.com>
14974 David Sherwood <david.sherwood@arm.com>
14976 * machmode.h (int_mode_for_size): New function.
14977 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
14978 instead of mode_for_size.
14979 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
14981 * combine.c (expand_field_assignment): Use int_mode_for_size
14982 instead of mode_for_size.
14983 (make_extraction): Likewise.
14984 (simplify_shift_const_1): Likewise.
14985 (simplify_comparison): Likewise.
14986 * dojump.c (do_jump): Likewise.
14987 * dwarf2out.c (mem_loc_descriptor): Likewise.
14988 * emit-rtl.c (init_derived_machine_modes): Likewise.
14989 * expmed.c (flip_storage_order): Likewise.
14990 (convert_extracted_bit_field): Likewise.
14991 * expr.c (copy_blkmode_from_reg): Likewise.
14992 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
14993 * internal-fn.c (expand_mul_overflow): Likewise.
14994 * lower-subreg.c (simple_move): Likewise.
14995 * optabs-libfuncs.c (init_optabs): Likewise.
14996 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14997 * tree.c (vector_type_mode): Likewise.
14998 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
14999 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15000 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15001 * tree-vect-stmts.c (vectorizable_load): Likewise.
15002 (vectorizable_store): Likewise.
15004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15005 Alan Hayward <alan.hayward@arm.com>
15006 David Sherwood <david.sherwood@arm.com>
15008 * coretypes.h (pod_mode): New type.
15009 (scalar_int_mode_pod): New typedef.
15010 * machmode.h (pod_mode): New class.
15011 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15012 * genmodes.c (emit_mode_int_n): Update accordingly.
15013 * lower-subreg.h (target_lower_subreg): Change type to
15014 scalar_int_mode_pod.
15015 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15016 scalar_int_mode_pod.
15018 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15019 Alan Hayward <alan.hayward@arm.com>
15020 David Sherwood <david.sherwood@arm.com>
15022 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
15023 machine_mode to scalar_int_mode.
15024 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
15025 (rs6000_option_override_internal): Remove cast to int.
15026 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
15027 machine_mode to scalar_int_mode.
15028 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
15029 (rs6000_option_override_internal): Remove cast to int.
15030 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
15031 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
15033 * config/s390/s390.c (s390_expand_builtin): Likewise.
15034 * coretypes.h (scalar_int_mode): New type.
15035 (opt_scalar_int_mode): New typedef.
15036 * machmode.h (scalar_int_mode): New class.
15037 (scalar_int_mode::includes_p): New function.
15038 (byte_mode): Change type to scalar_int_mode.
15039 (word_mode): Likewise.
15040 (ptr_mode): Likewise.
15041 * emit-rtl.c (byte_mode): Likewise.
15042 (word_mode): Likewise.
15043 (ptr_mode): Likewise.
15044 (init_derived_machine_modes): Update accordingly.
15045 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
15046 and MODE_PARTIAL_INT.
15047 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
15048 opt_scalar_int_mode.
15050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15051 Alan Hayward <alan.hayward@arm.com>
15052 David Sherwood <david.sherwood@arm.com>
15054 * target.def (libgcc_floating_mode_supported_p): Take a
15056 * doc/tm.texi: Regenerate.
15057 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
15059 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15060 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
15063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15064 Alan Hayward <alan.hayward@arm.com>
15065 David Sherwood <david.sherwood@arm.com>
15067 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
15068 * doc/tm.texi: Regenerate.
15069 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
15070 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
15071 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
15072 * targhooks.h (default_floatn_mode): Likewise.
15073 * targhooks.c (default_floatn_mode): Likewise.
15074 * tree.c (build_common_tree_nodes): Update accordingly.
15076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15077 Alan Hayward <alan.hayward@arm.com>
15078 David Sherwood <david.sherwood@arm.com>
15080 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
15081 (mode_iterator::iterate_p): Likewise.
15082 (mode_iterator::get_wider): Likewise.
15083 * expr.c (init_expr_target): Use opt_scalar_float_mode.
15085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15086 Alan Hayward <alan.hayward@arm.com>
15087 David Sherwood <david.sherwood@arm.com>
15089 * coretypes.h (opt_scalar_float_mode): New typedef.
15090 * machmode.h (float_mode_for_size): New function.
15091 * emit-rtl.c (double_mode): Delete.
15092 (init_emit_once): Use float_mode_for_size.
15093 * stor-layout.c (layout_type): Likewise.
15094 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
15096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15097 Alan Hayward <alan.hayward@arm.com>
15098 David Sherwood <david.sherwood@arm.com>
15100 * output.h (assemble_real): Take a scalar_float_mode.
15101 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
15102 * config/arm/arm.md (consttable_4): Likewise.
15103 (consttable_8): Likewise.
15104 (consttable_16): Likewise.
15105 * config/mips/mips.md (consttable_float): Likewise.
15106 * config/s390/s390.c (s390_output_pool_entry): Likewise.
15107 * varasm.c (assemble_real): Take a scalar_float_mode.
15108 (output_constant_pool_2): Update accordingly.
15109 (output_constant): Likewise.
15111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15112 Alan Hayward <alan.hayward@arm.com>
15113 David Sherwood <david.sherwood@arm.com>
15115 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
15116 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
15117 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
15118 (native_encode_real): Likewise.
15119 (native_interpret_real): Likewise.
15120 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
15121 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15124 Alan Hayward <alan.hayward@arm.com>
15125 David Sherwood <david.sherwood@arm.com>
15127 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
15128 <scalar_float_mode>. Simplify.
15129 (gen_extend_conv_libfunc): Likewise.
15131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15132 Alan Hayward <alan.hayward@arm.com>
15133 David Sherwood <david.sherwood@arm.com>
15135 * coretypes.h (scalar_float_mode): New type.
15136 * machmode.h (mode_traits::from_int): Use machine_mode if
15137 USE_ENUM_MODES is defined.
15138 (is_a): New function.
15140 (dyn_cast): Likewise.
15141 (scalar_float_mode): New class.
15142 (scalar_float_mode::includes_p): New function.
15143 (is_float_mode): Likewise.
15144 * gdbhooks.py (MachineModePrinter): New class.
15145 (build_pretty_printer): Use it for scalar_float_mode.
15146 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
15147 (format_helper::format_helper): Turn into a template.
15148 * genmodes.c (get_mode_class): New function.
15149 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
15150 or machine_mode if none.
15151 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
15152 as_a <scalar_float_mode>.
15153 * dwarf2out.c (mem_loc_descriptor): Likewise.
15154 (insert_float): Likewise.
15155 (add_const_value_attribute): Likewise.
15156 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15157 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
15158 (expand_unop): Update accordingly.
15159 (expand_abs_nojump): Likewise.
15160 (expand_copysign_absneg): Take a scalar_float_mode.
15161 (expand_copysign_bit): Likewise.
15162 (expand_copysign): Update accordingly.
15164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15165 Alan Hayward <alan.hayward@arm.com>
15166 David Sherwood <david.sherwood@arm.com>
15168 * coretypes.h (opt_mode): New class.
15169 * machmode.h (opt_mode): Likewise.
15170 (opt_mode::else_void): New function.
15171 (opt_mode::require): Likewise.
15172 (opt_mode::exists): Likewise.
15173 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
15174 (GET_MODE_2XWIDER_MODE): Likewise.
15175 (mode_iterator::get_wider): Update accordingly.
15176 (mode_iterator::get_2xwider): Likewise.
15177 (mode_iterator::get_known_wider): Likewise, turning into a template.
15178 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
15179 forcing a wider mode to exist.
15180 * config/cr16/cr16.h (LONG_REG_P): Likewise.
15181 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15182 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
15183 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
15184 * lower-subreg.c (init_lower_subreg): Likewise.
15185 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
15186 on the final iteration.
15187 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
15188 a wider mode exists before asking for a move pattern.
15189 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
15190 forcing a wider mode to exist.
15191 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
15192 returning false if no such mode exists.
15193 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
15194 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
15195 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
15196 Avoid checking for a MODE_INT if we already know the mode is not a
15198 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
15199 forcing a wider mode to exist.
15200 (expmed_mult_highpart_optab): Likewise.
15201 (expmed_mult_highpart): Likewise.
15202 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
15204 * lto-streamer-in.c (lto_input_mode_table): Likewise.
15205 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
15206 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15207 * internal-fn.c (expand_mul_overflow): Update use of
15208 GET_MODE_2XWIDER_MODE.
15209 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15210 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
15211 GET_MODE_WIDER_MODE.
15212 (convert_plusminus_to_widen): Likewise.
15213 * tree-switch-conversion.c (array_value_type): Likewise.
15214 * var-tracking.c (emit_note_insn_var_location): Likewise.
15215 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15216 Return false inside rather than outside the loop if no wider mode
15218 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
15219 and GET_MODE_2XWIDER_MODE
15220 (can_compare_p): Use else_void.
15221 * gdbhooks.py (OptMachineModePrinter): New class.
15222 (build_pretty_printer): Use it for opt_mode.
15224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15225 Alan Hayward <alan.hayward@arm.com>
15226 David Sherwood <david.sherwood@arm.com>
15228 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
15229 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
15231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15232 Alan Hayward <alan.hayward@arm.com>
15233 David Sherwood <david.sherwood@arm.com>
15235 * machmode.h (mode_traits): New structure.
15236 (get_narrowest_mode): New function.
15237 (mode_iterator::start): Likewise.
15238 (mode_iterator::iterate_p): Likewise.
15239 (mode_iterator::get_wider): Likewise.
15240 (mode_iterator::get_known_wider): Likewise.
15241 (mode_iterator::get_2xwider): Likewise.
15242 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
15243 (FOR_EACH_MODE): Likewise.
15244 (FOR_EACH_MODE_FROM): Likewise.
15245 (FOR_EACH_MODE_UNTIL): Likewise.
15246 (FOR_EACH_WIDER_MODE): Likewise.
15247 (FOR_EACH_2XWIDER_MODE): Likewise.
15248 * builtins.c (expand_builtin_strlen): Use new mode iterators.
15249 * combine.c (simplify_comparison): Likewise
15250 * config/i386/i386.c (type_natural_mode): Likewise.
15251 * cse.c (cse_insn): Likewise.
15252 * dse.c (find_shift_sequence): Likewise.
15253 * emit-rtl.c (init_derived_machine_modes): Likewise.
15254 (init_emit_once): Likewise.
15255 * explow.c (hard_function_value): Likewise.
15256 * expmed.c (extract_fixed_bit_field_1): Likewise.
15257 (extract_bit_field_1): Likewise.
15258 (expand_divmod): Likewise.
15259 (emit_store_flag_1): Likewise.
15260 * expr.c (init_expr_target): Likewise.
15261 (convert_move): Likewise.
15262 (alignment_for_piecewise_move): Likewise.
15263 (widest_int_mode_for_size): Likewise.
15264 (emit_block_move_via_movmem): Likewise.
15265 (copy_blkmode_to_reg): Likewise.
15266 (set_storage_via_setmem): Likewise.
15267 (compress_float_constant): Likewise.
15268 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15269 * optabs-query.c (get_best_extraction_insn): Likewise.
15270 * optabs.c (expand_binop): Likewise.
15271 (expand_twoval_unop): Likewise.
15272 (expand_twoval_binop): Likewise.
15273 (widen_leading): Likewise.
15274 (widen_bswap): Likewise.
15275 (expand_parity): Likewise.
15276 (expand_unop): Likewise.
15277 (prepare_cmp_insn): Likewise.
15278 (prepare_float_lib_cmp): Likewise.
15279 (expand_float): Likewise.
15280 (expand_fix): Likewise.
15281 (expand_sfix_optab): Likewise.
15282 * postreload.c (move2add_use_add2_insn): Likewise.
15283 * reg-stack.c (reg_to_stack): Likewise.
15284 * reginfo.c (choose_hard_reg_mode): Likewise.
15285 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15286 * stor-layout.c (mode_for_size): Likewise.
15287 (smallest_mode_for_size): Likewise.
15288 (mode_for_vector): Likewise.
15289 (finish_bitfield_representative): Likewise.
15290 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
15291 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
15292 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15293 * var-tracking.c (prepare_call_arguments): Likewise.
15295 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15296 Alan Hayward <alan.hayward@arm.com>
15297 David Sherwood <david.sherwood@arm.com>
15299 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
15300 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
15301 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
15302 * machmode.h (mode_size): Move earlier in file.
15303 (mode_precision): Likewise.
15304 (mode_inner): Likewise.
15305 (mode_nunits): Likewise.
15306 (mode_unit_size): Likewise.
15307 (unit_unit_precision): Likewise.
15308 (mode_wider): Likewise.
15309 (mode_2xwider): Likewise.
15310 (machine_mode): New class.
15311 (mode_to_bytes): New function.
15312 (mode_to_bits): Likewise.
15313 (mode_to_precision): Likewise.
15314 (mode_to_inner): Likewise.
15315 (mode_to_unit_size): Likewise.
15316 (mode_to_unit_precision): Likewise.
15317 (mode_to_nunits): Likewise.
15318 (GET_MODE_SIZE): Use mode_to_bytes.
15319 (GET_MODE_BITSIZE): Use mode_to_bits.
15320 (GET_MODE_PRECISION): Use mode_to_precision.
15321 (GET_MODE_INNER): Use mode_to_inner.
15322 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
15323 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
15324 (GET_MODE_NUNITS): Use mode_to_nunits.
15325 * system.h (ALWAYS_INLINE): New macro.
15326 * config/powerpcspe/powerpcspe-c.c
15327 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
15328 int for arg1_mode and arg2_mode.
15330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15331 Alan Hayward <alan.hayward@arm.com>
15332 David Sherwood <david.sherwood@arm.com>
15334 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15335 Prefix mode names with E_ in case statements.
15336 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15337 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
15338 (aarch64_split_simd_move): Likewise.
15339 (aarch64_gen_storewb_pair): Likewise.
15340 (aarch64_gen_loadwb_pair): Likewise.
15341 (aarch64_gen_store_pair): Likewise.
15342 (aarch64_gen_load_pair): Likewise.
15343 (aarch64_get_condition_code_1): Likewise.
15344 (aarch64_constant_pool_reload_icode): Likewise.
15345 (get_rsqrte_type): Likewise.
15346 (get_rsqrts_type): Likewise.
15347 (get_recpe_type): Likewise.
15348 (get_recps_type): Likewise.
15349 (aarch64_gimplify_va_arg_expr): Likewise.
15350 (aarch64_simd_container_mode): Likewise.
15351 (aarch64_emit_load_exclusive): Likewise.
15352 (aarch64_emit_store_exclusive): Likewise.
15353 (aarch64_expand_compare_and_swap): Likewise.
15354 (aarch64_gen_atomic_cas): Likewise.
15355 (aarch64_emit_bic): Likewise.
15356 (aarch64_emit_atomic_swap): Likewise.
15357 (aarch64_emit_atomic_load_op): Likewise.
15358 (aarch64_evpc_trn): Likewise.
15359 (aarch64_evpc_uzp): Likewise.
15360 (aarch64_evpc_zip): Likewise.
15361 (aarch64_evpc_ext): Likewise.
15362 (aarch64_evpc_rev): Likewise.
15363 (aarch64_evpc_dup): Likewise.
15364 (aarch64_gen_ccmp_first): Likewise.
15365 (aarch64_gen_ccmp_next): Likewise.
15366 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15367 (alpha_emit_xfloating_libcall): Likewise.
15368 (emit_insxl): Likewise.
15369 (alpha_arg_type): Likewise.
15370 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
15371 (arc_preferred_simd_mode): Likewise.
15372 (arc_secondary_reload): Likewise.
15373 (get_arc_condition_code): Likewise.
15374 (arc_print_operand): Likewise.
15375 (arc_legitimate_constant_p): Likewise.
15376 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15377 * config/arc/arc.md (casesi_load): Likewise.
15378 (casesi_compact_jump): Likewise.
15379 * config/arc/predicates.md (proper_comparison_operator): Likewise.
15380 (cc_use_register): Likewise.
15381 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15382 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15383 (arm_init_iwmmxt_builtins): Likewise.
15384 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
15385 (neon_expand_vector_init): Likewise.
15386 (arm_attr_length_move_neon): Likewise.
15387 (maybe_get_arm_condition_code): Likewise.
15388 (arm_emit_vector_const): Likewise.
15389 (arm_preferred_simd_mode): Likewise.
15390 (arm_output_iwmmxt_tinsr): Likewise.
15391 (thumb1_output_casesi): Likewise.
15392 (thumb2_output_casesi): Likewise.
15393 (arm_emit_load_exclusive): Likewise.
15394 (arm_emit_store_exclusive): Likewise.
15395 (arm_expand_compare_and_swap): Likewise.
15396 (arm_evpc_neon_vuzp): Likewise.
15397 (arm_evpc_neon_vzip): Likewise.
15398 (arm_evpc_neon_vrev): Likewise.
15399 (arm_evpc_neon_vtrn): Likewise.
15400 (arm_evpc_neon_vext): Likewise.
15401 (arm_validize_comparison): Likewise.
15402 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
15403 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
15404 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
15405 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
15406 (c6x_preferred_simd_mode): Likewise.
15407 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
15408 (epiphany_rtx_costs): Likewise.
15409 * config/epiphany/predicates.md (proper_comparison_operator):
15411 * config/frv/frv.c (condexec_memory_operand): Likewise.
15412 (frv_emit_move): Likewise.
15413 (output_move_single): Likewise.
15414 (output_condmove_single): Likewise.
15415 (frv_hard_regno_mode_ok): Likewise.
15416 (frv_matching_accg_mode): Likewise.
15417 * config/h8300/h8300.c (split_adds_subs): Likewise.
15418 (h8300_rtx_costs): Likewise.
15419 (h8300_print_operand): Likewise.
15420 (compute_mov_length): Likewise.
15421 (output_logical_op): Likewise.
15422 (compute_logical_op_length): Likewise.
15423 (compute_logical_op_cc): Likewise.
15424 (h8300_shift_needs_scratch_p): Likewise.
15425 (output_a_shift): Likewise.
15426 (compute_a_shift_length): Likewise.
15427 (compute_a_shift_cc): Likewise.
15428 (expand_a_rotate): Likewise.
15429 (output_a_rotate): Likewise.
15430 * config/i386/i386.c (classify_argument): Likewise.
15431 (function_arg_advance_32): Likewise.
15432 (function_arg_32): Likewise.
15433 (function_arg_64): Likewise.
15434 (function_value_64): Likewise.
15435 (ix86_gimplify_va_arg): Likewise.
15436 (ix86_legitimate_constant_p): Likewise.
15437 (put_condition_code): Likewise.
15438 (split_double_mode): Likewise.
15439 (ix86_avx256_split_vector_move_misalign): Likewise.
15440 (ix86_expand_vector_logical_operator): Likewise.
15441 (ix86_split_idivmod): Likewise.
15442 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
15443 (ix86_build_const_vector): Likewise.
15444 (ix86_build_signbit_mask): Likewise.
15445 (ix86_match_ccmode): Likewise.
15446 (ix86_cc_modes_compatible): Likewise.
15447 (ix86_expand_branch): Likewise.
15448 (ix86_expand_sse_cmp): Likewise.
15449 (ix86_expand_sse_movcc): Likewise.
15450 (ix86_expand_int_sse_cmp): Likewise.
15451 (ix86_expand_vec_perm_vpermi2): Likewise.
15452 (ix86_expand_vec_perm): Likewise.
15453 (ix86_expand_sse_unpack): Likewise.
15454 (ix86_expand_int_addcc): Likewise.
15455 (ix86_split_to_parts): Likewise.
15456 (ix86_vectorize_builtin_gather): Likewise.
15457 (ix86_vectorize_builtin_scatter): Likewise.
15458 (avx_vpermilp_parallel): Likewise.
15459 (inline_memory_move_cost): Likewise.
15460 (ix86_tieable_integer_mode_p): Likewise.
15461 (x86_maybe_negate_const_int): Likewise.
15462 (ix86_expand_vector_init_duplicate): Likewise.
15463 (ix86_expand_vector_init_one_nonzero): Likewise.
15464 (ix86_expand_vector_init_one_var): Likewise.
15465 (ix86_expand_vector_init_concat): Likewise.
15466 (ix86_expand_vector_init_interleave): Likewise.
15467 (ix86_expand_vector_init_general): Likewise.
15468 (ix86_expand_vector_set): Likewise.
15469 (ix86_expand_vector_extract): Likewise.
15470 (emit_reduc_half): Likewise.
15471 (ix86_emit_i387_round): Likewise.
15472 (ix86_mangle_type): Likewise.
15473 (ix86_expand_round_sse4): Likewise.
15474 (expand_vec_perm_blend): Likewise.
15475 (canonicalize_vector_int_perm): Likewise.
15476 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
15477 (expand_vec_perm_1): Likewise.
15478 (expand_vec_perm_interleave3): Likewise.
15479 (expand_vec_perm_even_odd_pack): Likewise.
15480 (expand_vec_perm_even_odd_1): Likewise.
15481 (expand_vec_perm_broadcast_1): Likewise.
15482 (ix86_vectorize_vec_perm_const_ok): Likewise.
15483 (ix86_expand_vecop_qihi): Likewise.
15484 (ix86_expand_mul_widen_hilo): Likewise.
15485 (ix86_expand_sse2_abs): Likewise.
15486 (ix86_expand_pextr): Likewise.
15487 (ix86_expand_pinsr): Likewise.
15488 (ix86_preferred_simd_mode): Likewise.
15489 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
15490 * config/i386/sse.md (*andnot<mode>3): Likewise.
15491 (<mask_codefor><code><mode>3<mask_name>): Likewise.
15492 (*<code><mode>3): Likewise.
15493 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
15494 (ia64_expand_atomic_op): Likewise.
15495 (ia64_arg_type): Likewise.
15496 (ia64_mode_to_int): Likewise.
15497 (ia64_scalar_mode_supported_p): Likewise.
15498 (ia64_vector_mode_supported_p): Likewise.
15499 (expand_vec_perm_broadcast): Likewise.
15500 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
15501 (iq2000_function_arg_advance): Likewise.
15502 (iq2000_function_arg): Likewise.
15503 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
15504 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
15505 (m68k_libcall_value): Likewise.
15506 (m68k_function_value): Likewise.
15507 (sched_attr_op_type): Likewise.
15508 * config/mcore/mcore.c (mcore_output_move): Likewise.
15509 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
15511 (microblaze_function_arg): Likewise.
15512 * config/mips/mips.c (mips16_build_call_stub): Likewise.
15513 (mips_print_operand): Likewise.
15514 (mips_mode_ok_for_mov_fmt_p): Likewise.
15515 (mips_vector_mode_supported_p): Likewise.
15516 (mips_preferred_simd_mode): Likewise.
15517 (mips_expand_vpc_loongson_even_odd): Likewise.
15518 (mips_expand_vec_unpack): Likewise.
15519 (mips_expand_vi_broadcast): Likewise.
15520 (mips_expand_vector_init): Likewise.
15521 (mips_expand_vec_reduc): Likewise.
15522 (mips_expand_msa_cmp): Likewise.
15523 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
15524 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
15525 (cc_flags_for_mode): Likewise.
15526 * config/msp430/msp430.c (msp430_print_operand): Likewise.
15527 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
15528 (nds32_output_casesi_pc_relative): Likewise.
15529 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15530 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
15531 (nvptx_gen_unpack): Likewise.
15532 (nvptx_gen_pack): Likewise.
15533 (nvptx_gen_shuffle): Likewise.
15534 (nvptx_gen_wcast): Likewise.
15535 (nvptx_preferred_simd_mode): Likewise.
15536 * config/pa/pa.c (pa_secondary_reload): Likewise.
15537 * config/pa/predicates.md (base14_operand): Likewise.
15538 * config/powerpcspe/powerpcspe-c.c
15539 (altivec_resolve_overloaded_builtin): Likewise.
15540 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
15542 (rs6000_preferred_simd_mode): Likewise.
15543 (output_vec_const_move): Likewise.
15544 (rs6000_expand_vector_extract): Likewise.
15545 (rs6000_split_vec_extract_var): Likewise.
15546 (reg_offset_addressing_ok_p): Likewise.
15547 (rs6000_legitimate_offset_address_p): Likewise.
15548 (rs6000_legitimize_address): Likewise.
15549 (rs6000_emit_set_const): Likewise.
15550 (rs6000_const_vec): Likewise.
15551 (rs6000_emit_move): Likewise.
15552 (spe_build_register_parallel): Likewise.
15553 (rs6000_darwin64_record_arg_recurse): Likewise.
15554 (swap_selector_for_mode): Likewise.
15555 (spe_init_builtins): Likewise.
15556 (paired_init_builtins): Likewise.
15557 (altivec_init_builtins): Likewise.
15558 (do_load_for_compare): Likewise.
15559 (rs6000_generate_compare): Likewise.
15560 (rs6000_expand_float128_convert): Likewise.
15561 (emit_load_locked): Likewise.
15562 (emit_store_conditional): Likewise.
15563 (rs6000_output_function_epilogue): Likewise.
15564 (rs6000_handle_altivec_attribute): Likewise.
15565 (rs6000_function_value): Likewise.
15566 (emit_fusion_gpr_load): Likewise.
15567 (emit_fusion_p9_load): Likewise.
15568 (emit_fusion_p9_store): Likewise.
15569 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
15570 (fusion_gpr_mem_load): Likewise.
15571 (fusion_addis_mem_combo_load): Likewise.
15572 (fusion_addis_mem_combo_store): Likewise.
15573 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
15574 (fusion_gpr_mem_load): Likewise.
15575 (fusion_addis_mem_combo_load): Likewise.
15576 (fusion_addis_mem_combo_store): Likewise.
15577 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15579 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
15580 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15581 (rs6000_preferred_simd_mode): Likewise.
15582 (output_vec_const_move): Likewise.
15583 (rs6000_expand_vector_extract): Likewise.
15584 (rs6000_split_vec_extract_var): Likewise.
15585 (reg_offset_addressing_ok_p): Likewise.
15586 (rs6000_legitimate_offset_address_p): Likewise.
15587 (rs6000_legitimize_address): Likewise.
15588 (rs6000_emit_set_const): Likewise.
15589 (rs6000_const_vec): Likewise.
15590 (rs6000_emit_move): Likewise.
15591 (rs6000_darwin64_record_arg_recurse): Likewise.
15592 (swap_selector_for_mode): Likewise.
15593 (paired_init_builtins): Likewise.
15594 (altivec_init_builtins): Likewise.
15595 (rs6000_expand_float128_convert): Likewise.
15596 (emit_load_locked): Likewise.
15597 (emit_store_conditional): Likewise.
15598 (rs6000_output_function_epilogue): Likewise.
15599 (rs6000_handle_altivec_attribute): Likewise.
15600 (rs6000_function_value): Likewise.
15601 (emit_fusion_gpr_load): Likewise.
15602 (emit_fusion_p9_load): Likewise.
15603 (emit_fusion_p9_store): Likewise.
15604 * config/rx/rx.c (rx_gen_move_template): Likewise.
15605 (flags_from_mode): Likewise.
15606 * config/s390/predicates.md (s390_alc_comparison): Likewise.
15607 (s390_slb_comparison): Likewise.
15608 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
15609 (s390_vector_mode_supported_p): Likewise.
15610 (s390_cc_modes_compatible): Likewise.
15611 (s390_match_ccmode_set): Likewise.
15612 (s390_canonicalize_comparison): Likewise.
15613 (s390_emit_compare_and_swap): Likewise.
15614 (s390_branch_condition_mask): Likewise.
15615 (s390_rtx_costs): Likewise.
15616 (s390_secondary_reload): Likewise.
15617 (__SECONDARY_RELOAD_CASE): Likewise.
15618 (s390_expand_cs): Likewise.
15619 (s390_preferred_simd_mode): Likewise.
15620 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
15621 * config/sh/sh.c (sh_print_operand): Likewise.
15622 (dump_table): Likewise.
15623 (sh_secondary_reload): Likewise.
15624 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15625 * config/sh/sh.md (casesi_worker_1): Likewise.
15626 (casesi_worker_2): Likewise.
15627 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
15628 (fcc_comparison_operator): Likewise.
15629 * config/sparc/sparc.c (sparc_expand_move): Likewise.
15630 (emit_soft_tfmode_cvt): Likewise.
15631 (sparc_preferred_simd_mode): Likewise.
15632 (output_cbranch): Likewise.
15633 (sparc_print_operand): Likewise.
15634 (sparc_expand_vec_perm_bmask): Likewise.
15635 (vector_init_bshuffle): Likewise.
15636 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15637 (spu_vector_mode_supported_p): Likewise.
15638 (spu_expand_insv): Likewise.
15639 (spu_emit_branch_or_set): Likewise.
15640 (spu_handle_vector_attribute): Likewise.
15641 (spu_builtin_splats): Likewise.
15642 (spu_builtin_extract): Likewise.
15643 (spu_builtin_promote): Likewise.
15644 (spu_expand_sign_extend): Likewise.
15645 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15646 (tilegx_simd_int): Likewise.
15647 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
15648 (tilepro_simd_int): Likewise.
15649 * config/v850/v850.c (const_double_split): Likewise.
15650 (v850_print_operand): Likewise.
15651 (ep_memory_offset): Likewise.
15652 * config/vax/vax.c (vax_rtx_costs): Likewise.
15653 (vax_output_int_move): Likewise.
15654 (vax_output_int_add): Likewise.
15655 (vax_output_int_subtract): Likewise.
15656 * config/visium/predicates.md (visium_branch_operator): Likewise.
15657 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
15658 (visium_print_operand_address): Likewise.
15659 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15660 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
15661 (xtensa_expand_conditional_branch): Likewise.
15662 (xtensa_copy_incoming_a7): Likewise.
15663 (xtensa_output_literal): Likewise.
15664 * dfp.c (decimal_real_maxval): Likewise.
15665 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15668 Alan Hayward <alan.hayward@arm.com>
15669 David Sherwood <david.sherwood@arm.com>
15671 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
15672 (mode_nunits_inline): Likewise.
15673 (mode_inner_inline): Likewise.
15674 (mode_unit_size_inline): Likewise.
15675 (mode_unit_precision_inline): Likewise.
15676 (emit_insn_modes_h): Likewise. Also emit a #define of the
15678 (emit_mode_wider): Add an E_ prefix to mode names.
15679 (emit_mode_complex): Likewise.
15680 (emit_mode_inner): Likewise.
15681 (emit_mode_adjustments): Likewise.
15682 (emit_mode_int_n): Likewise.
15683 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
15684 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
15685 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
15686 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
15687 (CRC32_BUILTIN, ENTRY): Likewise.
15688 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
15689 (aarch64_pop_regs): Likewise.
15690 (aarch64_process_components): Likewise.
15691 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
15692 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
15693 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
15694 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
15695 * config/arm/arm.c (arm_init_libfuncs): Likewise.
15696 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
15698 * config/i386/i386-builtin.def (pcmpestr): Likewise.
15699 (pcmpistr): Likewise.
15700 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
15701 * config/mmix/mmix.c (mmix_output_condition): Likewise.
15702 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15704 * config/rl78/rl78.c (mduc_regs): Likewise.
15705 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15706 (htm_expand_builtin): Likewise.
15707 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
15708 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
15709 * config/xtensa/xtensa.c (print_operand): Likewise.
15710 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
15711 (NUM_MODE_VECTOR_INT): Likewise.
15712 * genoutput.c (null_operand): Likewise.
15713 (output_operand_data): Likewise.
15714 * genrecog.c (print_parameter_value): Likewise.
15715 * lra.c (debug_operand_data): Likewise.
15717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15719 * dwarf2out.c (scompare_loc_descriptor_wide)
15720 (scompare_loc_descriptor_narrow): New functions, split out from...
15721 (scompare_loc_descriptor): ...here.
15722 * expmed.c (emit_store_flag_int): New function, split out from...
15723 (emit_store_flag): ...here.
15725 2017-08-30 Richard Biener <rguenther@suse.de>
15727 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
15728 (dwarf2out_early_finish): Move setting of AT_pubnames from
15729 early debug output to early finish.
15731 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15733 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
15734 and -mdata-region to the assembler.
15736 2017-08-30 Richard Biener <rguenther@suse.de>
15738 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
15740 (gen_subprogram_die): Add DW_AT_object_pointer only early.
15741 (dwarf2out_early_global_decl): Only generate a DIE for the
15742 abstract origin if it doesn't already exist or is a declaration DIE.
15743 (resolve_addr): Do not add the linkage name twice when
15744 generating a stub DIE for the DW_TAG_GNU_call_site target.
15746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15748 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15749 Use machine_mode rather than int for arg1_mode.
15751 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
15754 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
15755 that the second argument of the built-in functions to unpack
15756 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
15757 switch statement instead a lot of if statements.
15758 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
15759 Allow 64-bit values to be in Altivec registers as well as
15760 traditional floating point registers.
15761 (pack<mode>, FMOVE128_VSX iterator): Likewise.
15763 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
15765 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
15766 MAX_REGS_PER_ADDRESS == 1.
15768 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
15770 * config/i386/i386.opt (flag_fentry): Do not init to -1.
15771 * config/i386/i386.c (ix86_option_override_internal): Simplify
15772 setting of opts->x_flag_entry.
15774 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15775 Jakub Jelinek <jakub@redhat.com>
15776 Richard Biener <rguenther@suse.de>
15778 PR tree-optimization/81503
15779 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
15780 folded constant fits in the target type; reorder tests for clarity.
15782 2017-08-29 Martin Liska <mliska@suse.cz>
15784 * passes.def: Include pass_lower_switch.
15785 * stmt.c (dump_case_nodes): Remove and move to
15786 tree-switch-conversion.
15787 (case_values_threshold): Likewise.
15788 (expand_switch_as_decision_tree_p): Likewise.
15789 (emit_case_decision_tree): Likewise.
15790 (expand_case): Likewise.
15791 (balance_case_nodes): Likewise.
15792 (node_has_low_bound): Likewise.
15793 (node_has_high_bound): Likewise.
15794 (node_is_bounded): Likewise.
15795 (emit_case_nodes): Likewise.
15796 (struct simple_case_node): New struct.
15797 (add_case_node): Remove.
15798 (emit_case_dispatch_table): Use vector instead of case_list.
15799 (reset_out_edges_aux): Remove.
15800 (compute_cases_per_edge): Likewise.
15801 (expand_case): Build list of simple_case_node.
15802 (expand_sjlj_dispatch_table): Use it.
15803 * tree-switch-conversion.c (struct case_node): Moved from
15804 stmt.c and adjusted.
15805 (emit_case_nodes): Likewise.
15806 (node_has_low_bound): Likewise.
15807 (node_has_high_bound): Likewise.
15808 (node_is_bounded): Likewise.
15809 (case_values_threshold): Likewise.
15810 (reset_out_edges_aux): Likewise.
15811 (compute_cases_per_edge): Likewise.
15812 (add_case_node): Likewise.
15813 (dump_case_nodes): Likewise.
15814 (balance_case_nodes): Likewise.
15815 (expand_switch_as_decision_tree_p): Likewise.
15816 (emit_jump): Likewise.
15817 (emit_case_decision_tree): Likewise.
15818 (try_switch_expansion): Likewise.
15819 (do_jump_if_equal): Likewise.
15820 (emit_cmp_and_jump_insns): Likewise.
15821 (fix_phi_operands_for_edge): New function.
15822 (record_phi_operand_mapping): Likewise.
15823 (class pass_lower_switch): New pass.
15824 (pass_lower_switch::execute): New function.
15825 (make_pass_lower_switch): Likewise.
15826 (conditional_probability):
15827 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
15828 * tree-pass.h: Add make_pass_lower_switch.
15830 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15833 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
15836 2017-08-29 Richard Biener <rguenther@suse.de>
15838 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
15839 we do not add a DW_AT_inline attribute twice.
15840 (gen_subprogram_die): Remove code setting DW_AT_inline on
15841 DECL_ABSTRACT_P nodes.
15843 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
15845 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
15846 calls to internal functions.
15847 (gimplify_modify_expr): Likewise.
15848 * tree-call-cdce.c (use_internal_fn): Likewise.
15849 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
15850 (convert_to_divmod): Set the nothrow flag.
15851 * tree-if-conv.c (predicate_mem_writes): Likewise.
15852 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15853 (vectorizable_call): Likewise.
15854 (vectorizable_store): Likewise.
15855 (vectorizable_load): Likewise.
15856 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
15857 (vect_recog_mask_conversion_pattern): Likewise.
15859 2017-08-29 Martin Liska <mliska@suse.cz>
15862 * gcc.c (driver_handle_option): Add new argument.
15863 * opts-common.c (handle_option): Pass
15864 target_option_override_hook.
15865 * opts-global.c (lang_handle_option): Add new option.
15866 (set_default_handlers): Add new argument.
15867 (decode_options): Likewise.
15868 * opts.c (target_handle_option): Likewise.
15869 (common_handle_option): Call target_option_override_hook.
15870 * opts.h (struct cl_option_handler_func): Add hook for
15871 target option override.
15872 (struct cl_option_handlers): Likewise.
15873 (set_default_handlers): Add new argument.
15874 (decode_options): Likewise.
15875 (common_handle_option): Likewise.
15876 (target_handle_option): Likewise.
15877 * toplev.c (toplev::main): Pass targetm.target_option.override
15880 2017-08-29 Richard Biener <rguenther@suse.de>
15881 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
15883 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
15884 life to the active subtree.
15886 2017-08-28 Jeff Law <law@redhat.com>
15888 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
15889 derive_equivalences.
15890 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
15891 Code moved into....
15892 (edge_info::derive_equivalences): New private member function
15894 * tree-ssa-dom.c (class edge_info): Changed from a struct
15895 to a class. Add ctor/dtor, methods and data members.
15896 (edge_info::edge_info): Renamed from allocate_edge_info.
15897 Initialize additional members.
15898 (edge_info::~edge_info): New.
15899 (free_dom_edge_info): Delete the edge info.
15900 (record_edge_info): Use new class & associated member functions.
15901 Tighten forms for testing for edge equivalences.
15902 (record_temporary_equivalences): Iterate over the simple
15903 equivalences rather than assuming there's only one per edge.
15904 (cprop_into_successor_phis): Iterate over the simple
15905 equivalences rather than assuming there's only one per edge.
15906 (optimize_stmt): Use operand_equal_p rather than pointer
15907 equality for mini-DSE code.
15909 2017-08-28 Nathan Sidwell <nathan@acm.org>
15911 * gcc.c (execute): Fold SIGPIPE handling into switch
15912 statement. Adjust internal error message.
15914 2017-08-28 Richard Biener <rguenther@suse.de>
15917 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
15918 Do nothing for removed DIEs.
15920 2017-08-28 Richard Biener <rguenther@suse.de>
15922 PR tree-optimization/81977
15923 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
15926 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
15929 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
15930 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
15931 using targetm.gen_mem_thread_fence.
15933 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
15936 * config/i386/i386.md (*<btsc><mode>): Change operand 2
15937 predicate to register_operand. Reorder operands.
15938 (*btr<mode>): Ditto.
15939 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
15940 (*btr<mode>_mask): Ditto.
15942 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
15944 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
15945 * config/rs6000/xmmintrin.h: New file.
15946 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
15948 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15951 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
15952 parameter and_insn and return it.
15953 (recombine_lvx_pattern): Insert a copy to ensure availability of
15954 the base register of the copied masking operation at the point of
15955 the instruction replacement.
15956 (recombine_stvx_pattern): Likewise.
15958 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
15960 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
15961 undocumented switches.
15962 (-mpower9-dform-vector): Likewise.
15963 (-mpower9-dform): Likewise.
15964 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
15965 comments to delete references to -mpower9-dform* switches.
15966 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
15967 Delete reference to -mpower9-dform* switches, test for
15968 -mpower9-vector instead.
15969 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
15970 (OTHER_P9_VECTOR_MASKS): Likewise.
15971 (POWERPC_MASKS): Likewise.
15972 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
15973 tests against -mpower9-dform* to -mpower9-vector. Delete code
15974 that checked for -mpower9-dform* consistancy with other options.
15975 Add test for -mpower9-misc to enable other power9 switches.
15976 (rs6000_init_hard_regno_mode_ok): Likewise.
15977 (rs6000_option_override_internal): Likewise.
15978 (rs6000_emit_prologue): Likewise.
15979 (rs6000_emit_epilogue): Likewise.
15980 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
15981 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
15982 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
15984 (emit_fusion_p9_store): Likewise.
15985 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
15986 resetting these macros if the assembler does not support ISA 3.0
15988 (TARGET_P9_DFORM_VECTOR): Likewise.
15989 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
15990 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
15992 2017-08-25 Alan Modra <amodra@gmail.com>
15994 PR rtl-optimization/81747
15995 * cse.c (cse_extended_basic_block): Don't attempt to record
15996 equivalences for degenerate conditional branches that branch
15997 to their fall-through.
15999 2017-08-24 Martin Sebor <msebor@redhat.com>
16001 PR middle-end/81908
16002 * gimple-fold.c (size_must_be_zero_p): New function.
16003 (gimple_fold_builtin_memory_op): Call it.
16005 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16007 * config/rs6000/mm_malloc.h: New file.
16009 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16011 PR tree-optimization/81913
16012 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16013 analysis when either IVs in condition can wrap.
16015 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16017 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16018 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
16020 2017-08-24 Richard Biener <rguenther@suse.de>
16023 * targhooks.c (default_target_can_inline_p): Properly
16024 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
16025 is present and always compare.
16026 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
16027 infer -mfpmath=sse from TARGET_SSE_P.
16028 (ix86_can_inline_p): Properly use target_option_default_node when
16029 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
16031 2017-08-24 Richard Biener <rguenther@suse.de>
16034 * dwarf2out.c (output_die): Handle flag_generate_offload like
16036 (output_comp_unit): Likewise.
16037 (gen_array_type_die): Likewise.
16038 (dwarf2out_early_finish): Likewise.
16039 (note_variable_value_in_expr): Likewise.
16040 (dwarf2out_finish): Likewise. Adjust assert.
16041 * cgraphunit.c (symbol_table::compile): Move setting of
16042 flag_generate_offload earlier ...
16043 (symbol_table::finalize_compilation_unit): ... here, before
16044 early debug finalization.
16046 2017-08-24 Richard Biener <rguenther@suse.de>
16048 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
16049 and ipa-fnsummary.h.
16050 (ix86_can_inline_p): When ix86_fpmath flags do not match
16051 check whether the callee uses FP math at all.
16053 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
16055 PR middle-end/81931
16056 * tree-ssanames.c (get_nonzero_bits): Use element_precision
16057 instead of TYPE_PRECISION.
16059 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
16060 Alan Hayward <alan.hayward@arm.com>
16061 David Sherwood <david.sherwood@arm.com>
16063 * combine.c (make_extraction): Use subreg_offset_from_lsb.
16065 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
16067 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
16069 (ix86_frame::stack_realign_allocate): New field.
16070 (struct machine_frame_state): Modify comments.
16071 (machine_frame_state::sp_realigned_fp_end): New field.
16072 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
16073 layout calculation.
16074 (sp_valid_at): Add assertion to assure no attempt to access invalid
16075 offset of a realigned stack.
16076 (fp_valid_at): Likewise.
16077 (choose_baseaddr): Modify comments.
16078 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
16079 ix86_expand_prologue.
16080 (ix86_expand_prologue): Modify stack realignment and allocation.
16081 (ix86_expand_epilogue): Modify comments.
16082 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
16083 avx2_runtime, avx512f, and avx512f_runtime.
16085 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
16087 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
16088 (mstackrealign): Do not init to -1.
16089 * config/i386/i386.c (ix86_option_override_internal):
16090 Check opts_set, not opts when setting default value of
16091 opts->x_ix86_force_align_arg_pointer.
16093 2017-08-23 Richard Biener <rguenther@suse.de>
16095 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
16096 lang_hooks.decl_printable_name.
16097 * print-rtl-function.c (print_rtx_function): Likewise.
16098 * tree-pretty-print.c (dump_function_header): Likewise.
16100 2017-08-23 Richard Biener <rguenther@suse.de>
16103 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
16104 -g0 at compile-time.
16106 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16108 PR middle-end/19706
16109 * doc/sourcebuild.texi (Other hardware attributes):
16112 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16114 PR middle-end/19706
16115 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
16116 Fix single-use check.
16118 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
16120 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
16122 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16124 * doc/install.texi: Modify to add more details on running selected
16127 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16129 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
16130 is combined with -mabi=ms.
16131 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
16134 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16136 PR tree-optimization/81488
16137 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
16138 and cached_basis fields.
16139 (MAX_SPREAD): New constant.
16140 (alloc_cand_and_find_basis): Initialize new fields.
16141 (clear_visited): New function.
16142 (create_phi_basis_1): Rename from create_phi_basis, set visited
16143 and cached_basis fields.
16144 (create_phi_basis): New wrapper function.
16145 (phi_add_costs_1): Rename from phi_add_costs, add spread
16146 parameter, set visited field, short-circuit when limits reached.
16147 (phi_add_costs): New wrapper function.
16148 (record_phi_increments_1): Rename from record_phi_increments, set
16150 (record_phi_increments): New wrapper function.
16151 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
16152 (phi_incr_cost): New wrapper function.
16153 (all_phi_incrs_profitable_1): Rename from
16154 all_phi_incrs_profitable, set visited field.
16155 (all_phi_incrs_profitable): New wrapper function.
16157 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16158 Alan Hayward <alan.hayward@arm.com>
16159 David Sherwood <david.sherwood@arm.com>
16161 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
16162 that takes the outer and inner modes.
16163 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
16164 comparison as the canonical test for a paradoxical subreg.
16165 * combine.c (simplify_set): Use paradoxical_subreg_p.
16166 (make_extraction): Likewise.
16167 (force_to_mode): Likewise.
16168 (rtx_equal_for_field_assignment_p): Likewise.
16169 (gen_lowpart_for_combine): Likewise.
16170 (simplify_comparison): Likewise.
16171 * cse.c (equiv_constant): Likewise.
16172 * expmed.c (store_bit_field_1): Likewise.
16173 * final.c (alter_subreg): Likewise.
16174 * fwprop.c (propagate_rtx): Likewise.
16175 (forward_propagate_subreg): Likewise.
16176 * ira-conflicts.c (ira_build_conflicts): Likewise.
16177 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
16178 * lra-constraints.c (curr_insn_transform): Likewise.
16179 (split_reg): Likewise.
16180 * lra-eliminations.c (move_plus_up): Likewise.
16181 (lra_eliminate_regs_1): Likewise.
16182 * recog.c (general_operand): Likewise.
16183 * ree.c (combine_reaching_defs): Likewise.
16184 * reload.c (push_reload): Likewise.
16185 (find_reloads): Likewise.
16186 * reload1.c (elimination_effects): Likewise.
16187 (compute_reload_subreg_offset): Likewise.
16188 (choose_reload_regs): Likewise.
16189 * rtlanal.c (subreg_lsb_1): Likewise.
16190 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16191 (simplify_subreg): Likewise.
16192 * var-tracking.c (track_loc_p): Likewise.
16193 * emit-rtl.c (byte_lowpart_offset): Likewise.
16194 (paradoxical_subreg_p): Delete out-of-line definition.
16196 2017-08-22 Jeff Law <law@redhat.com>
16198 PR tree-optimization/81741
16199 PR tree-optimization/71947
16200 * tree-ssa-dom.c: Include tree-inline.h.
16201 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
16202 equivalences if one is more expensive to compute than the other.
16203 * tree-ssa-scopedtables.h (class const_or_copies): Make
16204 record_const_or_copy_raw method private.
16205 (class avail_exprs_stack): New method simplify_binary_operation.
16206 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
16207 avail_exprs_stack::simplify_binary_operation as needed.
16208 (avail_exprs_stack::simplify_binary_operation): New function.
16210 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
16212 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
16213 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
16214 (DOT_SYMBOLS): Likewise.
16215 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
16216 (RELOCATABLE_NEEDS_FIXUP): Likewise.
16217 (RS6000_ABI_NAME): Likewise.
16218 (TARGET_CMODEL): Likewise.
16219 (TOC_SECTION_ASM_OP): Likewise.
16220 (SET_CMODEL): New macro.
16221 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
16223 2017-08-22 Richard Biener <rguenther@suse.de>
16225 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
16226 to allow for free-lang-data replacements similar to verify_type_variant.
16228 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16229 Alan Hayward <alan.hayward@arm.com>
16230 David Sherwood <david.sherwood@arm.com>
16232 * config/aarch64/aarch64.md (casesi): Use DImode rather than
16233 VOIDmode for the LABEL_REF.
16235 2017-08-22 Richard Biener <rguenther@suse.de>
16237 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
16239 2017-08-22 Richard Biener <rguenther@suse.de>
16241 * common.opt (feliminate-dwarf2-dups): Ignore.
16242 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
16243 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
16244 same_die_p_wrap, compute_section_prefix,
16245 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
16246 (comdat_symbol_id, comdat_symbol_number): Likewise.
16247 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
16249 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
16250 (output_die): Mark unreachable path unreachable.
16251 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
16252 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
16253 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
16254 (dwarf2out_early_finish): Likewise.
16256 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
16258 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
16260 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
16263 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
16264 not VAR_P. Filter attribute warnings with OPT_Wattributes.
16265 (avr_attribute_table) <io, io_low, address>: Initialize
16266 .decl_required with true.
16268 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
16270 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
16271 undocumented debugging options.
16272 (-mvsx-scalar-double): Likewise.
16273 (-mallow-df-permute): Likewise.
16274 (-mvectorize-builtins): Likewise.
16275 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16276 (rs6000_builtin_vectorized_function): Likewise.
16277 (rs6000_builtin_md_vectorized_function): Likewise.
16278 (rs6000_opt_vars): Likewise.
16280 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
16283 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
16284 (*btrq_imm): Rename from *btrq.
16285 (*btcq_imm): Rename from *btcq.
16286 (btsc): New code attribute.
16287 (*<btsc><mode>): New insn pattern.
16288 (*btr<mode>): Ditto.
16289 (*<btsc><mode>_mask): New insn_and_split pattern.
16290 (*btr<mode>_mask): Ditto.
16292 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16293 Alan Hayward <alan.hayward@arm.com>
16294 David Sherwood <david.sherwood@arm.com>
16296 * function.c (pad_below): Simplify padding calculation.
16298 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16299 Alan Hayward <alan.hayward@arm.com>
16300 David Sherwood <david.sherwood@arm.com>
16302 * target.def (function_prologue): Remove frame size argument.
16303 (function_epilogue): Likewise.
16304 * doc/tm.texi: Regenerate.
16305 * final.c (final_start_function): Update call to function_prologue.
16306 (final_end_function): Update call to function_epilogue.
16307 (default_function_pro_epilogue): Remove frame size argument.
16308 * output.h (default_function_pro_epilogue): Likewise.
16309 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
16310 (arm_output_function_prologue): Likewise.
16311 * config/frv/frv.c (frv_function_prologue): Likewise.
16312 (frv_function_epilogue): Likewise.
16313 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
16314 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
16315 (ia64_output_function_epilogue): Likewise.
16316 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
16317 (m32r_output_function_epilogue): Likewise.
16318 * config/microblaze/microblaze.c (microblaze_function_prologue)
16319 (microblaze_function_epilogue): Likewise.
16320 * config/mips/mips.c (mips_output_function_prologue): Likewise.
16321 (mips_output_function_epilogue): Likewise.
16322 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
16323 (mmix_target_asm_function_epilogue): Likewise.
16324 * config/msp430/msp430.c (msp430_start_function): Likewise.
16325 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
16326 (nds32_asm_function_epilogue): Likewise.
16327 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
16328 * config/pa/pa.c (pa_output_function_prologue): Likewise.
16329 (pa_output_function_epilogue): Likewise.
16330 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
16331 (rs6000_output_function_epilogue): Likewise.
16332 * config/rl78/rl78.c (rl78_start_function): Likewise.
16333 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
16334 (rs6000_output_function_epilogue): Likewise.
16335 * config/rx/rx.c (rx_output_function_prologue): Likewise.
16336 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
16337 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
16338 (sparc_asm_function_epilogue): Likewise.
16340 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16342 * tree.h (type_has_mode_precision_p): New function.
16343 * convert.c (convert_to_integer_1): Use it.
16344 * expr.c (expand_expr_real_2): Likewise.
16345 (expand_expr_real_1): Likewise.
16346 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
16347 * match.pd: Likewise.
16348 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
16349 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
16350 * tree-tailcall.c (process_assignment): Likewise.
16351 * tree-vect-loop.c (vectorizable_reduction): Likewise.
16352 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
16353 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
16354 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16355 (vectorizable_assignment): Likewise.
16356 (vectorizable_shift): Likewise.
16357 (vectorizable_operation): Likewise.
16358 * tree-vrp.c (register_edge_assert_for_2): Likewise.
16360 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
16362 * match.pd: Add pow (C, x) simplification.
16364 2017-08-21 Richard Biener <rguenther@suse.de>
16366 PR tree-optimization/81900
16367 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
16368 for blocks with abnormal predecessors.
16369 (compute_antic): Do not set visited flag prematurely.
16371 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
16374 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
16376 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16378 * stor-layout.h (vector_type_mode): Move to...
16379 * tree.h (vector_type_mode): ...here.
16380 * stor-layout.c (vector_type_mode): Move to...
16381 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
16383 2017-08-21 Richard Biener <rguenther@suse.de>
16385 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
16386 register_external_die hooks.
16387 (debug_false_tree_charstarstar_uhwistar): Declare.
16388 (debug_nothing_tree_charstar_uhwi): Likewise.
16389 * debug.c (do_nothing_debug_hooks): Adjust.
16390 (debug_false_tree_charstarstar_uhwistar): New do nothing.
16391 (debug_nothing_tree_charstar_uhwi): Likewise.
16392 * dbxout.c (dbx_debug_hooks): Adjust.
16393 (xcoff_debug_hooks): Likewise.
16394 * sdbout.c (sdb_debug_hooks): Likewise.
16395 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16396 * dwarf2out.c (macinfo_label_base): New global.
16397 (dwarf2out_register_external_die): New function for the
16398 register_external_die hook.
16399 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
16400 (dwarf2_debug_hooks): Use them.
16401 (dwarf2_lineno_debug_hooks): Adjust.
16402 (struct die_struct): Add with_offset flag.
16403 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
16404 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
16405 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
16406 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
16407 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
16408 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
16409 defining section names for the early LTO debug variants.
16410 (reset_indirect_string): New helper.
16411 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
16412 (print_dw_val): Add support for offsetted symbol references.
16413 (get_ultimate_context): Split out from is_cxx.
16414 (is_cxx): Use get_ultimate_context.
16415 (is_fortran): Add decl overload.
16416 (compute_comp_unit_symbol): Split out worker from
16417 compute_section_prefix.
16418 (compute_section_prefix): Call compute_comp_unit_symbol and
16419 set comdat_type_p here.
16420 (output_die): Skip DIE symbol output for the LTO added one.
16421 Handle DIE symbol references with offset.
16422 (output_comp_unit): Guard section name mangling properly.
16423 For LTO debug sections emit a symbol at the section beginning
16424 which we use to refer to its DIEs.
16425 (add_abstract_origin_attribute): For DIEs registered via
16426 dwarf2out_register_external_die directly refer to the early
16427 DIE rather than indirectly through the shadow one we created.
16428 Remove obsolete call to dwarf2out_abstract_function for
16429 non-function/block origins.
16430 (gen_array_type_die): When generating early LTO debug do
16431 not emit DW_AT_string_length.
16432 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
16433 late when in LTO. As suggested place a gcc_unreachable for
16434 the DECL_ABSTRACT_P case.
16435 (gen_subprogram_die): Avoid another specification DIE
16436 for early built declarations/definitions for the late LTO case.
16437 (gen_variable_die): Add type references for late duplicated VLA dies
16439 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
16440 we have the abstract instance already.
16441 (process_scope_var): Adjust decl DIE contexts in LTO which
16442 first puts them in limbo.
16443 (gen_decl_die): Do not generate type DIEs late apart from
16444 types for VLAs or for decls we do not yet have a DIE. Do not
16445 call dwarf2out_abstract_function late.
16446 (dwarf2out_early_global_decl): Make sure to create DIEs
16447 for abstract instances of a decl first.
16448 (dwarf2out_late_global_decl): Adjust comment.
16449 (output_macinfo_op): With multiple macro sections use
16450 macinfo_label_base to distinguish labels.
16451 (output_macinfo): Likewise. Update macinfo_label_base.
16452 Pass in the line info label.
16453 (note_variable_value_in_expr): When generating LTO resolve
16454 all variable values here by generating DIEs as needed.
16455 (init_sections_and_labels): Add early LTO debug flag parameter
16456 and generate different sections and names if set. Add generation
16457 counter for the labels so we can have multiple of them.
16458 (reset_dies): Helper to allow DIEs to be output multiple times.
16459 (dwarf2out_finish): When outputting DIEs to the fat part of an
16460 LTO object first reset DIEs.
16461 (dwarf2out_early_finish): Output early DIEs when generating LTO.
16462 (modified_type_die): Check for decl_ultimate_origin being self
16464 (gen_type_die_with_usage): Likewise.
16465 (gen_typedef_die): Allow decl_ultimate_origin being self.
16466 (set_decl_abstract_flags): Remove.
16467 (set_block_abstract_flags): Likewise.
16468 (dwarf2out_abstract_function): Treat the early generated DIEs
16469 as the abstract copy and only add DW_AT_inline and
16470 DW_AT_artificial here and call set_decl_origin_self.
16471 If the DIE has an abstract origin don't do anything.
16472 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
16473 if we have none yet (Go fails to build one, PR78628).
16474 (variably_modified_type_p): Prevent endless recursion for Ada
16475 cyclic pointer types.
16476 * lto-streamer-in.c: Include debug.h.
16477 (dref_queue): New global.
16478 (lto_read_tree_1): Stream in DIE references.
16479 (lto_input_tree): Register DIE references.
16480 (input_function): Stream DECL_DEBUG_ARGS.
16481 * lto-streamer-out.c: Include debug.h.
16482 (lto_write_tree_1): Output DIE references.
16483 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
16484 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
16485 (output_function): Stream DECL_DEBUG_ARGS.
16486 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
16487 Stream DECL_ABSTRACT_ORIGIN.
16488 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
16489 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
16490 DECL_CONTEXT for file-scope decls.
16491 * lto-streamer.h (struct dref_entry): Declare.
16492 (dref_queue): Likewise.
16493 * cfgexpand.c (pass_expand::execute): Do not call the
16494 outlining_inline_function hook here.
16495 * lto-wrapper.c (debug_obj): New global.
16496 (tool_cleanup): Unlink it if required.
16497 (debug_objcopy): New function.
16498 (run_gcc): Handle early debug sections in the IL files by
16499 extracting them to separate files, partially linkin them and
16500 feeding the result back as result to the linker.
16501 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
16502 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
16503 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
16504 sections into a separate segment.
16505 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
16507 (darwin_asm_dwarf_section): Likewise.
16508 (darwin_asm_output_dwarf_offset): Likewise.
16509 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
16511 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16512 Alan Hayward <alan.hayward@arm.com>
16513 David Sherwood <david.sherwood@arm.com>
16515 * read-md.h (md_reader::record_potential_iterator_use): Replace
16516 pointer argument with an rtx and an index.
16517 * read-rtl.c (iterator_group::apply_iterator): Likewise.
16518 (apply_mode_iterator): Likewise.
16519 (apply_code_iterator): Likewise.
16520 (apply_int_iterator): Likewise.
16521 (apply_subst_iterator): Likewise.
16522 (record_iterator_use): Likewise.
16523 (record_attribute_use): Likewise.
16524 (md_reader::record_potential_iterator_use): Likewise. Update calls
16525 to record_iterator_use and apply_iterator.
16526 (iterator_use): Replace ptr with x and index.
16527 (attribute_use): Likewise.
16528 (apply_attribute_uses): Update calls to apply_iterator.
16529 (apply_iterators): Likewise. Update initialization of iterator_use.
16530 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
16531 and record_potential_iterator_use.
16532 (rtx_reader::read_rtx_operand): Likewise.
16534 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16535 Alan Hayward <alan.hayward@arm.com>
16536 David Sherwood <david.sherwood@arm.com>
16538 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
16541 2017-08-21 Richard Biener <rguenther@suse.de>
16543 PR middle-end/81884
16544 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
16545 at struct end conservatively when comparing common bases.
16547 2017-08-21 Richard Biener <rguenther@suse.de>
16549 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
16550 (mem_ref_in_stmt): Remove.
16551 (determine_max_movement): Use ref index to get at the reference.
16552 (invariantness_dom_walker::before_dom_children): Deal with
16553 lim data already initialized.
16554 (gather_mem_refs_stmt): Initialize lim data and record ref index.
16556 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16558 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
16559 (TARGET_ISA_ROUND): Ditto.
16560 (TARGET_ROUND): Ditto.
16561 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
16562 * config/i386/i386.md: Ditto.
16563 * config/i386/sse.md: Ditto.
16564 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
16565 with OPTION_MASK_ISA_SSE4_1.
16567 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16570 * doc/extend.texi (x86 Built-in Functions): Correct the name of
16571 __builtin_ia32_lzcnt_u16.
16573 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
16576 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
16577 (rs6000_set_current_function): Rewrite function to use it.
16579 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
16582 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
16583 and TYPE_WARN_IF_NOT_ALIGN.
16584 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
16585 (handle_warn_if_not_align): New.
16586 (place_union_field): Call handle_warn_if_not_align.
16587 (place_field): Call handle_warn_if_not_align.
16588 Copy TYPE_WARN_IF_NOT_ALIGN.
16589 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
16590 (layout_type): Likewise.
16591 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
16593 (tree_decl_common): Add warn_if_not_align.
16594 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
16595 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
16596 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
16597 (DECL_WARN_IF_NOT_ALIGN): Likewise.
16598 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
16599 * doc/extend.texi: Document warn_if_not_aligned attribute.
16600 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
16602 2017-08-17 Martin Liska <mliska@suse.cz>
16605 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
16606 (get_data_dependence): Use it as pointer type.
16607 (distribute_loop): Likewise.
16609 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16611 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
16612 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
16613 (p8_vmrgow_<mode>_direct): New define_insn.
16614 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
16615 handle endianness for vmrgew and vmrgow permute patterns.
16617 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16619 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
16620 * config/rs6000/rs6000-cpus.def: Remove comment.
16621 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
16622 (POWERPC_MASKS): Likewise.
16623 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
16624 use of TARGET_VSX_TIMODE.
16625 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
16626 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
16627 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
16628 (rs6000_option_override_internal): Remove dead code.
16629 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
16630 (rs6000_legitimize_reload_address): Likewise.
16631 (rs6000_legitimate_address_p): Likewise.
16632 (rs6000_opt_masks): Delete "vsx-timode".
16633 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
16634 from function comment.
16635 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
16636 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
16637 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
16639 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
16640 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
16641 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
16642 (VSX_TI): Likewise.
16644 (define_peephole2): Likewise.
16646 2017-08-17 Martin Sebor <msebor@redhat.com>
16649 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
16650 past the end of an array.
16651 (test_pp_format): Add test cases.
16653 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
16655 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
16656 missing ECF_NOTHROW flags.
16658 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16661 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
16662 operands residing in integer registers.
16663 (*vsx_le_perm_load_<mode>): Likewise.
16664 (*vsx_le_perm_store_<mode>): Likewise.
16665 (define_peephole2): Add peepholes to optimize the above.
16667 2017-08-17 Marek Polacek <polacek@redhat.com>
16669 PR middle-end/81814
16670 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
16671 to mimic what shorten_compare did. Change the return type to bool.
16672 (fold_cond_expr_with_comparison): Update call to
16673 operand_equal_for_comparison_p.
16674 (fold_ternary_loc): Likewise.
16676 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
16678 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
16680 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
16682 2017-08-17 Richard Biener <rguenther@suse.de>
16684 * tree-ssa-structalias.c (solve_graph): When propagating
16685 to successors update the graphs succ edges and avoid duplicate work.
16687 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
16690 * config/i386/i386.c (ix86_option_override_internal): Save target
16691 specific options after ix86_stack_protector_guard_reg was changed.
16693 2017-08-17 Richard Biener <rguenther@suse.de>
16695 PR tree-optimization/81827
16696 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
16697 (new_var_info): Initialize it conservatively.
16698 (get_call_vi): Mark register vars.
16699 (new_scalar_tmp_constraint_exp): Likewise.
16700 (handle_rhs_call): Likewise.
16701 (handle_const_call): Likewise.
16702 (create_function_info_for): Likewise.
16703 (solve_constraints): Sort varinfos to separate register from
16704 non-register vars to pack points-to solution bitmaps during
16707 2017-08-17 Marek Polacek <polacek@redhat.com>
16709 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
16711 2017-08-17 Richard Biener <rguenther@suse.de>
16713 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
16714 to true when overflow is undefined and we saturated the result.
16716 2017-08-17 Alan Modra <amodra@gmail.com>
16719 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
16720 Don't use store multiple if only one reg needs saving.
16721 (interesting_frame_related_regno): New function.
16722 (rs6000_frame_related): Don't emit frame info for regs that
16724 (rs6000_emit_epilogue): Likewise.
16726 2017-08-16 Nathan Sidwell <nathan@acm.org>
16728 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
16729 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
16730 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
16731 * tree.c (free_lang_data_in_type): Use else-if chain. Always
16732 clear TYPE_LANG_1. Remove obsolete member-function stripping.
16733 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
16734 (verify_type): Adjust for TYPE_BINFO move.
16735 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
16736 process TYPE_BINFO directly.
16737 (hash_tree): Likewise.
16738 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16740 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16743 2017-08-16 David Malcolm <dmalcolm@redhat.com>
16745 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
16747 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
16750 * config/i386/i386.md (*anddi_1_btr): Change predicates of
16751 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
16752 Add ix86_binary_operator_ok to insn constraint.
16753 (*iordi_1_bts): Ditto.
16754 (*xordi_1_btc): Ditto.
16755 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
16756 Update corresponding peephole2 pattern.
16760 2017-08-16 Bin Cheng <bin.cheng@arm.com>
16762 PR tree-optimization/81832
16763 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
16764 copy loop header which has IFN_LOOP_DIST_ALIAS call.
16766 2017-08-16 Marek Polacek <polacek@redhat.com>
16769 * fold-const.c (fold_indirect_ref_1): Restore original behavior
16770 regarding size_zero_node.
16772 2017-08-16 Martin Liska <mliska@suse.cz>
16775 * config.gcc: Respect previously set extra_objs in case
16778 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
16780 PR tree-optimization/81835
16781 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
16782 the phi SSA_NAME. Check that the condition in a COND_EXPR does
16783 not depend on the phi.
16785 2017-08-16 Alan Modra <amodra@gmail.com>
16787 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
16790 2017-08-16 Alan Modra <amodra@gmail.com>
16792 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
16793 (save_reg_p): ..into this. Update all callers.
16794 (first_reg_to_save): Simplify.
16796 2017-08-16 Alan Modra <amodra@gmail.com>
16798 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
16801 2017-08-15 Joseph Myers <joseph@codesourcery.com>
16805 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
16806 constant count if that count is less than 32.
16808 2017-08-15 Nathan Sidwell <nathan@acm.org>
16810 * gcc.c (execute): Emit friendlier message if inferior is killed
16811 by an external cause.
16813 2017-08-15 Richard Biener <rguenther@suse.de>
16815 PR tree-optimization/81790
16816 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
16817 CONSTRUCTORs from simplifying and VN.
16819 2017-08-14 Martin Sebor <msebor@redhat.com>
16821 * builtin-attrs.def: Add comments.
16823 2017-08-14 Martin Sebor <msebor@redhat.com>
16826 * doc/extend.texi (attribute nonstring): Document new attribute.
16828 2017-08-14 Martin Sebor <msebor@redhat.com>
16831 * tree-diagnostic.c (default_tree_printer): Handle %G.
16832 * gimple-pretty-print.h (percent_G_format): Declare new function.
16833 * gimple-pretty-print.c (percent_G_format): Define.
16834 * tree-pretty-print.c (percent_K_format): Add argument.
16836 2017-08-14 Martin Sebor <msebor@redhat.com>
16838 PR translation/79998
16839 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
16840 Remove a stray space.
16842 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
16845 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
16846 (*iordi_1_bts): Ditto.
16847 (*xordi_1_btc): Ditto.
16849 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16852 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
16853 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16855 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
16856 quoted strings, and make more translator-friendly.
16857 (darwin_rs6000_override_options): Likewise.
16858 (rs6000_option_override_internal): Likewise.
16859 (rs6000_return_in_memory): Fix overlong line.
16860 (init_cmulative_args): Use quoted strings, and make more
16861 translator-friendly.
16862 (rs6000_pass_by_reference): Fix overlong line.
16863 (def_builtin): Use quoted strings.
16864 (altivec_expand_predicate_builtin): Use quoted strings, and make
16865 more translator-friendly.
16866 (htm_expand_builtin): Use quoted strings.
16867 (cpu_expand_builtin): Use quoted strings, and make more
16868 translator-friendly.
16869 (altivec_expand_builtin): Likewise.
16870 (paired_expand_predicate_builtin): Likewise.
16871 (rs6000_invalid_builtin): Likewise.
16872 (builtin_function_type): Use quoted strings.
16873 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
16874 more translator-friendly.
16875 (rs6000_trampoline_init): Likewise.
16876 (rs6000_handle_altivec_attribute): Likewise.
16877 (rs6000_inner_target_options): Use quoted strings.
16878 (rs6000_disable_incompatible_switches): Likewise.
16879 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
16880 strings, and make more translator-friendly.
16881 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
16883 2017-08-14 Bin Cheng <bin.cheng@arm.com>
16885 PR tree-optimization/81799
16886 * tree-loop-distribution.c (version_loop_by_alias_check): Force
16887 cond_expr to simple gimple operand.
16889 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
16891 PR middle-end/46932
16892 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
16894 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
16898 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
16899 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
16900 TARGET_GASISR_PROLOGUES.
16901 * config/avr/avr.c (avr_option_override): Same.
16902 (avr_pass_pre_proep::execute): Same.
16904 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
16907 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
16908 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
16910 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
16912 * config/i386/i386.md (*load_tp_<mode>): Redefine as
16913 define_insn_and_split. Split to a memory load from 0 in
16914 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
16915 using PTR mode iterator.
16916 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
16917 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
16918 (*add_tp_<mode>): Redefine as define_insn_and_split.
16919 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
16920 address space. Merge with *add_tp_x32 using PTR mode iterator.
16921 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
16922 Split to an add with a memory load from 0 in
16923 DEFAULT_TLS_SEG_REG address space.
16925 2017-08-12 Andrew Pinski <apinski@cavium.com>
16927 * config/aarch64/aarch64-option-extensions.def (rdma):
16928 Fix feature string to what Linux prints out in /proc/cpuinfo.
16930 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
16933 * dwarf2out.c (modified_type_die): For C typedef types that have
16934 an ultimate origin, process the ultimate origin instead of the
16936 (gen_typedef_die): Assert that input DECLs have no ultimate
16938 (gen_type_die_with_usage): For typedef variants that have an
16939 ultimate origin, just call gen_decl_die on the original DECL.
16940 (process_scope_var): Avoid creating DIEs for local typedefs and
16941 concrete static variables.
16943 2017-08-12 Alan Modra <amodra@gmail.com>
16947 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
16948 match gnu-user.h startfile.
16949 (ENDFILE_LINUX_SPEC): Similarly.
16951 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
16954 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
16956 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
16958 2017-08-11 Tamar Christina <tamar.christina@arm.com>
16959 * config/aarch64/aarch64.md (mov<mode>): Change.
16960 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16961 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
16962 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
16964 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
16966 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
16967 for storage order barriers.
16969 2017-08-11 Martin Liska <mliska@suse.cz>
16972 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
16973 variables of void type.
16975 2017-08-11 Martin Liska <mliska@suse.cz>
16977 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
16978 TARGET_SUPPORTS_ALIASES.
16979 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
16980 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
16981 (optimize_weakref): Likewise.
16982 * symtab.c (symtab_node::noninterposable_alias): Likewise.
16983 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
16984 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
16986 2017-08-11 Martin Liska <mliska@suse.cz>
16989 * config/i386/i386.c (make_resolver_func): Do complete
16990 refactoring of the function.
16992 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16995 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
16996 * config/i386/i386.c (ix86_stack_protect_guard): Use
16997 ix86_stack_protect_guard_symbol_str to generate varible declaration.
16998 * doc/invoke.texi (x86 Options): Document
16999 -mstack-protector-guard-symbol= option.
17001 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17003 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17004 * config/i386/i386.c (ix86_split_stack_guard): New function.
17005 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17006 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17007 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17008 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17009 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17010 (split_stack_space_check): Call ix86_split_stack_guard.
17012 2017-08-10 Martin Sebor <msebor@redhat.com>
17014 * print-tree.c (print_node): Print location using the established
17016 Replace spaces with colons.
17017 (debug_raw, debug): Ditto.
17019 2017-08-10 Martin Sebor <msebor@redhat.com>
17022 * pretty-print.c (pp_format): Correct the handling of %s precision.
17024 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
17027 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17029 (ix86_finalize_stack_frame_flags): This. Also clear
17030 frame_pointer_needed if -fno-omit-frame-pointer is used without
17032 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17033 with ix86_finalize_stack_frame_flags.
17034 (ix86_expand_epilogue): Likewise.
17035 (ix86_expand_split_stack_prologue): Likewise.
17036 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
17038 2017-08-10 Martin Liska <mliska@suse.cz>
17041 * c-attribs.c (handle_target_attribute):
17042 Report warning for an empty string argument of target attribute.
17044 2017-08-09 Jakub Jelinek <jakub@redhat.com>
17047 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
17049 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
17050 or DECL_NONLOCAL labels.
17051 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
17052 or DECL_NONLOCAL labels here.
17054 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
17056 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
17057 to indicate when early gimple folding has been disabled.
17058 (rs6000_gimple_fold_builtin): Add debug content.
17059 (rs6000_invalid_builtin): Fix whitespace.
17060 (rs6000_expand_builtin): Fix whitespace.
17061 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
17063 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
17066 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
17067 SAVE_MULTIPLE if not all the registers that saves, should be saved.
17069 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
17071 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
17072 (qdf24xx): Likewise.
17073 * config/aarch64/aarch64.md: Include falkor.md.
17074 * config/aarch64/falkor.md: New.
17076 2017-08-09 Marek Polacek <polacek@redhat.com>
17079 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
17080 * diagnostic.c (emit_diagnostic): Add a comment.
17081 (emit_diagnostic_valist): New function.
17083 2017-08-09 Marek Polacek <polacek@redhat.com>
17086 * input.c (make_location): New overload.
17087 * input.h (make_location): Declare.
17089 2017-08-08 Alan Modra <amodra@gmail.com>
17090 H.J. Lu <hongjiu.lu@intel.com>
17093 * gcc.c (NO_PIE_SPEC): Delete.
17094 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
17096 (LINK_PIE_SPEC): ..to here.
17097 (LINK_COMMAND_SPEC): Support -no-pie.
17098 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
17099 chain of crtbegin*.o selection, update for PIE_SPEC changes and
17101 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
17102 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
17103 (ENDFILE_CRTEND_SPEC): Similarly.
17105 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
17108 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
17109 (mstack-protector-guard-offset=): Ditto.
17110 * config/i386/i386.c (ix86_option_override): Handle
17111 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
17113 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
17114 ix86_stack_protect_guard_offset variables.
17115 (TARGET_STACK_PROTECT_GUARD): Always define.
17116 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
17117 and -mstack-protector-guard-offset= options.
17119 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17121 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
17122 boundary case for the last candidate.
17124 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17126 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
17127 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
17129 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17131 PR middle-end/19706
17132 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
17133 * config/aarch64/aarch64-builtins.c
17134 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
17135 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
17136 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
17138 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17139 Andrew Pinski <pinskia@gmail.com>
17141 PR middle-end/19706
17142 * internal-fn.def (XORSIGN): New.
17143 * optabs.def (xorsign_optab): New.
17144 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
17145 (convert_expand_mult_copysign): New.
17146 (pass_optimize_widening_mul::execute): Call
17147 convert_expand_mult_copysign.
17149 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17151 PR tree-optimization/81354
17152 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
17153 Insert on edges rather than explicitly creating landing pads.
17154 (analyze_candidates_and_replace): Commit edge inserts.
17156 2017-08-08 Richard Biener <rguenther@suse.de>
17158 PR middle-end/81719
17159 * tree-ssa-loop-niter.c: Include tree-dfa.h.
17160 (expand_simple_operations): Also look through ADDR_EXPRs with
17161 MEM_REF bases treating them as POINTER_PLUS_EXPR.
17163 2017-08-08 Richard Biener <rguenther@suse.de>
17165 PR tree-optimization/81723
17166 * tree-vect-slp.c (struct bst_traits): New hash traits.
17167 (bst_fail): New global.
17168 (vect_build_slp_tree_2): New worker, split out from ...
17169 (vect_build_slp_tree): ... this now wrapping it with using
17170 bst_fail set to cache SLP tree build fails. Properly handle
17172 (vect_analyze_slp_instance): Allocate and free bst_fail.
17174 2017-08-08 Martin Liska <mliska@suse.cz>
17177 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
17178 LABEL_DECLs that can be from a different function.
17180 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17182 PR tree-optimization/81744
17183 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
17184 loop's number of iterations.
17186 2017-08-08 Martin Liska <mliska@suse.cz>
17188 * asan.c: Include header files.
17189 * attribs.c (build_decl_attribute_variant): New function moved
17191 (build_type_attribute_qual_variant): Likewise.
17192 (cmp_attrib_identifiers): Likewise.
17193 (simple_cst_list_equal): Likewise.
17194 (omp_declare_simd_clauses_equal): Likewise.
17195 (attribute_value_equal): Likewise.
17196 (comp_type_attributes): Likewise.
17197 (build_type_attribute_variant): Likewise.
17198 (lookup_ident_attribute): Likewise.
17199 (remove_attribute): Likewise.
17200 (merge_attributes): Likewise.
17201 (merge_type_attributes): Likewise.
17202 (merge_decl_attributes): Likewise.
17203 (merge_dllimport_decl_attributes): Likewise.
17204 (handle_dll_attribute): Likewise.
17205 (attribute_list_equal): Likewise.
17206 (attribute_list_contained): Likewise.
17207 * attribs.h (lookup_attribute): New function moved from tree.[ch].
17208 (lookup_attribute_by_prefix): Likewise.
17209 * bb-reorder.c: Include header files.
17210 * builtins.c: Likewise.
17211 * calls.c: Likewise.
17212 * cfgexpand.c: Likewise.
17213 * cgraph.c: Likewise.
17214 * cgraphunit.c: Likewise.
17215 * convert.c: Likewise.
17216 * dwarf2out.c: Likewise.
17217 * final.c: Likewise.
17218 * fold-const.c: Likewise.
17219 * function.c: Likewise.
17220 * gimple-expr.c: Likewise.
17221 * gimple-fold.c: Likewise.
17222 * gimple-pretty-print.c: Likewise.
17223 * gimple.c: Likewise.
17224 * gimplify.c: Likewise.
17225 * hsa-common.c: Likewise.
17226 * hsa-gen.c: Likewise.
17227 * internal-fn.c: Likewise.
17228 * ipa-chkp.c: Likewise.
17229 * ipa-cp.c: Likewise.
17230 * ipa-devirt.c: Likewise.
17231 * ipa-fnsummary.c: Likewise.
17232 * ipa-inline.c: Likewise.
17233 * ipa-visibility.c: Likewise.
17235 * lto-cgraph.c: Likewise.
17236 * omp-expand.c: Likewise.
17237 * omp-general.c: Likewise.
17238 * omp-low.c: Likewise.
17239 * omp-offload.c: Likewise.
17240 * omp-simd-clone.c: Likewise.
17241 * opts-global.c: Likewise.
17242 * passes.c: Likewise.
17243 * predict.c: Likewise.
17244 * sancov.c: Likewise.
17245 * sanopt.c: Likewise.
17246 * symtab.c: Likewise.
17247 * toplev.c: Likewise.
17248 * trans-mem.c: Likewise.
17249 * tree-chkp.c: Likewise.
17250 * tree-eh.c: Likewise.
17251 * tree-into-ssa.c: Likewise.
17252 * tree-object-size.c: Likewise.
17253 * tree-parloops.c: Likewise.
17254 * tree-profile.c: Likewise.
17255 * tree-ssa-ccp.c: Likewise.
17256 * tree-ssa-live.c: Likewise.
17257 * tree-ssa-loop.c: Likewise.
17258 * tree-ssa-sccvn.c: Likewise.
17259 * tree-ssa-structalias.c: Likewise.
17260 * tree-ssa.c: Likewise.
17261 * tree-streamer-in.c: Likewise.
17262 * tree-vectorizer.c: Likewise.
17263 * tree-vrp.c: Likewise.
17264 * tsan.c: Likewise.
17265 * ubsan.c: Likewise.
17266 * varasm.c: Likewise.
17267 * varpool.c: Likewise.
17268 * tree.c: Remove functions moved to attribs.[ch].
17269 * tree.h: Likewise.
17270 * config/aarch64/aarch64.c: Add attrs.h header file.
17271 * config/alpha/alpha.c: Likewise.
17272 * config/arc/arc.c: Likewise.
17273 * config/arm/arm.c: Likewise.
17274 * config/avr/avr.c: Likewise.
17275 * config/bfin/bfin.c: Likewise.
17276 * config/c6x/c6x.c: Likewise.
17277 * config/cr16/cr16.c: Likewise.
17278 * config/cris/cris.c: Likewise.
17279 * config/darwin.c: Likewise.
17280 * config/epiphany/epiphany.c: Likewise.
17281 * config/fr30/fr30.c: Likewise.
17282 * config/frv/frv.c: Likewise.
17283 * config/ft32/ft32.c: Likewise.
17284 * config/h8300/h8300.c: Likewise.
17285 * config/i386/winnt.c: Likewise.
17286 * config/ia64/ia64.c: Likewise.
17287 * config/iq2000/iq2000.c: Likewise.
17288 * config/lm32/lm32.c: Likewise.
17289 * config/m32c/m32c.c: Likewise.
17290 * config/m32r/m32r.c: Likewise.
17291 * config/m68k/m68k.c: Likewise.
17292 * config/mcore/mcore.c: Likewise.
17293 * config/microblaze/microblaze.c: Likewise.
17294 * config/mips/mips.c: Likewise.
17295 * config/mmix/mmix.c: Likewise.
17296 * config/mn10300/mn10300.c: Likewise.
17297 * config/moxie/moxie.c: Likewise.
17298 * config/msp430/msp430.c: Likewise.
17299 * config/nds32/nds32-isr.c: Likewise.
17300 * config/nds32/nds32.c: Likewise.
17301 * config/nios2/nios2.c: Likewise.
17302 * config/nvptx/nvptx.c: Likewise.
17303 * config/pa/pa.c: Likewise.
17304 * config/pdp11/pdp11.c: Likewise.
17305 * config/powerpcspe/powerpcspe.c: Likewise.
17306 * config/riscv/riscv.c: Likewise.
17307 * config/rl78/rl78.c: Likewise.
17308 * config/rx/rx.c: Likewise.
17309 * config/s390/s390.c: Likewise.
17310 * config/sh/sh.c: Likewise.
17311 * config/sol2.c: Likewise.
17312 * config/sparc/sparc.c: Likewise.
17313 * config/spu/spu.c: Likewise.
17314 * config/stormy16/stormy16.c: Likewise.
17315 * config/tilegx/tilegx.c: Likewise.
17316 * config/tilepro/tilepro.c: Likewise.
17317 * config/v850/v850.c: Likewise.
17318 * config/vax/vax.c: Likewise.
17319 * config/visium/visium.c: Likewise.
17320 * config/xtensa/xtensa.c: Likewise.
17322 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17325 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
17326 constraints since the -mupper-regs-* switches have been
17328 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
17329 into a vector from a double word element that was extracted from
17330 another vector, and eliminate extra XXPERMDI instructions.
17331 (vsx_concat_<mode>_2): Likewise.
17332 (vsx_concat_<mode>_3): Likewise.
17333 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
17334 concat to allow optimizing inserts from previous extracts.
17336 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
17338 * config/i386/i386.c (ix86_stack_protect_guard): Generate
17339 memory reference to a SSP offset in TLS address space.
17340 (ix86_print_operand) <case '@'>: Remove.
17341 (ix86_print_operand_punct_valid_p): Remove '@' code.
17342 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
17343 UNSPEC_SP_TLS_TEST.
17344 (stack_tls_protect_set_<mode>): Remove.
17345 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
17346 (stack_tls_protect_test_<mode>): Remove.
17347 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
17349 2017-08-07 Olivier Hainque <hainque@adacore.com>
17352 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
17354 2017-08-07 Douglas Rupp <rupp@adacore.com>
17356 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
17357 variable was referenced as multidir in command.
17359 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17362 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
17365 2017-08-07 Martin Liska <mliska@suse.cz>
17367 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
17368 * config/rl78/rl78.c: Add include of attribs.h.
17369 * config/sh/sh.c: Likewise.
17370 * config/v850/v850.c: Likewise.
17372 2017-08-07 Tom de Vries <tom@codesourcery.com>
17374 PR middle-end/78266
17375 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
17377 2017-08-07 Martin Liska <mliska@suse.cz>
17379 * config/mips/mips.c: Include attribs.h.
17381 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
17384 * doc/invoke.texi: Document change in behvaior for -Ofast for
17387 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
17389 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
17391 (aarch64_pop_regs): Likewise.
17392 (aarch64_gen_load_pair): Likewise.
17393 (aarch64_save_callee_saves): Likewise.
17394 (aarch64_restore_callee_saves): Likewise.
17396 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17398 * config/i386/i386.c: Revert the last change.
17400 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17403 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17405 (ix86_finalize_stack_frame_flags): This. Also clear
17406 frame_pointer_needed if -fno-omit-frame-pointer is used without
17408 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17409 with ix86_finalize_stack_frame_flags.
17410 (ix86_expand_epilogue): Likewise.
17411 (ix86_expand_split_stack_prologue): Likewise.
17413 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17416 * config/i386/i386.c (get_builtin_code_for_version): Set priority
17417 to P_AES for Westmere.
17419 2017-08-07 Jonathan Yong <10walls@gmail.com>
17421 * config/i386/mingw.opt (fset-stack-executable): Removed.
17422 * config/i386/cygming.opt (fset-stack-executable): Moved
17424 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
17426 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
17428 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
17430 2017-08-07 Marek Polacek <polacek@redhat.com>
17432 PR middle-end/81737
17433 * fold-const.c (fold_indirect_ref_1): Check type_domain.
17435 2017-08-07 Martin Liska <mliska@suse.cz>
17437 * attribs.h (canonicalize_attr_name): New function.
17438 (cmp_attribs): Move from c-format.c and adjusted.
17439 (is_attribute_p): Moved from tree.h.
17440 * tree-inline.c: Add new includes.
17441 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
17442 (private_is_attribute_p): Remove.
17443 (private_lookup_attribute): Likewise.
17444 (private_lookup_attribute_by_prefix): Simplify.
17445 (remove_attribute): Use is_attribute_p.
17446 * tree.h: Remove removed declarations.
17448 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17450 PR middle-end/81698
17451 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
17452 instead of computing it in the function. Formatting fix.
17453 (expand_case): Don't rely on default_edge being the first edge,
17454 clear it if removing it, pass default_edge to
17455 emit_case_dispatch_table.
17456 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
17459 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
17461 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
17462 insn in the function, emit NOP after the insn.
17464 2017-08-06 Tom de Vries <tom@codesourcery.com>
17466 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
17469 2017-08-06 Tom de Vries <tom@codesourcery.com>
17471 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
17474 2017-08-04 Yury Gribov <tetra2005@gmail.com>
17476 PR tree-optimization/57371
17477 * match.pd: New pattern.
17479 2017-08-04 Marek Polacek <polacek@redhat.com>
17481 PR middle-end/81695
17482 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
17483 perform the computation in offset_int.
17485 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17487 PR tree-optimization/81136
17488 * tree-vectorizer.h: Include tree-hash-traits.h.
17489 (vec_base_alignments): New typedef.
17490 (vec_info): Add a base_alignments field.
17491 (vect_record_base_alignments): Declare.
17492 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
17494 (DR_IS_CONDITIONAL_IN_STMT): New macro.
17495 (create_data_ref): Add an is_conditional_in_stmt argument.
17496 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
17497 the is_conditional_in_stmt field.
17498 (data_ref_loc): Add an is_conditional_in_stmt field.
17499 (get_references_in_stmt): Set the is_conditional_in_stmt field.
17500 (find_data_references_in_stmt): Update call to create_data_ref.
17501 (graphite_find_data_references_in_stmt): Likewise.
17502 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
17503 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
17504 (vect_record_base_alignment): New function.
17505 (vect_record_base_alignments): Likewise.
17506 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
17507 for nested statements even if we fail to compute a misalignment.
17508 Use pooled base alignments for unconditional references.
17509 (vect_find_same_alignment_drs): Compare base addresses instead
17511 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
17512 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
17514 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17516 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
17517 Add an explicit name for the enum. Use auto_vec for slp_instances
17518 and grouped_stores.
17519 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
17521 (_bb_vec_info): Add a constructor and destructor.
17522 (vinfo_for_stmt): Return NULL for uids of -1 as well.
17523 (destroy_loop_vec_info): Delete.
17524 (vect_destroy_datarefs): Likewise.
17525 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
17526 (vec_info::vec_info): New function.
17527 (vec_info::~vec_info): Likewise.
17528 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
17529 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
17530 destroy_loop_vec_info.
17531 * tree-vect-loop.c (new_loop_vec_info): Replace with...
17532 (_loop_vec_info::_loop_vec_info): ...this.
17533 (destroy_loop_vec_info): Replace with...
17534 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
17535 the stmt_vec_infos. Leave handling of vec_info information to its
17536 destructor. Remove explicit vector releases.
17537 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
17538 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
17539 * tree-vect-slp.c (new_bb_vec_info): Replace with...
17540 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
17541 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
17542 (destroy_bb_vec_info): Replace with...
17543 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
17544 information to its destructor.
17545 (vect_slp_analyze_bb_1): Use new and delete instead of
17546 new_bb_vec_info and destroy_bb_vec_info.
17547 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
17550 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17552 * tree-data-ref.h (subscript): Add access_fn field.
17553 (data_dependence_relation): Add could_be_independent_p.
17554 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
17555 (same_access_functions): Move to tree-data-ref.c.
17556 * tree-data-ref.c (ref_contains_union_access_p): New function.
17557 (access_fn_component_p): Likewise.
17558 (access_fn_components_comparable_p): Likewise.
17559 (dr_analyze_indices): Add a reference to access_fn_component_p.
17560 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
17562 (constant_access_functions): Likewise.
17563 (add_other_self_distances): Likewise.
17564 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
17565 (initialize_data_dependence_relation): Use XCNEW and remove
17566 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
17567 of access functions that have the same type. Allow the
17568 subsequence to end with different bases in some circumstances.
17569 Record the chosen access functions in SUB_ACCESS_FN.
17570 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
17571 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
17572 (subscript_dependence_tester_1): Likewise dra and drb.
17573 (build_classic_dist_vector): Update calls accordingly.
17574 (subscript_dependence_tester): Likewise.
17575 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
17576 DDR_COULD_BE_INDEPENDENT_P.
17577 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
17578 comp_alias_ddrs instead of may_alias_ddrs.
17579 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17581 (vect_analyze_data_ref_dependence): Use it if
17582 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
17583 distance vectors if that fails.
17584 (dependence_distance_ge_vf): New function.
17585 (vect_prune_runtime_alias_test_list): Use it. Don't clear
17586 LOOP_VINFO_MAY_ALIAS_DDRS.
17588 2017-08-04 Richard Biener <rguenther@suse.de>
17590 PR middle-end/81705
17591 * fold-const.c (fold_binary_loc): Properly restrict
17592 minus_var0 && minus_var1 case when associating undefined overflow
17595 2017-08-04 Tom de Vries <tom@codesourcery.com>
17597 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
17599 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17601 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17602 Don't start diagnostic messages with a capital letter.
17603 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17605 (rs6000_invalid_builtin): Likewise.
17606 (rs6000_trampoline_init): Likewise.
17608 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17611 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
17612 after setting changeable df flags.
17614 2017-08-03 Richard Biener <rguenther@suse.de>
17616 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
17617 up if the use is in USE - X.
17619 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
17621 * toplev.c (dumpfile.h): New include.
17622 (internal_error_reentered): New static function. Use it...
17623 (internal_error_function): ...here to handle reentered internal_error.
17625 2017-08-03 Richard Biener <rguenther@suse.de>
17627 PR middle-end/81148
17628 * fold-const.c (split_tree): Add minus_var and minus_con
17629 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
17630 here but always use minus_*.
17631 (associate_trees): Assert we never associate with MINUS_EXPR
17632 and NULL first operand. Do not recurse for PLUS_EXPR operands
17633 when associating as MINUS_EXPR either.
17634 (fold_binary_loc): Track minus_var and minus_con.
17636 2017-08-03 Tom de Vries <tom@codesourcery.com>
17639 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
17640 ACCEL_COMPILER, apply finish_options on
17641 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
17643 2017-08-03 Tom de Vries <tom@codesourcery.com>
17646 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
17647 function_entry_patch_area_size > 0.
17649 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17652 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
17653 instead of 10??LU, perform unit multiplication in wide_int,
17654 don't change alloc_object_size_limit if the limit is larger
17657 PR tree-optimization/81655
17658 PR tree-optimization/81588
17659 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
17660 the case when ranges[i].low and high are 1 for unsigned type with
17663 PR middle-end/81052
17664 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
17665 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
17667 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17669 * tree-vrp.h: Add include guard.
17671 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
17674 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
17675 (ud2): New insn pattern.
17676 * config/i386/i386.c (ix86_expand_epilogue):
17677 For naked functions, generate ud2 instead of trap insn.
17679 2017-08-02 Marek Polacek <polacek@redhat.com>
17682 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
17684 2017-08-02 Tom de Vries <tom@codesourcery.com>
17685 Cesar Philippidis <cesar@codesourcery.com>
17687 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
17688 Add missing edge probabilities.
17690 2017-08-02 Tamar Christina <tamar.christina@arm.com>
17692 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
17693 Correct endianness.
17695 2017-08-02 Jakub Jelinek <jakub@redhat.com>
17697 PR middle-end/79499
17698 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
17699 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
17700 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
17701 prologue_seq sequences - if any.
17703 2017-08-02 Richard Biener <rguenther@suse.de>
17705 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
17706 via vectors if supported, integer extracts via punning if supported
17707 or otherwise vector extracts.
17709 2017-08-02 Richard Biener <rguenther@suse.de>
17711 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
17713 (bitmap_insert_into_set): ... this.
17715 2017-08-02 Richard Biener <rguenther@suse.de>
17717 PR tree-optimization/81633
17719 2015-08-17 Alan Hayward <alan.hayward@arm.com>
17721 PR tree-optimization/71752
17722 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
17724 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
17726 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
17727 (machine_function::call_ms2sysv_pad_out): Remove field.
17728 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
17729 (ix86_compute_frame_layout): Likewise.
17731 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
17734 * config/i386/i386.c (ix86_set_func_type): Disallow naked
17735 attribute with interrupt attribute.
17737 2017-08-01 Andrew Pinski <apinski@cavium.com>
17739 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
17740 BIT_INSERT_EXPR's operand 1
17741 to see if the types precision matches.
17743 2017-08-01 Martin Liska <mliska@suse.cz>
17745 PR middle-end/70140
17746 * builtins.c (expand_builtin_memcpy_args): Remove.
17747 (expand_builtin_memcpy): Call newly added function
17748 expand_builtin_memory_copy_args.
17749 (expand_builtin_memcpy_with_bounds): Likewise.
17750 (expand_builtin_mempcpy): Remove last argument.
17751 (expand_builtin_mempcpy_with_bounds): Likewise.
17752 (expand_builtin_memory_copy_args): New function created from
17753 expand_builtin_mempcpy_args with small modifications.
17754 (expand_builtin_mempcpy_args): Remove.
17755 (expand_builtin_stpcpy): Remove unused argument.
17756 (expand_builtin): Likewise.
17757 (expand_builtin_with_bounds): Likewise.
17759 2017-08-01 Martin Liska <mliska@suse.cz>
17762 Make mempcpy more optimal (PR middle-end/70140).
17764 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17767 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
17768 __builtin_vec_cmpne verify both arguments are compatible vectors
17769 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
17770 verify arg1_type is a pointer or array type. For __builtin_vec_st,
17771 move computation of aligned to after checking the argument types.
17775 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17776 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
17779 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
17780 Doug Rupp <rupp@adacore.com>
17781 Olivier Hainque <hainque@adacore.com>
17783 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
17784 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
17786 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
17787 for TARGET_OS_CPP_BUILTIN.
17788 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
17789 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
17791 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
17792 passing required abi options to the assembler for EABI configurations.
17793 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
17794 of .text.hot and .text.unlikely sections for kernel modules when
17795 using ARM style exceptions.
17796 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
17797 options. Add EXTRA_CC1_SPEC.
17798 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
17800 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
17802 (ARM_TARGET2_DWARF_FORMAT): Define.
17803 * config/arm/t-vxworks: Adjust multilib control to removal of the
17804 Diab command line options.
17806 2017-08-01 Martin Liska <mliska@suse.cz>
17808 PR gcov-profile/81561
17809 * gcov.c (unblock): Make unblocking safe as we need to preserve
17810 index correspondence of blocks and block_lists.
17812 2017-08-01 Richard Biener <rguenther@suse.de>
17814 PR tree-optimization/81181
17815 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
17816 (compute_antic): ... end of iteration here.
17818 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
17820 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
17821 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
17822 (ftree-slp-vectorize): Likewise.
17823 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
17824 can no longer be set independent of flag_tree_loop_vectorize.
17825 * omp-general.c (emp_max_vf): Likewise.
17826 * opts.c (enable_fdo_optimizations): Remove references to
17827 flag_tree_vectorize, these are now implicit.
17828 (common_handle_option): Remove handling for OPT_ftree_vectorize,
17829 and leave it for the options machinery.
17831 2017-08-01 Martin Liska <mliska@suse.cz>
17833 PR middle-end/70140
17834 * builtins.c (expand_builtin_memcpy_args): Remove.
17835 (expand_builtin_memcpy): Call newly added function
17836 expand_builtin_memory_copy_args.
17837 (expand_builtin_memcpy_with_bounds): Likewise.
17838 (expand_builtin_mempcpy): Remove last argument.
17839 (expand_builtin_mempcpy_with_bounds): Likewise.
17840 (expand_builtin_memory_copy_args): New function created from
17841 expand_builtin_mempcpy_args with small modifications.
17842 (expand_builtin_mempcpy_args): Remove.
17843 (expand_builtin_stpcpy): Remove unused argument.
17844 (expand_builtin): Likewise.
17845 (expand_builtin_with_bounds): Likewise.
17847 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17850 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
17851 print "ds:" only for immediates in generic address space.
17853 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17856 * config/i386/i386.c (ix86_funciton_naked): New prototype.
17857 (ix86_function_ok_for_sibcall): Return false for naked functions.
17859 2017-08-01 Richard Biener <rguenther@suse.de>
17861 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
17862 (compute_antic): Seed worklist with exit block predecessors.
17863 * cfganal.c (dfs_find_deadend): For a cycle return the source
17864 of the edge closing it.
17866 2017-08-01 Tamar Christina <tamar.christina@arm.com>
17868 * config/aarch64/aarch64.c
17869 (aarch64_can_const_movi_rtx_p): Move 0 check.
17871 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17873 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
17874 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
17876 * match.pd: Ditto in address comparison pattern.
17878 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17880 PR tree-optimization/81627
17881 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
17882 closed ssa form for store-store chain.
17884 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17886 PR tree-optimization/81620
17887 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
17888 for store-store chain.
17890 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17892 PR tree-optimization/81588
17893 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
17894 ranges[i].in_p, invert comparison code ccode. For >/>=,
17895 swap rhs1 and rhs2 and comparison code unconditionally,
17896 for </<= don't do that. Don't swap rhs1/rhs2 again if
17897 ranges[i].in_p, instead invert comparison code ccode if
17898 opcode or oe->rank is BIT_IOR_EXPR.
17901 * optabs.def (vec_extract_optab, vec_init_optab): Change from
17902 a direct optab to conversion optab.
17903 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
17904 with GET_MODE_INNER as last argument instead of optab_handler.
17905 * expmed.c (extract_bit_field_1): Likewise. Use vector from
17906 vector extraction if possible and optab is available.
17907 * expr.c (store_constructor): Use convert_optab_handler instead
17908 of optab_handler. Use vector initialization from smaller
17909 vectors if possible and optab is available.
17910 * tree-vect-stmts.c (vectorizable_load): Likewise.
17911 * doc/md.texi (vec_extract, vec_init): Document that the optabs
17912 now have two modes.
17913 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
17914 of vec_init from half-sized vectors with the same element mode.
17915 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
17916 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
17917 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
17918 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
17919 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
17920 after mode in gen_vec_extract* calls.
17921 (vec_extract<mode>): Renamed to ...
17922 (vec_extract<mode><ssescalarmodelower>): ... this.
17923 (vec_extract<mode><ssehalfvecmodelower>): New expander.
17924 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
17925 element mode after mode in gen_vec_init* calls.
17926 (VEC_INIT_HALF_MODE): New mode iterator.
17927 (vec_init<mode>): Renamed to ...
17928 (vec_init<mode><ssescalarmodelower>): ... this.
17929 (vec_init<mode><ssehalfvecmodelower>): New expander.
17930 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
17931 (vec_extractv2sfsf): ... this.
17932 (vec_initv2sf): Renamed to ...
17933 (vec_initv2sfsf): ... this.
17934 (vec_extractv2si): Renamed to ...
17935 (vec_extractv2sisi): ... this.
17936 (vec_initv2si): Renamed to ...
17937 (vec_initv2sisi): ... this.
17938 (vec_extractv4hi): Renamed to ...
17939 (vec_extractv4hihi): ... this.
17940 (vec_initv4hi): Renamed to ...
17941 (vec_initv4hihi): ... this.
17942 (vec_extractv8qi): Renamed to ...
17943 (vec_extractv8qiqi): ... this.
17944 (vec_initv8qi): Renamed to ...
17945 (vec_initv8qiqi): ... this.
17946 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
17947 (vec_init<mode>): Renamed to ...
17948 (vec_init<mode><VEC_base_l>): ... this.
17949 (vec_extract<mode>): Renamed to ...
17950 (vec_extract<mode><VEC_base_l>): ... this.
17951 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
17952 (vec_initv2sfsf): ... this.
17953 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
17954 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17955 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
17956 element mode after mode in gen_vec_init* calls.
17957 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
17958 (vec_init<mode><Vel>): ... this.
17959 (vec_extract<mode>): Renamed to ...
17960 (vec_extract<mode><Vel>): ... this.
17961 * config/aarch64/iterators.md (Vel): New mode attribute.
17962 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
17963 Add element mode after mode in gen_vec_extract* calls.
17964 * config/s390/vector.md (non_vec_l): New mode attribute.
17965 (vec_extract<mode>): Renamed to ...
17966 (vec_extract<mode><non_vec_l>): ... this.
17967 (vec_init<mode>): Renamed to ...
17968 (vec_init<mode><non_vec_l>): ... this.
17969 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
17970 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
17972 * config/arm/iterators.md (V_elem_l): New mode attribute.
17973 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
17974 (vec_extract<mode><V_elem_l>): ... this.
17975 (vec_extractv2di): Renamed to ...
17976 (vec_extractv2didi): ... this.
17977 (vec_init<mode>): Renamed to ...
17978 (vec_init<mode><V_elem_l>): ... this.
17979 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
17980 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
17981 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
17982 Add element mode after gen_vec_extract* calls.
17983 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
17984 (vec_init<mode><unitmode>): ... this.
17985 (vec_extract<mode>): Renamed to ...
17986 (vec_extract<mode><unitmode>): ... this.
17987 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
17988 (vec_init<mode><unitmode>): ... this.
17989 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
17990 (vec_initv2sfsf): ... this.
17991 (vec_extractv2sf): Renamed to ...
17992 (vec_extractv2sfsf): ... this.
17993 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
17994 Add element mode after gen_vec_extract* calls.
17995 * config/mips/mips.md (unitmode): New mode iterator.
17996 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
17997 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
17998 * config/spu/spu.md (inner_l): New mode attribute.
17999 (vec_init<mode>): Renamed to ...
18000 (vec_init<mode><inner_l>): ... this.
18001 (vec_extract<mode>): Renamed to ...
18002 (vec_extract<mode><inner_l>): ... this.
18003 * config/sparc/sparc.md (veltmode): New mode iterator.
18004 (vec_init<VMALL:mode>): Renamed to ...
18005 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18006 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18007 (vec_initv2sisi): ... this.
18008 (vec_initv2sf): Renamed to ...
18009 (vec_initv2sfsf): ... this.
18010 (vec_extractv2sf): Renamed to ...
18011 (vec_extractv2sfsf): ... this.
18012 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18013 (vec_init<mode>): Renamed to ...
18014 (vec_init<mode><VEC_base_l>): ... this.
18015 (vec_extract<mode>): Renamed to ...
18016 (vec_extract<mode><VEC_base_l>): ... this.
18017 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18018 (vec_initv2sfsf): ... this.
18019 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
18020 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18021 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
18022 gen_vec_init* calls.
18024 2017-08-01 Richard Biener <rguenther@suse.de>
18026 PR tree-optimization/81297
18027 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
18028 TREE_OVERFLOW from INTEGER_CSTs.
18030 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
18032 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
18034 2017-07-31 Carl Love <cel@us.ibm.com>
18036 * config/rs6000/rs6000-c: Add support for built-in functions
18037 vector signed char vec_xl_be (signed long long, signed char *);
18038 vector unsigned char vec_xl_be (signed long long, unsigned char *);
18039 vector signed int vec_xl_be (signed long long, signed int *);
18040 vector unsigned int vec_xl_be (signed long long, unsigned int *);
18041 vector signed long long vec_xl_be (signed long long, signed long long *);
18042 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
18043 vector signed short vec_xl_be (signed long long, signed short *);
18044 vector unsigned short vec_xl_be (signed long long, unsigned short *);
18045 vector double vec_xl_be (signed long long, double *);
18046 vector float vec_xl_be (signed long long, float *);
18047 * config/rs6000/altivec.h (vec_xl_be): Add #define.
18048 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
18049 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
18051 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
18052 (altivec_expand_builtin): Add switch statement to call
18053 altivec_expand_xl_be for each builtin.
18054 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
18055 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
18056 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
18057 __builtin_vsx_le_be_v16qi.
18058 * doc/extend.texi: Update the built-in documentation file for the
18059 new built-in functions.
18061 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18064 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
18066 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18068 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18070 * config.gcc: Add z14.
18071 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
18072 CPU model numbers for z13s and z14.
18073 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
18075 * config/s390/s390-opts.h (enum processor_type): Rename
18076 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18077 * config/s390/s390.c (processor_table): Add field for CPU name to
18078 be passed to Binutils.
18079 (s390_asm_output_machine_for_arch): Use the new field in
18080 processor_table for Binutils.
18081 (s390_expand_builtin): Replace arch12 with z14.
18082 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18083 (s390_get_sched_attrmask): Likewise.
18084 (s390_get_unit_mask): Likewise.
18085 * config/s390/s390.opt: Add z14 to processor_type enum.
18087 2017-07-31 Martin Jambor <mjambor@suse.cz>
18090 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
18091 regardless of optimization level.
18093 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
18094 Martin Liska <mliska@suse.cz>
18096 * predict.def: Remove old comment and adjust probability.
18097 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
18098 PREDICT statements.
18100 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18103 * config/i386/i386.c (ix86_function_naked): New function.
18104 (ix86_can_use_return_insn_p): Return false for naked functions.
18105 (ix86_expand_prologue): Skip prologue for naked functions.
18106 (ix86_expand_epilogue): Skip epilogue for naked functions
18107 and emit trap instruction.
18108 (ix86_warn_func_return): New function.
18109 (ix86_attribute_table): Add "naked" attribute specification.
18110 (TARGET_WARN_FUNC_RETURN): Define.
18111 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
18113 2017-07-31 Martin Liska <mliska@suse.cz>
18115 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
18116 (dump_gimple_bb_header): Always dump BB info.
18117 (pp_cfg_jump): Do not append info about BB when dumping a jump.
18119 2017-07-31 Martin Liska <mliska@suse.cz>
18122 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
18123 also with current_function_decl non-null equality.
18125 2017-07-31 Jakub Jelinek <jakub@redhat.com>
18128 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
18129 change type to the element type, instead add eltype variable and
18130 use it where we are interested in the element type.
18132 PR tree-optimization/81603
18133 * ipa-polymorphic-call.c
18134 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
18135 offset arithmetic in offset_int, bail out if the resulting bit offset
18136 doesn't fit into shwi.
18138 2017-07-31 Martin Liska <mliska@suse.cz>
18140 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
18141 (gimplify_save_expr): Fix comment.
18143 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
18146 * config/i386/i386.c (ix86_function_arg): Update arguments for
18148 (ix86_compute_frame_layout): Set the initial stack offset to
18149 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
18150 INCOMING_FRAME_SP_OFFSET.
18151 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
18152 stack before exception handler returns.
18153 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
18154 the 'ERROR_CODE' for exception handler.
18156 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
18158 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
18159 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
18160 (ASM_OUTPUT_REG_POP): Ditto.
18161 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
18162 instead of asm_fprintf to output pure string.
18164 2017-07-29 Jakub Jelinek <jakub@redhat.com>
18166 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
18167 to imported_module_or_decl hook.
18168 (debug_nothing_tree_tree_tree_bool): Remove.
18169 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
18170 * debug.c (do_nothing_debug_hooks): Use
18171 debug_nothing_tree_tree_tree_bool_bool instead of
18172 debug_nothing_tree_tree_tree_bool.
18173 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18174 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
18175 * sdbout.c (sdb_debug_hooks): Likewise.
18176 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
18177 (gen_namespace_die): Add DW_AT_export_symbols attribute if
18179 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
18180 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
18181 attribute, don't add anything.
18183 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18185 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
18186 (fold_build2_stat_loc): Likewise.
18187 (fold_build3_stat_loc): Likewise.
18188 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
18189 (fold_build1_loc): Remove macro.
18190 (fold_build2_loc): Likewise.
18191 (fold_build3_loc): Likewise.
18193 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18195 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
18196 (gimple_build_debug_bind_source_stat): Likewise.
18197 * gimple.h (gimple_build_debug_bind): Remove macro.
18198 (gimple_build_debug_bind_source): Likewise.
18200 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18202 * bitmap.c (bitmap_alloc): Adjust.
18203 (bitmap_gc_alloc): Likewise.
18204 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
18206 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18208 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
18209 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
18210 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
18211 (bitmap_gc_alloc_stat): Likewise.
18212 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
18214 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18216 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
18217 * rtl.h (shallow_copy_rtx): Remove macro.
18219 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18221 * emit-rtl.c (gen_raw_REG): Adjust.
18222 * gengenrtl.c (gendef): Likewise.
18223 * rtl.c (rtx_alloc_stat): Remove _stat from name.
18224 * rtl.h (rtx_alloc): Remove macro.
18226 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18228 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
18229 (build_tree_list_stat): Likewise.
18230 * tree.h (build_tree_list): Remove macro.
18231 (build_tree_list_vec): Likewise.
18233 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18235 * tree.c (make_vector_stat): Remove _stat from name.
18236 (build_vector_stat): Likewise.
18237 * tree.h (make_vector_stat): Remove macro.
18238 (build_vector_stat): Likewise.
18240 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18242 * tree.h (build_var_debug_value): Remove prototype.
18244 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18246 * tree.c (tree_cons_stat): Remove _stat from name.
18247 * tree.h (tree_cons): Remove macro.
18249 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18251 * tree.c (build_vl_exp_stat): Remove _stat from name.
18252 * tree.h (build_vl_exp): Remove macro.
18254 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18256 * tree.c (build_decl_stat): Remove _stat from name.
18257 * tree.h (build_decl): Remove macro.
18259 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18261 * gimple.c (gimple_build_with_ops_stat): Adjust.
18262 (gimple_alloc_stat): Remove _stat from name.
18263 * gimple.h (gimple_alloc): Remove macro.
18265 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18267 * tree.c (make_tree_vec_stat): Remove _stat from name.
18268 (grow_tree_vec_stat): Likewise.
18269 * tree.h (make_tree_vec_stat): Adjust prototype.
18270 (grow_tree_vec_stat): Likewise.
18271 (make_tree_vec): Remove macro.
18272 (grow_tree_vec): Likewise.
18274 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18276 * fold-const.c (fold_build1_stat_loc): Adjust.
18277 (fold_build2_stat_loc): Likewise.
18278 (fold_build3_stat_loc): Likewise.
18279 * tree.c (build0_stat): Remove _stat from name.
18280 (build1_stat): Likewise.
18281 (build2_stat): Likewise.
18282 (build3_stat): Likewise.
18283 (build4_stat): Likewise.
18284 (build5_stat): Likewise.
18285 * tree.h (build1_loc): Remove macro, and rename _stat function
18287 (build2_loc): Likewise.
18288 (build3_loc): Likewise.
18289 (build4_loc): Likewise.
18290 (build5_loc): Likewise.
18292 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18294 * tree.c (make_int_cst_stat): Remove _stat from name.
18295 * tree.h (make_int_cst_stat): Adjust prototype.
18296 (make_int_cst): Remove macro.
18298 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18300 * tree.c (make_tre_binfo_stat): Remove _stat from name.
18301 * tree.h (make_tree_binfo_stat): Adjust prototype.
18302 (make_tree_binfo): Remove.
18304 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18306 * tree.c (copy_node_stat): Rename to copy_node.
18307 (build_distinct_type_copy): Adjust.
18308 * tree.h (copy_node_stat): Adjust prototype.
18309 (copy_node): Remove macro.
18311 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18313 * tree.c (make_node_stat): rename to make_node.
18314 (build_tree_list_stat): Adjust.
18315 (build0_stat): Likewise.
18316 (build2_stat): Likewise.
18317 (build3_stat): Likewise.
18318 (build4_stat): Likewise.
18319 (build5_stat): Likewise.
18320 (build_decl_stat): Likewise.
18321 * tree.h (make_node_stat): Adjust prototype.
18322 (make_node): remove macro.
18324 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
18326 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
18327 (PPC_FEATURE2_SCV): Likewise.
18328 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
18330 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18332 * config/aarch64/aarch64.c
18333 (aarch64_internal_mov_immediate): Add new special pattern.
18334 * config/aarch64/aarch64.md (*movdi_aarch64):
18335 Add reg/32bit const mov case.
18337 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18338 Richard Sandiford <richard.sandiford@linaro.org>
18340 * config/aarch64/aarch64.md (mov<mode>): Generalize.
18341 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18342 Add integer and movi cases.
18343 (movi-split-hf-df-sf split, fp16): New.
18344 (enabled): Added TARGET_FP_F16INST.
18345 * config/aarch64/iterators.md (GPF_HF): New.
18346 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
18348 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18350 * config/aarch64/aarch64.c
18351 (aarch64_simd_container_mode): Add prototype.
18352 (aarch64_expand_mov_immediate): Add HI support.
18353 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
18354 (aarch64_can_const_movi_rtx_p): New.
18355 (aarch64_preferred_reload_class):
18356 Remove restrictions of using FP registers for certain SIMD operations.
18357 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
18358 (aarch64_valid_floating_const): Add integer move validation.
18359 (aarch64_simd_imm_scalar_p): Remove.
18360 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
18361 (aarch64_legitimate_constant_p): Expand list of supported cases.
18362 * config/aarch64/aarch64-protos.h
18363 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
18364 (aarch64_reinterpret_float_as_int): New.
18365 (aarch64_simd_imm_scalar_p): Remove.
18366 * config/aarch64/constraints.md (Uvi): New.
18367 (Dd): Split into Ds and new Dd.
18368 * config/aarch64/aarch64.md (*movsi_aarch64):
18370 (*movdi_aarch64): Add SIMD mov case.
18372 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18374 * tree-predcom.c: (struct chain): Handle store-store chain in which
18375 stores for elimination only store loop invariant values.
18376 (execute_pred_commoning_chain): Ditto.
18377 (prepare_initializers_chain_store_elim): Ditto.
18378 (prepare_finalizers): Ditto.
18379 (is_inv_store_elimination_chain): New function.
18380 (initialize_root_vars_store_elim_1): New function.
18382 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18384 * tree-predcom.c: Revise general description of the pass.
18385 (enum chain_type): New enum type for store elimination.
18386 (struct chain): New field supporting store elimination.
18387 (struct component): Ditto.
18388 (dump_chain): Dump store-stores chain.
18389 (release_chain): Release resources.
18390 (split_data_refs_to_components): Compute and create component
18391 contains only stores for elimination.
18392 (get_chain_last_ref_at): New function.
18393 (make_invariant_chain): Initialization.
18394 (make_rooted_chain): Specify chain type in parameter and record it.
18395 (add_looparound_copies): Skip for store-stores chain.
18396 (determine_roots_comp): Compute type of chain and pass it to
18398 (initialize_root_vars_store_elim_2): New function.
18399 (finalize_eliminated_stores): New function.
18400 (remove_stmt): Handle store for elimination.
18401 (execute_pred_commoning_chain): Execute predictive commoning on
18402 store-store chains.
18403 (determine_unroll_factor): Skip unroll for store-stores chain.
18404 (prepare_initializers_chain_store_elim): New function.
18405 (prepare_initializers_chain): Hanlde store-store chain.
18406 (prepare_finalizers_chain, prepare_finalizers): New function.
18407 (tree_predictive_commoning_loop): Return integer value indicating
18408 if loop is unrolled or lcssa form is corrupted.
18409 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
18411 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18413 * tree-predcom.c (initialize_root): Delete.
18414 (execute_pred_commoning_chain): Initialize root vars and replace
18415 reference of non-combined chain directly, rather than call above
18418 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18420 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
18421 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
18423 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18425 * tree-predcom.c (struct chain): New field init_seq.
18426 (release_chain): Release init_seq.
18427 (prepare_initializers_chain): Record intialization stmts in above
18429 (insert_init_seqs): New function.
18430 (tree_predictive_commoning_loop): Call insert_init_seqs.
18432 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18434 * tree-predcom.c (determine_roots_comp): Skip trivial components.
18436 2017-07-28 Richard Biener <rguenther@suse.de>
18438 * match.pd: Remove superfluous :c.
18439 * genmatch.c (simplify::id): Add member.
18440 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
18442 (current_id): New global.
18443 (dt_node::parent): Move from ...
18444 (dt_operand::parent): ... here. Add for_id member.
18445 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
18446 (decision_tree::find_node): Relax order requirement when
18447 merging DT_TRUE nodes to ones inbetween the current simplify
18448 and the one we try to merge with. Add diagnostic whenever
18449 we need to enforce pattern order by not merging.
18450 (decision_tree::insert): Set current_id.
18451 (decision_tree::print_node): Dump parent node and for_id.
18452 (parser::last_id): Add member.
18453 (parser::push_simplify): Assign unique id.
18454 (parser::parser): Initialize last_id.
18456 2017-07-28 Martin Liska <mliska@suse.cz>
18459 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
18460 gimple_build_debug_bind.
18462 2017-07-28 Richard Biener <rguenther@suse.de>
18464 PR tree-optimization/81502
18465 * match.pd: Add pattern combining BIT_INSERT_EXPR with
18467 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
18469 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
18470 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
18471 for BIT_FIELD_REF args.
18472 * fold-const.c (make_bit_field_ref): Likewise.
18473 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
18475 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18478 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
18479 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
18480 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
18481 Or it into SANITIZER_UNDEFINED.
18482 * ubsan.c: Include gimple-fold.h and varasm.h.
18483 (ubsan_expand_ptr_ifn): New function.
18484 (instrument_pointer_overflow): New function.
18485 (maybe_instrument_pointer_overflow): New function.
18486 (instrument_object_size): Formatting fix.
18487 (pass_ubsan::execute): Call instrument_pointer_overflow
18488 and maybe_instrument_pointer_overflow.
18489 * internal-fn.c (expand_UBSAN_PTR): New function.
18490 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
18491 * sanitizer.def (__ubsan_handle_pointer_overflow,
18492 __ubsan_handle_pointer_overflow_abort): New builtins.
18493 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
18494 * internal-fn.def (UBSAN_PTR): New internal function.
18495 * opts.c (sanitizer_opts): Add pointer-overflow.
18496 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
18497 * fold-const.c (build_range_check): Compute pointer range check in
18498 integral type if pointer arithmetics would be needed. Formatting
18501 2017-07-28 Martin Liska <mliska@suse.cz>
18504 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
18505 parameters that are of a variable-length.
18507 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
18509 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
18511 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
18512 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
18513 (CRT_CALL_STATIC_FUNCTION): Likewise.
18514 (ASM_DEFAULT_SPEC): New define.
18515 (ASM_SPEC32): Likewise.
18516 (ASM_SPEC64): Likewise.
18517 (ASM_SPEC_COMMON): Likewise.
18518 (ASM_SPEC): Likewise.
18519 (INVALID_64BIT): Likewise.
18520 (LINK_OS_DEFAULT_SPEC): Likewise.
18521 (LINK_OS_SPEC32): Likewise.
18522 (LINK_OS_SPEC64): Likewise.
18523 (POWERPC_LINUX): Likewise.
18524 (PTRDIFF_TYPE): Likewise.
18525 (RESTORE_FP_PREFIX): Likewise.
18526 (RESTORE_FP_SUFFIX): Likewise.
18527 (SAVE_FP_PREFIX): Likewise.
18528 (SAVE_FP_SUFFIX): Likewise.
18529 (SIZE_TYPE): Likewise.
18530 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
18531 (TARGET_64BIT): Likewise.
18532 (TARGET_64BIT): Likewise.
18533 (TARGET_AIX): Likewise.
18534 (WCHAR_TYPE_SIZE): Likewise.
18535 (WCHAR_TYPE): Undefine.
18536 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
18537 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
18538 (CPP_OS_RTEMS_SPEC): Delete.
18539 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
18540 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
18542 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
18544 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18546 PR tree-optimization/81578
18547 * tree-parloops.c (build_new_reduction): Bail out if
18548 reduction_code isn't one of the standard OpenMP reductions.
18549 Move the details printing after that decision.
18551 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18553 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
18554 related to reload_in_progress.
18555 (splat_input_operand): Likewise.
18556 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
18558 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
18560 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
18561 (TARGET_INSTANTIATE_DECLS): Likewise.
18562 (legitimate_indexed_address_p): Delete reload_in_progress code.
18563 (rs6000_debug_legitimate_address_p): Likewise.
18564 (rs6000_eliminate_indexed_memrefs): Likewise.
18565 (rs6000_emit_le_vsx_store): Likewise.
18566 (rs6000_emit_move_si_sf_subreg): Likewise.
18567 (rs6000_emit_move): Likewise.
18568 (register_to_reg_type): Likewise.
18569 (rs6000_pre_atomic_barrier): Likewise.
18570 (rs6000_machopic_legitimize_pic_address): Likewise.
18571 (rs6000_allocate_stack_temp): Likewise.
18572 (rs6000_address_for_fpconvert): Likewise.
18573 (rs6000_address_for_altivec): Likewise.
18574 (rs6000_secondary_memory_needed_rtx): Delete function.
18575 (rs6000_check_sdmode): Likewise.
18576 (rs6000_alloc_sdmode_stack_slot): Likewise.
18577 (rs6000_instantiate_decls): Likewise.
18578 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
18579 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
18580 Delete reload_in_progress.
18581 (*vec_reload_and_plus_<mptrsize>): Likewise.
18582 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
18583 (vsx_div_v2di): Likewise.
18584 (vsx_udiv_v2di): Likewise.
18586 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18588 * config/rs6000/rs6000.opt (mlra): Replace with stub.
18589 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
18590 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
18591 (rs6000_debug_reg_global): Delete print of LRA status.
18592 (rs6000_option_override_internal): Delete dead LRA related code.
18593 (rs6000_lra_p): Delete function.
18594 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
18596 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
18598 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
18599 * config/riscv/rtems.h: New file.
18601 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18602 Sudakshina Das <sudi.das@arm.com>
18604 * config/aarch64/aarch64.md
18605 (define_split for and<mode>3nr_compare): Move
18606 non aarch64_logical_operand to a register.
18607 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
18608 register immediate operand to a register.
18609 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
18611 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18613 PR middle-end/81564
18614 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
18616 2017-07-27 Richard Biener <rguenther@suse.de>
18618 PR tree-optimization/81573
18619 PR tree-optimization/81494
18620 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
18621 multi defuse cycle case.
18623 2017-07-27 Richard Biener <rguenther@suse.de>
18625 PR tree-optimization/81571
18626 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
18629 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
18631 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
18632 earlier and only if MASK_FPU is set. Adjust formatting.
18634 2017-07-27 Martin Liska <mliska@suse.cz>
18636 * opt-functions.awk: Add validation of value of Init.
18637 * optc-gen.awk: Pass new argument.
18639 2017-07-27 Martin Liska <mliska@suse.cz>
18641 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
18642 Fix wrong condition.
18644 2017-07-27 Martin Liska <mliska@suse.cz>
18646 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
18647 BBs and edges seen by autoFDO.
18649 2017-07-27 Richard Biener <rguenther@suse.de>
18651 PR tree-optimization/81502
18652 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
18653 with incompatible but same sized type.
18654 (execute_update_addresses_taken): Likewise.
18656 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
18658 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
18659 flag_tree_loop_vectorize rather than flag_tree_vectorize.
18661 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18664 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
18665 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
18666 Change s_operand to memory_operand.
18668 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
18670 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
18671 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
18672 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
18673 Emit instructions rather than returning an expression. Handle TFmode
18674 and KFmode by casting to TImode.
18675 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
18676 (rs6000_emit_le_vsx_store): Likewise.
18677 * config/rs6000/vsx.md (VSX_TI): New iterator.
18678 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
18679 (*vsx_le_undo_permute_<mode>): Likewise.
18680 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
18681 emit the split sequence.
18682 (*vsx_le_perm_store_<mode>): Likewise.
18684 2017-07-27 Jakub Jelinek <jakub@redhat.com>
18686 PR tree-optimization/81555
18687 PR tree-optimization/81556
18688 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
18689 if true, force CHANGED for the recursive invocation.
18690 (reassociate_bb): Remember original length of ops array, pass
18691 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
18693 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
18694 attributes for noipa attribute. For naked attribute use
18695 lookup_attribute first before lookup_attribute_spec.
18696 * final.c (rest_of_handle_final): Disable IPA RA for functions with
18698 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
18699 for functions with noipa attribute.
18700 (cgraph_externally_visible_p): Return true for functions with noipa
18702 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
18703 for functions with noipa attribute.
18704 * doc/extend.texi: Document noipa function attribute.
18705 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
18706 also for functions with noipa attribute.
18707 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
18709 2017-07-26 Andrew Pinski <apinski@cavium.com>
18711 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
18712 vec_unalign_load_cost and vec_unalign_store_cost.
18714 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
18716 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
18717 -mvsx-small-integer option.
18718 (ISA_3_0_MASKS_IEEE): Likewise.
18719 (OTHER_VSX_VECTOR_MASKS): Likewise.
18720 (POWERPC_MASKS): Likewise.
18721 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
18722 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
18723 code, only testing for DImode being allowed in non-VSX floating
18725 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
18726 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
18728 (rs6000_option_override_internal): Delete -mvsx-small-integer.
18729 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
18730 TARGET_P8_VECTOR test.
18731 (rs6000_secondary_reload_simple_move): Likewise.
18732 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
18733 since TARGET_P9_VECTOR was already tested.
18734 (rs6000_opt_masks): Remove -mvsx-small-integer.
18735 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
18736 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18738 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
18739 test for TARGET_VEXTRACTUB was used, and that uses
18741 (p9 extract splitter): Likewise.
18742 (vsx_extract_<mode>_di_p9): Likewise.
18743 (vsx_extract_<mode>_store_p9): Likewise.
18744 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
18745 for TARGET_P9_VECTOR was used. Delete code that is now dead with
18746 the elimination of TARGET_VSX_SMALL_INTEGER.
18747 (vsx_extract_<mode>_p8): Likewise.
18748 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
18749 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
18750 (vsx_set_<mode>_p9): Likewise.
18751 (vsx_set_v4sf_p9): Likewise.
18752 (vsx_set_v4sf_p9_zero): Likewise.
18753 (vsx_insert_extract_v4sf_p9): Likewise.
18754 (vsx_insert_extract_v4sf_p9_2): Likewise.
18755 * config/rs6000/rs6000.md (sign extend splitter): Change
18756 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
18757 (floatsi<mode>2_lfiwax_mem): Likewise.
18758 (floatunssi<mode>2_lfiwzx_mem): Likewise.
18759 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
18760 since a test for TARGET_P9_VECTOR was used.
18761 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18762 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
18763 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18764 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18765 TARGET_P8_VECTOR test.
18766 (fix_trunc<mode>si2_stfiwx): Likewise.
18767 (fix_trunc<mode>si2_internal): Likewise.
18768 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
18769 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18771 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18772 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18773 TARGET_P8_VECTOR test.
18774 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18775 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
18776 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18778 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18779 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
18780 since a test for TARGET_P9_VECTOR was used.
18781 (splitter for loading small constants): Likewise.
18783 2017-07-26 Andrew Pinski <apinski@cavium.com>
18785 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
18788 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
18791 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
18792 (fp_valid_at): Likewise.
18794 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18796 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
18797 (qdf24xx_addrcost_table): Likewise.
18798 (cortexa57_tunings): Update to use generic_branch_cost.
18799 (cortexa72_tunings): Likewise.
18800 (cortexa73_tunings): Likewise.
18801 (qdf24xx_tunings): Likewise.
18803 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18805 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
18806 (thunderx2t99_branch_cost): Likewise.
18807 (cortexa35_tunings): Update to use generic_branch_cost.
18808 (cortexa53_tunings): Likewise.
18809 (cortexa57_tunings): Likewise.
18810 (cortexa72_tunings): Likewise.
18811 (cortexa73_tunings): Likewise.
18812 (thunderx2t99_tunings): Likewise.
18814 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18816 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
18817 (sparc_option_override): Honour MASK_FSMULD.
18818 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
18819 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
18820 * config/sparc/sparc.opt (mfsmuld): New option.
18821 * doc/invoke.texi (mfsmuld): Document option.
18823 2017-07-26 Marek Polacek <polacek@redhat.com>
18825 PR middle-end/70992
18826 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
18828 2017-07-26 Richard Biener <rguenther@suse.de>
18830 * gimple-match-head.c (do_valueize): Return OP if valueize
18832 (get_def): New helper to get at the def stmt of a SSA name
18833 if valueize allows.
18834 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
18835 do_valueize to get at the def stmt.
18836 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
18838 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
18840 PR middle-end/46932
18841 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
18843 2017-07-26 Martin Liska <mliska@suse.cz>
18846 * function.c (expand_function_start): Make expansion of
18847 nonlocal_goto_save_area after parm_birth_insn.
18849 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18851 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
18852 from all CPU target flags enable members.
18854 2017-07-26 Richard Biener <rguenther@suse.de>
18856 * genmatch.c (dt_simplify::gen): Make iterator vars const.
18857 (decision_tree::gen): Make 'type' const.
18858 (write_predicate): Likewise.
18860 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18862 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
18863 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
18864 (rs6000_option_override_internal): Likewise.
18865 (rs6000_expand_vector_set): Likewise.
18866 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
18867 (TARGET_UPPER_REGS_SF): Likewise.
18868 (TARGET_UPPER_REGS_DI): Likewise.
18869 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
18870 (TARGET_DIRECT_MOVE_64BIT): Likewise.
18871 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
18872 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18873 (Splitters for DI constants in Altivec registers): Likewise.
18874 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
18875 (vsx_set_v4sf_p9): Likewise.
18876 (vsx_set_v4sf_p9_zero): Likewise.
18877 (vsx_insert_extract_v4sf_p9): Likewise.
18878 (vsx_insert_extract_v4sf_p9_2): Likewise.
18880 2017-07-25 Carl Love <cel@us.ibm.com>
18882 * doc/extend.texi: Update the built-in documentation file for the
18883 existing built-in functions
18884 vector signed char vec_cnttz (vector signed char);
18885 vector unsigned char vec_cnttz (vector unsigned char);
18886 vector signed short vec_cnttz (vector signed short);
18887 vector unsigned short vec_cnttz (vector unsigned short);
18888 vector signed int vec_cnttz (vector signed int);
18889 vector unsigned int vec_cnttz (vector unsigned int);
18890 vector signed long long vec_cnttz (vector signed long long);
18891 vector unsigned long long vec_cnttz (vector unsigned long long);
18893 2017-07-25 Andrew Pinski <apinski@cavium.com>
18895 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
18896 accesses where the use is for the first operand of a BIT_INSERT.
18898 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18901 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18902 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18904 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18906 * config/i386/gstabs.h: Delete.
18907 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
18909 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
18911 * config/i386/i386.c (ix86_decompose_address): Do not check for
18912 register RTX when looking at index_reg or base_reg.
18913 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
18915 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
18917 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
18918 to update EH info here.
18920 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18922 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
18924 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18926 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
18928 2017-07-25 Torsten Duwe <duwe@suse.de>
18930 * common.opt: Introduce -fpatchable-function-entry
18931 command line option, and its variables function_entry_patch_area_size
18932 and function_entry_patch_area_start.
18933 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
18934 including a two-value parser.
18935 * target.def (print_patchable_function_entry): New target hook.
18936 * targhooks.h (default_print_patchable_function_entry): New function.
18937 * targhooks.c (default_print_patchable_function_entry): Likewise.
18938 * toplev.c (process_options): Switch off IPA-RA if
18939 patchable function entries are being generated.
18940 * varasm.c (assemble_start_function): Look at the
18941 patchable-function-entry command line switch and current
18942 function attributes and maybe generate NOP instructions by
18943 calling the print_patchable_function_entry hook.
18944 * doc/extend.texi: Document patchable_function_entry attribute.
18945 * doc/invoke.texi: Document -fpatchable_function_entry
18946 command line option.
18947 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
18949 * doc/tm.texi: Re-generate.
18951 2017-07-25 Jakub Jelinek <jakub@redhat.com>
18954 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
18955 TARGET_AVX512DQ rather than TARGET_AVX512BW.
18957 2017-07-25 Tamar Christina <tamar.christina@arm.com>
18959 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
18961 2017-07-25 Richard Biener <rguenther@suse.de>
18963 PR tree-optimization/81455
18964 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
18965 not walk in cycles when looking for guards.
18967 2017-07-25 Richard Biener <rguenther@suse.de>
18969 PR tree-optimization/81529
18970 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
18971 when optimizing backedge uses.
18973 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
18975 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
18977 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
18978 to dl_section_ref. On AIX, append an expression to subtract
18979 the size of the section length to dl_section_ref.
18981 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
18983 * configure.ac: If any of the config.* scripts fail, exit 1.
18984 * configure: Regenerate.
18986 2017-07-25 Richard Biener <rguenther@suse.de>
18988 PR middle-end/81546
18989 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
18990 of immediate uses, be more verbose on errors.
18992 2017-07-25 Richard Biener <rguenther@suse.de>
18994 PR tree-optimization/81510
18995 * tree-vect-loop.c (vect_is_simple_reduction): When the
18996 reduction stmt is not inside the loop bail out.
18998 2017-07-25 Richard Biener <rguenther@suse.de>
19000 PR tree-optimization/81303
19001 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19002 profitability check against LOOP_VINFO_NITERSM1.
19004 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19006 * domwalk.c (cmp_bb_postorder): Simplify.
19007 (sort_bbs_postorder): New function. Use it...
19008 (dom_walker::walk): ...here to optimize common cases.
19010 2017-07-25 Martin Liska <mliska@suse.cz>
19013 * ipa-visibility.c (function_and_variable_visibility): Make the
19014 redirection just on target that supports aliasing.
19015 Fix GNU coding style.
19017 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
19020 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
19022 * config/arm/rtems.h: Likewise.
19023 * config/bfin/rtems.h: Likewise.
19024 * config/i386/rtemself.h: Likewise.
19025 * config/lm32/rtems.h: Likewise.
19026 * config/m32c/rtems.h: Likewise.
19027 * config/m68k/rtemself.h: Likewise.
19028 * config/microblaze/rtems.h: Likewise.
19029 * config/mips/rtems.h: Likewise.
19030 * config/moxie/rtems.h: Likewise.
19031 * config/nios2/rtems.h: Likewise.
19032 * config/powerpcspe/rtems.h: Likewise.
19033 * config/rs6000/rtems.h: Likewise.
19034 * config/rtems.h: Likewise.
19035 * config/sh/rtems.h: Likewise.
19036 * config/sh/rtemself.h: Likewise.
19037 * config/sparc/rtemself.h: Likewise.
19039 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19042 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
19043 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
19044 * tree-ssa-structalias.c (alias_get_name): Same.
19046 2017-07-25 Bin Cheng <bin.cheng@arm.com>
19049 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
19050 instructions if no du chain is found.
19052 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19054 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
19056 2017-07-25 Richard Biener <rguenther@suse.de>
19058 PR middle-end/81505
19059 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
19062 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19064 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
19065 upper-regs options.
19066 (ISA_2_7_MASKS_SERVER): Likewise.
19067 (ISA_3_0_MASKS_IEEE): Likewise.
19068 (OTHER_P8_VECTOR_MASKS): Likewise.
19069 (OTHER_VSX_VECTOR_MASKS): Likewise.
19070 (POWERPC_MASKS): Likewise.
19071 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
19072 duplicate list of options.
19073 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
19074 explicit -mupper-regs options.
19075 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
19076 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
19077 alias for -mupper-regs-df.
19078 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19079 (rs6000_init_hard_regno_mode_ok): Likewise.
19080 (rs6000_option_override_internal): Likewise.
19081 (rs6000_opt_masks): Likewise.
19082 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
19083 options in terms of whether -mvsx or -mpower8-vector was used.
19084 (TARGET_UPPER_REGS_DI): Likewise.
19085 (TARGET_UPPER_REGS_SF): Likewise.
19086 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
19087 -mupper-regs-* options.
19089 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19091 * passes.c (emergency_dump_function): Print some empty lines and a
19092 header before the RTL dump.
19094 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19096 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
19098 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
19101 * config/aarch64/aarch64.c (aarch64_classify_symbol):
19102 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
19104 2017-07-24 Carl Love <cel@us.ibm.com>
19106 * config/rs6000/rs6000-c.c: Add support for built-in functions
19107 vector float vec_extract_fp32_from_shorth (vector unsigned short);
19108 vector float vec_extract_fp32_from_shortl (vector unsigned short);
19109 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
19110 vec_extract_fp_from_shortl): Add defines for the two builtins.
19111 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
19112 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
19114 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
19115 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
19116 * doc/extend.texi: Update the built-in documentation file for the
19117 new built-in function.
19119 2017-07-24 Jakub Jelinek <jakub@redhat.com>
19122 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
19124 * doc/generic.texi: Likewise.
19125 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19126 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19128 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
19130 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
19131 (aarch64_mls_elt_merge<mode>): Likewise.
19133 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
19135 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19136 having __cxa_atexit.
19138 2017-07-23 Michael Collison <michael.collison@arm.com>
19140 * config/arm/arm.c (arm_option_override): Deprecate
19141 use of -mstructure-size-boundary.
19142 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
19143 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
19145 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19148 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19149 Reduce cost estimate for direct moves.
19151 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
19154 * config/i386/i386.c (ix86_option_override_internal): Disable
19155 BMI, BMI2 and TBM instructions for -m16.
19157 2017-07-21 Carl Love <cel@us.ibm.com>
19159 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19160 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19161 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19162 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19163 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19164 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19165 VMULOSW): New enum "unspec" values.
19166 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19167 altivec_vmulosw): New patterns.
19168 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19169 VMULOSW): Add definitions.
19171 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
19173 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
19174 (qdf24xx): Likewise.
19175 * config/aarch64/aarch64-options-extensions.def (rdma); New.
19176 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
19177 (AARCH64_FL_V8_1): Renumber.
19178 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
19179 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
19180 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
19181 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
19182 rdma to feature modifiers list.
19184 2017-07-21 Yury Gribov <tetra2005@gmail.com>
19186 PR middle-end/56727
19187 * ipa-visibility (function_and_variable_visibility): Convert
19188 recursive PLT call to direct call if appropriate.
19190 2017-07-21 Andrew Pinski <apinski@cavium.com>
19192 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
19193 operand 1 to see if the types precision matches.
19194 * fold-const.c (operand_equal_p): Likewise.
19196 2017-07-21 Richard Biener <rguenther@suse.de>
19198 PR tree-optimization/81303
19199 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
19200 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
19201 (vect_peeling_hash_get_lowest_cost): Adjust.
19202 (vect_enhance_data_refs_alignment): Likewise. Use
19203 vect_get_peeling_costs_all_drs to compute the penalty for no
19204 peeling to match up costs.
19206 2017-07-21 Richard Biener <rguenther@suse.de>
19208 PR tree-optimization/81500
19209 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
19210 we didn't identify a reduction path.
19212 2017-07-21 Tom de Vries <tom@codesourcery.com>
19213 Cesar Philippidis <cesar@codesourcery.com>
19215 PR gcov-profile/81442
19216 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
19219 2017-07-21 Tom de Vries <tom@codesourcery.com>
19222 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
19224 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
19225 nvptx_override_options_after_change.
19227 2017-07-21 Ulrich Drepper <drepper@redhat.com>
19229 * dwarf2out.c (output_file_names): Avoid double testing for
19230 dwarf_version >= 5.
19232 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
19234 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
19236 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19238 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
19240 (try_crossjump_to_edge): Do not punt on partitioned functions.
19242 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19244 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
19245 Put all BBs reachable only via paths crossing cold region to cold
19247 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
19249 2016-07-21 Richard Biener <rguenther@suse.de>
19251 PR tree-optimization/81303
19252 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
19253 into account prologue and epilogue iterations when raising
19254 min_profitable_iters to sth at least covering one vector iteration.
19256 2017-07-21 Tamar Christina <tamar.christina@arm.com>
19258 * config/arm/arm.c (arm_test_cpu_arch_dat):
19261 2017-07-20 Nathan Sidwell <nathan@acm.org>
19263 Remove TYPE_METHODS.
19264 * tree.h (TYPE_METHODS): Delete.
19265 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
19266 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
19267 (dbxout_type_methods): Scan TYPE_FIELDS.
19268 (dbxout_type): Don't check TYPE_METHODS here.
19269 * function.c (use_register_for_decl): Always ignore register for
19270 class types when not optimizing.
19271 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
19272 * tree.c (free_lang_data_in_type): Stitch out member functions and
19273 templates from TYPE_FIELDS.
19274 (build_distinct_type_copy, verify_type_variant,
19275 verify_type): Member fns are on TYPE_FIELDS.
19276 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
19277 * tree-pretty-print.c (dump_generic_node): Likewise.
19279 2017-07-20 Jakub Jelinek <jakub@redhat.com>
19282 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
19283 V2TImode and V4TImode.
19284 (ix86_expand_vector_extract): Likewise.
19285 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
19286 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
19287 (ssescalarmode): Handle V4TImode and V2TImode.
19288 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
19289 (*vec_extractv2ti, *vec_extractv4ti): New insns.
19290 (VEXTRACTI128_MODE): New mode iterator.
19291 (splitter for *vec_extractv?ti first element): New.
19292 (VEC_INIT_MODE): New mode iterator.
19293 (vec_init<mode>): Consolidate 3 expanders into one using
19294 VEC_INIT_MODE mode iterator.
19296 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19298 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
19299 non_spilled_static_chain_regno_p.
19301 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19303 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
19305 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
19307 * bb-reorder.c (connect_traces): Allow copying of blocks within
19310 2017-07-20 Richard Biener <rguenther@suse.de>
19312 * gimple.h (gimple_phi_result): Add gphi * overload.
19313 (gimple_phi_result_ptr): Likewise.
19314 (gimple_phi_arg): Likewise. Adjust index assert to only
19315 allow actual argument accesses rather than all slots available
19317 (gimple_phi_arg_def): Add gphi * overload.
19318 * tree-phinodes.c (make_phi_node): Initialize only actual
19320 (resize_phi_node): Clear memory not covered by old node,
19321 do not initialize excess argument slots.
19322 (reserve_phi_args_for_new_edge): Initialize new argument slot
19325 2017-07-20 Bin Cheng <bin.cheng@arm.com>
19327 PR tree-optimization/81388
19329 2016-07-21 Bin Cheng <bin.cheng@arm.com>
19331 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
19332 by removing computation of may_be_zero.
19334 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19335 Tom de Vries <tom@codesourcery.com>
19337 PR middle-end/81030
19338 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
19339 when gimple level profile disagrees with what RTL expander did.
19341 2017-07-20 Richard Biener <rguenther@suse.de>
19343 PR tree-optimization/61171
19344 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
19345 (vect_analyze_stmt): Add slp instance parameter.
19346 (vectorizable_reduction): Likewise.
19347 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
19348 (vect_is_simple_reduction): Deal with chains not detected
19349 as SLP reduction chain, specifically not properly associated
19350 chains containing a mix of plus/minus.
19351 (get_reduction_op): Remove.
19352 (get_initial_defs_for_reduction): Simplify, pass in whether
19353 this is a reduction chain, pass in the SLP node for the PHIs.
19354 (vect_create_epilog_for_reduction): Get the SLP instance as
19356 (vectorizable_reduction): Get the SLP instance as arg.
19357 During analysis remember the SLP node with the PHIs in the
19358 instance. Simplify getting at the vectorized reduction PHIs.
19359 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
19360 through SLP instance.
19361 (vect_slp_analyze_operations): Likewise.
19362 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
19363 (vect_transform_stmt): Likewise.
19365 2017-07-20 Tom de Vries <tom@codesourcery.com>
19367 PR tree-optimization/81489
19368 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
19369 read of phi arg location to before loop that modifies phi.
19371 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19373 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
19376 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19378 PR middle-end/81331
19379 * except.c (execute): Fix ordering issue.
19381 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19383 PR rtl-optimization/81423
19384 * combine.c (make_compound_operation_int): Don't try to optimize
19385 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
19387 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19389 PR rtl-optimization/81423
19390 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
19391 with a constant that is -1 in the truncated to mode.
19393 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19395 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
19396 (determine_unlikely_bbs): ... here.
19397 * predict.h (propagate_unlikely_bbs_forward): Declare.
19398 * cfgexpand.c (pass_expand::execute): Use it.
19399 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
19401 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
19402 propagate_unlikely_bbs_forward.
19404 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19406 PR middle-end/81331
19407 * except.c (maybe_add_nop_after_section_switch): New function.
19410 2017-07-19 Tom de Vries <tom@codesourcery.com>
19412 * gimple.h (gimple_phi_set_arg): Make assert more strict.
19414 2017-07-19 Tom de Vries <tom@codesourcery.com>
19416 * gimple.h (gimple_phi_arg): Make assert more strict.
19418 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
19420 * config.gcc (powerpc*-*-*): Add mmintrin.h.
19421 * config/rs6000/mmintrin.h: New file.
19422 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
19424 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19426 PR tree-optimization/81346
19427 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
19429 2017-07-19 Tom de Vries <tom@codesourcery.com>
19431 * config/nvptx/nvptx.md (VECIM): Add V2DI.
19433 2017-07-19 Tom de Vries <tom@codesourcery.com>
19435 * config/nvptx/nvptx-modes.def: Add V2DImode.
19436 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
19437 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
19438 (nvptx_output_mov_insn): Handle lack of mov.b128.
19439 (nvptx_print_operand): Handle 'H' and 'L' codes.
19440 (nvptx_vector_mode_supported): Allow V2DImode.
19441 (nvptx_preferred_simd_mode): New function.
19442 (nvptx_data_alignment): New function.
19443 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
19444 nvptx_preferred_simd_mode.
19445 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
19447 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
19449 2017-07-19 Tom de Vries <tom@codesourcery.com>
19451 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
19452 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
19453 (nvptx_vector_mode_supported): New function. Allow V2SImode.
19454 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
19455 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
19456 (mov<VECIM>_insn): New define_insn.
19457 (define_expand "mov<VECIM>): New define_expand.
19459 2017-07-19 Tom de Vries <tom@codesourcery.com>
19461 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
19463 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19465 PR tree-optimization/81346
19466 * fold-const.h (fold_div_compare, range_check_type): Declare.
19467 * fold-const.c (range_check_type): New function.
19468 (build_range_check): Use range_check_type.
19469 (fold_div_compare): No longer static, rewritten into
19470 a match.pd helper function.
19471 (fold_comparison): Don't call fold_div_compare here.
19472 * match.pd (X / C1 op C2): New optimization using fold_div_compare
19473 as helper function.
19475 2017-07-19 Nathan Sidwell <nathan@acm.org>
19477 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
19478 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
19479 * tree.c (find_decls_types_r, verify_type): Use
19480 TYPE_{MIN,MAX}_VALUE_RAW.
19481 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
19482 (hash_tree): Likewise.
19483 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19485 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19488 2017-07-18 Tom de Vries <tom@codesourcery.com>
19490 PR middle-end/81464
19491 * omp-expand.c (expand_omp_for_static_chunk): Handle
19492 equal-argument loop exit phi.
19494 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
19497 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
19498 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
19499 operand 2 predicate.
19500 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
19501 operand 2 predicate.
19502 (ror,rol -> rorx splitters): Use const_int_operand as
19503 operand 2 predicate.
19505 2017-06-18 Richard Biener <rguenther@suse.de>
19507 PR tree-optimization/81410
19508 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
19509 the gap in the ! slp_perm SLP case after each group.
19511 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19513 PR middle-end/81463
19514 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
19517 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19519 PR middle-end/81462
19520 * predict.c (set_even_probabilities): Cleanup; do not affect
19521 probabilities that are already known.
19522 (combine_predictions_for_bb): Call even when count is set.
19524 2017-07-18 Nathan Sidwell <nathan@acm.org>
19526 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
19529 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19532 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
19533 optimization for loop niter analysis.
19535 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
19538 * config/avr/avr.c (avr_optimize_casesi): Don't use
19539 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
19541 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
19543 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
19544 body_cost_vec from _vect_peel_extended_info.
19545 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
19546 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
19549 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19551 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
19553 2017-07-18 Richard Biener <rguenther@suse.de>
19555 PR tree-optimization/80620
19556 PR tree-optimization/81403
19557 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
19558 info when re-using a VN table entry.
19560 2017-07-18 Richard Biener <rguenther@suse.de>
19562 PR tree-optimization/81418
19563 * tree-vect-loop.c (vectorizable_reduction): Properly compute
19564 vectype_in. Verify that with lane-reducing reduction operations
19565 we have a single def-use cycle.
19567 2017-07-17 Carl Love <cel@us.ibm.com>
19569 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
19571 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19572 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19573 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19574 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19575 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19576 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19577 VMULOSW): New enum "unspec" values.
19578 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19579 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19580 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19581 altivec_vmulosw): New patterns.
19582 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19583 VMULOSW): Add definitions.
19584 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
19586 * config/alpha/alpha.c: Include predict.h.
19588 2017-07-17 Yury Gribov <tetra2005@gmail.com>
19590 * tree-vrp.c (compare_assert_loc): Fix comparison function
19591 to return predictable results.
19593 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19595 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
19597 (subdi3): Likewise.
19598 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
19599 * doc/invoke.texi (mexpand-adddi): Update text.
19601 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19603 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
19604 that also clobbers the CC register. The old expand code is moved
19606 (*arc_clzsi2): ... here.
19607 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
19608 the CC register. The old expand code is moved to ...
19609 (arc_ctzsi2): ... here.
19611 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19613 * config/arc/arc.opt (mindexed-loads): Use initial value
19614 TARGET_INDEXED_LOADS_DEFAULT.
19615 (mauto-modify-reg): Use initial value
19616 TARGET_AUTO_MODIFY_REG_DEFAULT.
19617 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19618 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19619 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19620 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19622 2017-07-17 Martin Liska <mliska@suse.cz>
19625 * opts.c (finish_options): Do not allow -fgnu-tm
19626 w/ -fsanitize={kernel-,}address. Say sorry.
19628 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19631 * tree-loop-distribution.c (classify_partition): Only assert on
19632 numer of iterations.
19633 (merge_dep_scc_partitions): Delete prameter. Update function call.
19634 (distribute_loop): Remove code handling loop with unknown niters.
19635 (pass_loop_distribution::execute): Skip loop with unknown niters.
19637 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19640 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
19641 function sort_partitions_by_post_order.
19643 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19645 PR tree-optimization/81374
19646 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
19647 the max index of basic blocks, rather than number of basic blocks.
19649 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19651 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
19653 (arc_legitimate_pic_operand_p): Likewise.
19654 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
19656 (arc_needs_pcl_p): Likewise.
19657 (arc_legitimate_pc_offset_p): Likewise.
19658 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
19659 function is also used in constrains.md.
19660 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
19661 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
19662 PLUS. Only return true/false in known cases, otherwise assert.
19663 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
19664 is already called in arc_legitimate_constant_p.
19665 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
19667 (LEGITIMATE_PIC_OPERAND_P): Use
19668 arc_raw_symbolic_reference_mentioned_p function.
19669 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
19674 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19675 Andrew Burgess <andrew.burgess@embecosm.com>
19677 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
19678 (arc_return_address_register): New function.
19679 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
19680 (arc_handle_fndecl_attribute): Add naked attribute.
19681 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19682 (TARGET_WARN_FUNC_RETURN): Likewise.
19683 (arc_allocate_stack_slots_for_args): New function.
19684 (arc_warn_func_return): Likewise.
19685 (machine_function): Change type fn_type.
19686 (arc_compute_function_type): Consider new naked function type,
19687 change function return type.
19688 (arc_must_save_register): Adapt to handle new
19689 arc_compute_function_type's return type.
19690 (arc_expand_prologue): Likewise.
19691 (arc_expand_epilogue): Likewise.
19692 (arc_return_address_regs): Delete.
19693 (arc_return_address_register): New function.
19694 (arc_epilogue_uses): Use above function.
19695 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
19696 (arc_function_type): Change encoding, add naked type.
19697 (ARC_INTERRUPT_P): Change to handle the new encoding.
19698 (ARC_FAST_INTERRUPT_P): Likewise.
19699 (ARC_NORMAL_P): Define.
19700 (ARC_NAKED_P): Likewise.
19701 (arc_compute_function_type): Delete prototype.
19702 * config/arc/arc.md (in_ret_delay_slot): Use
19703 arc_return_address_register function.
19704 (simple_return): Likewise.
19705 (p_return_i): Likewise.
19707 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19709 PR tree-optimization/81428
19710 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
19711 can't be built for those types.
19713 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19715 Remove stuff dead since r239246.
19717 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
19718 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
19719 (avr_inform_devices): Remove dead stuff.
19721 2017-07-17 Tamar Christina <tamar.christina@arm.com>
19723 * config/arm/arm_neon.h: Fix softp typo.
19725 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19727 PR tree-optimization/81365
19728 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
19729 aggregate moves onto bb predecessor edges, make sure there are no
19730 loads that could alias the lhs in between the start of bb and the
19733 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19736 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
19737 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
19738 [LSHIFTRT, outer_code = TRUNCATE]: Same.
19740 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19742 PR tree-optimization/81396
19743 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
19744 (init_symbolic_number): Initialize it to 1.
19745 (perform_symbolic_merge): Add n_ops from both operands into the new
19747 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
19748 without base_addr as useless if they need more than one operation.
19749 (bswap_replace): Handle !bswap case for NULL base_addr.
19751 2017-07-17 Tom de Vries <tom@codesourcery.com>
19754 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
19757 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
19759 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
19760 conditional builtin define __FIX_LEON3FT_B2BST.
19762 2017-07-17 Daniel Cederman <cederman@gaisler.com>
19764 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
19765 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
19768 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19770 PR rtl-optimization/81424
19771 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
19772 to remove potential trapping from operands if -fnon-call-exceptions.
19774 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19776 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
19777 profile_proability for scalling.
19778 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
19780 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19782 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
19784 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19786 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
19787 fixpoint arithmetics.
19789 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19791 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
19792 fixpoint arithmetics.
19794 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19796 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
19797 fixpoint arithmetics.
19799 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19801 * profile-count.h (profile_probability::from_reg_br_prob_note,
19802 profile_probability::to_reg_br_prob_note): New functions.
19803 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
19804 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
19805 * predict.c (probability_reliable_p): Update.
19806 (edge_probability_reliable_p): Update.
19807 (br_prob_note_reliable_p): Update.
19808 (invert_br_probabilities): Update.
19809 (add_reg_br_prob_note): New function.
19810 (combine_predictions_for_insn): Update.
19811 * asan.c (asan_clear_shadow): Update.
19812 * cfgbuild.c (compute_outgoing_frequencies): Update.
19813 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19814 (update_br_prob_note): Update.
19815 (rtl_verify_edges): Update.
19816 (purge_dead_edges): Update.
19817 (fixup_reorder_chain): Update.
19818 * emit-rtl.c (try_split): Update.
19819 * ifcvt.c (cond_exec_process_insns): Update.
19820 (cond_exec_process_if_block): Update.
19821 (dead_or_predicable): Update.
19822 * internal-fn.c (expand_addsub_overflow): Update.
19823 (expand_neg_overflow): Update.
19824 (expand_mul_overflow): Update.
19825 * loop-doloop.c (doloop_modify): Update.
19826 * loop-unroll.c (compare_and_jump_seq): Update.
19827 * optabs.c (emit_cmp_and_jump_insn_1): Update.
19828 * predict.h: Update.
19829 * reorg.c (mostly_true_jump): Update.
19831 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
19832 * config/alpha/alpha.c (emit_unlikely_jump): Update.
19833 * config/arc/arc.c: (emit_unlikely_jump): Update.
19834 * config/arm/arm.c: (emit_unlikely_jump): Update.
19835 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
19836 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
19837 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
19838 (ix86_print_operand): Update.
19839 (ix86_split_fp_branch): Update.
19840 (predict_jump): Update.
19841 * config/ia64/ia64.c (ia64_print_operand): Update.
19842 * config/mmix/mmix.c (mmix_print_operand): Update.
19843 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
19844 (rs6000_expand_split_stack_prologue): Update.
19845 * config/rs6000/rs6000.c: Update.
19846 * config/s390/s390.c (s390_expand_vec_strlen): Update.
19847 (s390_expand_vec_movstr): Update.
19848 (s390_expand_cs_tdsi): Update.
19849 (s390_expand_split_stack_prologue): Update.
19850 * config/sh/sh.c (sh_print_operand): Update.
19851 (expand_cbranchsi4): Update.
19852 (expand_cbranchdi4): Update.
19853 * config/sparc/sparc.c (output_v9branch): Update.
19854 * config/spu/spu.c (get_branch_target): Update.
19855 (ea_load_store_inline): Update.
19856 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
19857 * config/tilepro/tilepro.c: Update.
19859 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19861 * gimplify.c (mostly_copy_tree_r): Revert latest change.
19862 (gimplify_save_expr): Likewise.
19864 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19866 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
19868 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19870 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
19872 * timevar.def (TV_IPA_FNSUMMARY): Define.
19874 2017-07-16 Daniel Cederman <cederman@gaisler.com>
19876 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
19877 to back store errata sensitive sequence from being generated.
19878 (sqrtdf2_fix): Likewise.
19880 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19882 * tree-ssa-threadupdate.c (compute_path_counts,
19883 update_joiner_offpath_counts): Use profile_probability.
19885 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
19888 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19890 * config/arm/arm-c.c (arm_cpu_builtins): Define
19891 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19893 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
19895 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19896 array entries to represent __ieee128 versions of the
19897 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
19898 scalar_extract_sig, and scalar_insert_exp built-in functions.
19899 (altivec_resolve_overloaded_builtin): Add special case handling
19900 for the __builtin_scalar_insert_exp function, as represented by
19901 the P9V_BUILTIN_VEC_VSIEDP constant.
19902 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
19903 exponent support for __ieee128 argument.
19904 (VSESQP): Add scalar extract signature support for __ieee128
19906 (VSTDCNQP): Add scalar test negative support for __ieee128
19908 (VSIEQP): Add scalar insert exponent support for __int128 argument
19909 with __ieee128 result.
19910 (VSIEQPF): Add scalar insert exponent support for __ieee128
19911 argument with __ieee128 result.
19912 (VSTDCQP): Add scalar test data class support for __ieee128
19914 (VSTDCNQP): Add overload support for scalar test negative with
19915 __ieee128 argument.
19916 (VSTDCQP): Add overload support for scalar test data class
19917 __ieee128 argument.
19918 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
19919 UNSPEC_VSX_SXSIGDP.
19920 (UNSPEC_VSX_SIEXPQP): New constant.
19921 (xsxexpqp): New insn for VSX scalar extract exponent quad
19923 (xsxsigqp): New insn for VSX scalar extract significand quad
19925 (xsiexpqpf): New insn for VSX scalar insert exponent quad
19926 precision with floating point argument.
19927 (xststdcqp): New expand for VSX scalar test data class quad
19929 (xststdcnegqp): New expand for VSX scalar test negative quad
19931 (xststdcqp): New insn to match expansions for VSX scalar test data
19932 class quad precision and VSX scalar test negative quad precision.
19933 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
19934 special case operand checking to enforce that second operand of
19935 VSX scalar test data class with quad precision argument is a 7-bit
19937 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19938 prototypes and descriptions of __ieee128 versions of
19939 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
19940 scalar_test_data_class, and scalar_test_neg built-in functions.
19942 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19944 PR tree-optimization/81162
19945 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
19946 replace a negate with an add.
19948 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
19950 * doc/invoke.texi (arm/-mcpu): Document +crypto.
19952 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19954 * config/arm/arm-c.c (arm_cpu_builtins): Define
19955 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19957 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19959 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
19960 (armv8-r): Set ARM Cortex-R52 as default CPU.
19961 * config/arm/arm-tables.opt: Regenerate.
19962 * config/arm/arm-tune.md: Regenerate.
19963 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19965 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
19966 extension for -mcpu=cortex-r52.
19968 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19970 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
19971 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
19972 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
19973 (fp-armv8): Define it as FP_ARMv8 only.
19974 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
19975 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
19977 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
19978 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
19979 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
19980 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
19981 than TARGET_FPU_ARMV8.
19982 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
19983 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
19984 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
19986 * config/arm/neon.md (neon_vrint): Likewise.
19987 (neon_vcvt): Likewise.
19988 (neon_<fmaxmin_op><mode>): Likewise.
19989 (<fmaxmin><mode>3): Likewise.
19990 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
19991 * config/arm/predicates.md (arm_cond_move_operator): Check against
19992 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
19994 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
19996 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
19997 to top of function.
19999 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20001 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20002 loop in comment with memset.
20004 2017-07-14 Martin Liska <mliska@suse.cz>
20006 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20007 * dwarf2out.c (is_java): Remove the function.
20008 (output_pubname): Remove usage of the function.
20009 (lower_bound_default): Remove usage of DW_LANG_Java.
20010 (gen_compile_unit_die): Likewise.
20011 * gcc.c: Remove compiler defaults for .java and .zip files.
20012 * gimple-expr.c (remove_suffix): Change as there's no longer
20013 extension than 4-letter one.
20014 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20015 (gimplify_save_expr): Likewise.
20016 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20017 as it's possible even for other languages than Java.
20018 * langhooks.h (struct lang_hooks): Remove Java from a comment.
20019 * lto-opts.c (lto_write_options): Remove reference to Java.
20020 * opts.c (strip_off_ending): Update file extension handling.
20021 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
20022 * tree-eh.c (lower_resx): Likewise.
20023 * tree.c (free_lang_data_in_type): Remove dead code.
20024 (find_decls_types_r): Likewise.
20025 (build_common_builtin_nodes): Remove Java from a comment.
20026 (verify_type): Remove dead code.
20027 * varasm.c (assemble_external): Remove Java from a comment.
20029 2017-07-14 Martin Liska <mliska@suse.cz>
20031 * opts.c (finish_options): Add quotes.
20032 (common_handle_option): Likewise.
20034 2017-07-14 Martin Liska <mliska@suse.cz>
20036 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
20037 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
20038 Remove N_SO_PASCAL.
20039 * dwarf2out.c (lower_bound_default): Do not handle
20041 (gen_compile_unit_die): Likewise.
20042 * gcc.c: Remove default extension binding for GNU Pascal.
20043 * stmt.c: Remove Pascal language from a comment.
20044 * xcoffout.c: Likewise.
20046 2017-07-13 David Malcolm <dmalcolm@redhat.com>
20049 * diagnostic-show-locus.c (fixit_cmp): New function.
20050 (layout::layout): Sort m_fixit_hints.
20051 (column_range::column_range): Assert that the values are valid.
20052 (struct char_span): New struct.
20053 (correction::overwrite): New method.
20054 (struct source_line): New struct.
20055 (line_corrections::add_hint): Add assertions. Reimplement memcpy
20056 calls in terms of classes source_line and char_span, and
20057 correction::overwrite.
20058 (selftest::test_overlapped_fixit_printing_2): New function.
20059 (selftest::diagnostic_show_locus_c_tests): Call it.
20061 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
20063 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
20064 early if there is no lhs.
20066 2017-07-13 Martin Liska <mliska@suse.cz>
20068 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
20069 (gen_reference_type_die): Likewise.
20070 * stor-layout.c: Remove Pascal-related comment.
20072 2017-07-13 Martin Liska <mliska@suse.cz>
20074 * opts.c (finish_options): Add quotes to error messages.
20075 (parse_sanitizer_options): Likewise.
20077 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
20079 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
20081 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
20083 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
20085 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
20087 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
20089 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
20091 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
20094 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
20095 provides the hardware capability bits, define the macro
20096 __BUILTIN_CPU_SUPPORTS__.
20097 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
20098 if GLIBC does not provide the hardware capability bits. Add a
20099 gcc_unreachable call if the built-in cpu function is neither
20100 __builtin_cpu_is nor __builtin_cpu_supports.
20101 (rs6000_get_function_versions_dispatcher): Change the warning
20102 that an old GLIBC is used which does not export the capability
20103 bits to be an error.
20104 * doc/extend.texi (target_clones attribute): Document the
20105 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
20106 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
20107 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
20108 the macros defined by GCC if the newer GLIBC is available.
20110 2017-07-12 Jeff Law <law@redhat.com>
20112 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
20113 remaining includes slightly.
20114 * config/riscv/riscv-builtins.c: Include profile-count.h.
20116 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20119 * config/avr/avr.c (avr_set_current_function): In diagnostic
20120 messages: Quote keywords and (parts of) identifiers.
20121 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
20124 2017-07-12 Carl Love <cel@us.ibm.com>
20126 * config/rs6000/rs6000-c.c: Add support for built-in functions
20127 vector bool char vec_revb (vector bool char);
20128 vector bool short vec_revb (vector short char);
20129 vector bool int vec_revb (vector bool int);
20130 vector bool long long vec_revb (vector bool long long);
20131 * doc/extend.texi: Update the built-in documentation file for the
20132 new built-in functions.
20134 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20136 * config/s390/s390.md: Remove movcc splitter.
20138 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20140 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
20141 load/store on condition.
20143 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20146 * config/avr/avr.c (avr_encode_section_info)
20147 [progmem && !TREE_READONLY]: Error if progmem object needs
20150 2017-07-11 Michael Collison <michael.collison@arm.com>
20152 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
20155 2017-07-11 Carl Love <cel@us.ibm.com>
20157 * config/rs6000/rs6000-c.c: Add support for builtins
20158 vector unsigned int vec_parity_lsbb (vector signed int);
20159 vector unsigned int vec_parity_lsbb (vector unsigned int);
20160 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
20161 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
20162 vector unsigned long long vec_parity_lsbb (vector signed long long);
20163 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
20164 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
20165 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
20166 * doc/extend.texi: Update the built-in documentation file for the
20167 new built-in functions.
20169 2017-07-11 David Malcolm <dmalcolm@redhat.com>
20171 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
20172 (layout::m_primary_loc): New field.
20173 (layout::layout): Initialize new field. Move location filtering
20174 logic from here to...
20175 (layout::maybe_add_location_range): ...this new method. Add
20176 support for filtering to just the lines already specified by other
20178 (layout::will_show_line_p): New method.
20179 (gcc_rich_location::add_location_if_nearby): New method.
20180 (selftest::test_add_location_if_nearby): New test function.
20181 (selftest::diagnostic_show_locus_c_tests): Call it.
20182 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
20185 2017-07-11 Tom de Vries <tom@codesourcery.com>
20187 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
20188 (bb_first_real_insn): New function.
20189 (nvptx_single): Add extra initialization of broadcasted condition
20192 2017-07-11 Nathan Sidwell <nathan@acm.org>
20194 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
20196 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
20198 * doc/extend.texi (AVR Function Attributes): Remove weblink to
20199 Binutils doc as TEXI will mess them up.
20200 * doc/invoke.texi (AVR Options): Same here.
20202 2017-07-11 Daniel Cederman <cederman@gaisler.com>
20204 * config/sparc/sparc.opt (mfix-ut700): New option.
20205 (mfix-gr712rc): Likewise.
20206 (sparc_fix_b2bst): New variable.
20207 * doc/invoke.texi (SPARC options): Document them.
20208 (ARM options): Fix warnings.
20209 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
20210 instructions to prevent sequences that can trigger the store-store
20211 errata for certain LEON3FT processors.
20212 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
20213 (sparc_option_override): Set sparc_fix_b2bst appropriately.
20214 * config/sparc/sparc.md (fix_b2bst): New attribute.
20215 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
20217 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
20220 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
20222 (*rsqrtsf2_sse): Ditto.
20224 (div<mode>3): Macroize insn from divdf3 and divsf3
20225 using MODEF mode iterator.
20227 2017-07-10 Martin Sebor <msebor@redhat.com>
20229 PR tree-optimization/80397
20230 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
20231 instead of testing for equality to INTEGER_TYPE.
20233 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
20235 * config.gcc: Remove uclibc from arc target spec.
20237 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
20239 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
20241 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20244 * lto-wrapper.c (remove_option): New function.
20245 (merge_and_complain): Merge PIC/PIE options more realistically.
20247 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20249 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
20253 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
20254 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
20255 * config.in: Regenerate.
20256 * configure: Regenerate.
20257 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
20258 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
20259 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
20260 (TARGET_GASISR_PROLOGUES): ...target mask.
20261 * common/config/avr/avr-common.c
20262 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
20263 Set -mgas-isr-prologues.
20264 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
20265 INSERT_PASS_BEFORE for it.
20266 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
20267 * config/avr/avr.c (avr_option_override)
20268 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
20269 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
20270 (avr_attribute_table) <no_gccisr>: Add new function attribute.
20271 (avr_set_current_function) <is_no_gccisr>: Init machine field.
20272 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
20274 (make_avr_pass_pre_proep): New function.
20275 (emit_push_sfr) <treg>: Add argument to function and use it
20276 instead of TMP_REG.
20277 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
20278 and set machine->gasisr.yes.
20279 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
20280 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
20281 __gcc_isr.n_pushed to .L__stack_usage.
20282 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
20283 (avr_asm_final_postscan_insn): ...this new static function.
20284 * config/avr/avr.h (machine_function)
20285 <is_no_gccisr, use_L__stack_usage>: New fields.
20286 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
20287 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
20288 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
20289 (gasisr, *gasisr): New expander and insn.
20290 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20291 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
20292 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
20294 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
20296 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
20299 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20301 Move jump-tables out of .text again.
20304 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
20305 (ASM_OUTPUT_ADDR_VEC): New function.
20306 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
20307 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
20308 INSN_ADDRESSes as asm comment.
20309 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
20310 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
20311 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
20312 * config/avr/avr.md (*tablejump): Adjust comment.
20313 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
20314 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
20316 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
20317 (avr_output_addr_vec): New proto.
20318 (avr_log_t) <insn_addresses>: New field.
20320 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
20323 * config/i386/i386.c (ix86_function_arg_advance): Set
20324 outgoing_args_on_stack to true if there are outgoing arguments
20326 (ix86_function_arg): Likewise.
20327 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
20328 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
20329 * config/i386/i386.h (machine_function): Add
20330 outgoing_args_on_stack.
20332 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
20334 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20335 supporting pthreds.
20336 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
20338 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
20340 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
20341 (REAL_H): Remove $(MACHMODE_H).
20342 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
20344 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
20345 $(MACHMODE_H) and double-int.h.
20346 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
20348 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
20351 2017-07-07 Andrew Pinski <apinski@cavium.com>
20353 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
20354 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
20356 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20358 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
20359 Add warning if GCC was not configured to link against a GLIBC that
20360 exports the hardware capability bits.
20361 (make_resolver_func): Make resolver function private and not a
20362 COMDAT function. Create the name with clone_function_name instead
20363 of make_unique_name.
20366 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
20367 correct operand in doing the split.
20369 2017-07-07 Carl Love <cel@us.ibm.com>
20371 * config/rs6000/rs6000-c: Add support for built-in function
20372 vector unsigned short vec_pack_to_short_fp32 (vector float,
20374 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
20375 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
20376 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
20377 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
20378 (convert_4f32_8i16): Add define_expand.
20379 * doc/extend.texi: Update the built-in documentation file for the
20380 new built-in function.
20382 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20384 * config/sparc/m8.md: New file.
20385 * config/sparc/sparc.md: Include m8.md.
20387 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20389 * config/sparc/sparc.opt: New option -mvis4b.
20390 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
20391 (sparc_option_override): Handle VIS4B.
20392 (enum sparc_builtins): Define
20393 SPARC_BUILTIN_DICTUNPACK{8,16,32},
20394 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
20395 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
20396 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
20397 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
20398 (check_constant_argument): New function.
20399 (sparc_vis_init_builtins): Define builtins
20400 __builtin_vis_dictunpack{8,16,32},
20401 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
20402 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
20403 __builtin_vis_fpcmpde{8,16,32}shl and
20404 __builtin_vis_fpcmpur{8,16,32}shl.
20405 (sparc_expand_builtin): Check that the constant operands to
20406 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
20407 constant and in range.
20408 * config/sparc/sparc-c.c (sparc_target_macros): Handle
20410 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
20411 (SPARC_IMM5_P): Likewise.
20412 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
20413 (enabled): Handle vis4b.
20414 (UNSPEC_DICTUNPACK): New unspec.
20415 (UNSPEC_FPCMPSHL): Likewise.
20416 (UNSPEC_FPUCMPSHL): Likewise.
20417 (UNSPEC_FPCMPDESHL): Likewise.
20418 (UNSPEC_FPCMPURSHL): Likewise.
20419 (cpu_feature): New CPU feature `vis4b'.
20420 (dictunpack{8,16,32}): New insns.
20421 (FPCSMODE): New mode iterator.
20422 (fpcscond): New code iterator.
20423 (fpcsucond): Likewise.
20424 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
20425 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
20426 (fpcmpde{8,16,32}{si,di}shl): Likewise.
20427 (fpcmpur{8,16,32}{si,di}shl): Likewise.
20428 * config/sparc/constraints.md: Define constraints `q' for unsigned
20429 2-bit integer constants and `t' for unsigned 5-bit integer
20431 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
20433 (imm5_operand_dictunpack16): Likewise.
20434 (imm5_operand_dictunpack32): Likewise.
20435 (imm2_operand): Likewise.
20436 * doc/invoke.texi (SPARC Options): Document -mvis4b.
20437 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
20438 ditunpack* and fpcmp*shl builtins.
20440 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20442 * config.gcc: Handle m8 in --with-{cpu,tune} options.
20443 * config.in: Add HAVE_AS_SPARC6 define.
20444 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
20446 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
20448 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
20449 (CPP_CPU_SPEC): Handle m8.
20450 (ASM_CPU_SPEC): Likewise.
20451 * config/sparc/sparc-opts.h (enum processor_type): Add
20453 * config/sparc/sparc.c (m8_costs): New struct.
20454 (sparc_option_override): Handle TARGET_CPU_m8.
20455 (sparc32_initialize_trampoline): Likewise.
20456 (sparc64_initialize_trampoline): Likewise.
20457 (sparc_issue_rate): Likewise.
20458 (sparc_register_move_cost): Likewise.
20459 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
20460 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
20461 (ASM_CPU64_DEFAULT_SPEC): Likewise.
20462 (CPP_CPU_SPEC): Handle M8.
20463 (ASM_CPU_SPEC): Likewise.
20464 (AS_M8_FLAG): Define.
20465 * config/sparc/sparc.md: Add m8 to the cpu attribute.
20466 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
20467 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
20469 * configure: Regenerate.
20470 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
20473 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20475 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
20477 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
20478 ("*movdi_insn_sp32"): Do not set v3pipe.
20479 ("*movsi_insn"): Likewise.
20480 ("*movdi_insn_sp64"): Likewise.
20481 ("*movsf_insn"): Likewise.
20482 ("*movdf_insn_sp32"): Likewise.
20483 ("*movdf_insn_sp64"): Likewise.
20484 ("*zero_extendsidi2_insn_sp64"): Likewise.
20485 ("*sign_extendsidi2_insn"): Likewise.
20486 ("*mov<VM32:mode>_insn"): Likewise.
20487 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20488 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20489 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20490 ("<vlop:code><VL:mode>3"): Likewise.
20491 ("*not_<vlop:code><VL:mode>3"): Likewise.
20492 ("*nand<VL:mode>_vis"): Likewise.
20493 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
20494 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
20495 ("one_cmpl<VL:mode>2"): Likewise.
20496 ("faligndata<VM64:mode>_vis"): Likewise.
20497 ("alignaddrsi_vis"): Likewise.
20498 ("alignaddrdi_vis"): Likweise.
20499 ("alignaddrlsi_vis"): Likewise.
20500 ("alignaddrldi_vis"): Likewise.
20501 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20502 ("bmaskdi_vis"): Likewise.
20503 ("bmasksi_vis"): Likewise.
20504 ("bshuffle<VM64:mode>_vis"): Likewise.
20505 ("cmask8<P:mode>_vis"): Likewise.
20506 ("cmask16<P:mode>_vis"): Likewise.
20507 ("cmask32<P:mode>_vis"): Likewise.
20508 ("pdistn<P:mode>_vis"): Likewise.
20509 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20511 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20513 * config/sparc/sparc.md ("subtype"): New insn attribute.
20514 ("*wrgsr_sp64"): Set insn subtype.
20515 ("*rdgsr_sp64"): Likewise.
20516 ("alignaddrsi_vis"): Likewise.
20517 ("alignaddrdi_vis"): Likewise.
20518 ("alignaddrlsi_vis"): Likewise.
20519 ("alignaddrldi_vis"): Likewise.
20520 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20521 ("fexpand_vis"): Likewise.
20522 ("fpmerge_vis"): Likewise.
20523 ("faligndata<VM64:mode>_vis"): Likewise.
20524 ("bshuffle<VM64:mode>_vis"): Likewise.
20525 ("cmask8<P:mode>_vis"): Likewise.
20526 ("cmask16<P:mode>_vis"): Likewise.
20527 ("cmask32<P:mode>_vis"): Likewise.
20528 ("fchksm16_vis"): Likewise.
20529 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
20530 ("fmean16_vis"): Likewise.
20531 ("fp<plusminus_insn>64_vis"): Likewise.
20532 ("<plusminus_insn>v8qi3"): Likewise.
20533 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20534 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
20535 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
20536 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
20537 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
20538 ("*movqi_insn"): Likewise.
20539 ("*movhi_insn"): Likewise.
20540 ("*movsi_insn"): Likewise.
20541 ("movsi_pic_gotdata_op"): Likewise.
20542 ("*movdi_insn_sp32"): Likewise.
20543 ("*movdi_insn_sp64"): Likewise.
20544 ("movdi_pic_gotdata_op"): Likewise.
20545 ("*movsf_insn"): Likewise.
20546 ("*movdf_insn_sp32"): Likewise.
20547 ("*movdf_insn_sp64"): Likewise.
20548 ("*zero_extendhisi2_insn"): Likewise.
20549 ("*zero_extendqihi2_insn"): Likewise.
20550 ("*zero_extendqisi2_insn"): Likewise.
20551 ("*zero_extendqidi2_insn"): Likewise.
20552 ("*zero_extendhidi2_insn"): Likewise.
20553 ("*zero_extendsidi2_insn_sp64"): Likewise.
20554 ("ldfsr"): Likewise.
20555 ("prefetch_64"): Likewise.
20556 ("prefetch_32"): Likewise.
20557 ("tie_ld32"): Likewise.
20558 ("tie_ld64"): Likewise.
20559 ("*tldo_ldub_sp32"): Likewise.
20560 ("*tldo_ldub1_sp32"): Likewise.
20561 ("*tldo_ldub2_sp32"): Likewise.
20562 ("*tldo_ldub_sp64"): Likewise.
20563 ("*tldo_ldub1_sp64"): Likewise.
20564 ("*tldo_ldub2_sp64"): Likewise.
20565 ("*tldo_ldub3_sp64"): Likewise.
20566 ("*tldo_lduh_sp32"): Likewise.
20567 ("*tldo_lduh1_sp32"): Likewise.
20568 ("*tldo_lduh_sp64"): Likewise.
20569 ("*tldo_lduh1_sp64"): Likewise.
20570 ("*tldo_lduh2_sp64"): Likewise.
20571 ("*tldo_lduw_sp32"): Likewise.
20572 ("*tldo_lduw_sp64"): Likewise.
20573 ("*tldo_lduw1_sp64"): Likewise.
20574 ("*tldo_ldx_sp64"): Likewise.
20575 ("*mov<VM32:mode>_insn"): Likewise.
20576 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20577 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20579 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20581 * config/sparc/sparc.md ("type"): New insn type viscmp.
20582 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
20584 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
20585 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
20586 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20587 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
20589 ("n7_vis_logical_11cycle"): Likewise.
20590 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
20591 * config/sparc/niagara2.md ("niag3_vis": Likewise.
20592 * config/sparc/niagara.md ("niag_vis"): Likewise.
20593 * config/sparc/ultra3.md ("us3_fga"): Likewise.
20594 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
20596 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20598 * config/sparc/sparc.md: New instruction type `bmask'.
20599 (bmaskdi_vis): Use the `bmask' type.
20600 (bmasksi_vis): Likewise.
20601 * config/sparc/ultra3.md (us3_array): Likewise.
20602 * config/sparc/niagara7.md (n7_array): Likewise.
20603 * config/sparc/niagara4.md (n4_array): Likewise.
20604 * config/sparc/niagara2.md (niag2_vis): Likewise.
20605 (niag3_vis): Likewise.
20606 * config/sparc/niagara.md (niag_vis): Likewise.
20608 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20610 * ipa-comdats.c: Remove optimize check from gate.
20611 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
20612 for functions not optimized.
20613 (ipa_fn_summary_read): Skip optimize check.
20614 (ipa_fn_summary_write): Likewise.
20615 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
20617 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
20619 (can_inline_edge_p): Check flag_pcc_struct_return for match.
20620 (check_callers): Give up on caller which is not optimized.
20621 (inline_small_functions): Likewise.
20622 (ipa_inline): Do not give up when not optimizing.
20623 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
20624 away unoptimizes cdtors.
20625 (whole_program_function_and_variable_visibility): Do
20626 ipa_discover_readonly_nonaddressable_vars in LTO mode.
20627 * ipa.c (process_references): Do not check optimize.
20628 (symbol_table::remove_unreachable_nodes): Update optimize check.
20629 (set_writeonly_bit): Update optimize check.
20630 (pass_ipa_cdtor_merge::gate): Do not check optimize.
20631 (pass_ipa_single_use::gate): Remove.
20633 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20635 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
20636 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
20637 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
20638 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
20639 permute_load, permute_store, adjust_extract, adjust_splat,
20640 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
20641 replace_swap_with_copy, dump_swap_insn_table,
20642 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
20643 recombine_lvx_pattern, recombine_stvx_pattern,
20644 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
20645 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
20646 to file rs6000-p8swap.c.
20647 * config/rs6000/rs6000-p8swap.c: New file.
20648 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
20649 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
20650 and rs6000*-*-* targets.
20652 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20654 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
20656 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20658 * lto-wrapper.c (merge_and_complain): Do not merge
20659 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
20660 fsigned_zeros, ftrapping_math, fwrapv.
20661 (append_compiler_options): Do not track these options.
20662 (append_linker_options): Likewie
20664 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20666 * cgraphunit.c (cgraph_node::finalize_function): When
20667 !flag_toplevel_reorde set no_reorder flag.
20668 (varpool_node::finalize_decl): Likewise.
20669 (symbol_table::compile): Drop no toplevel reorder path.
20671 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20673 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
20674 edges; zero probability is not better than uninitialized.
20676 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20678 * asan.h (asan_sanitize_allocas_p): Declare.
20679 * asan.c (asan_sanitize_allocas_p): New function.
20680 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
20681 (handle_builtin_alloca): Likewise.
20682 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
20683 if !asan_sanitize_allocas_p.
20684 * params.def (asan-instrument-allocas): Add new option.
20685 * params.h (ASAN_PROTECT_ALLOCAS): Define.
20686 * opts.c (common_handle_option): Disable allocas sanitization for
20689 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20691 * asan.c: Include gimple-fold.h.
20692 (get_last_alloca_addr): New function.
20693 (handle_builtin_stackrestore): Likewise.
20694 (handle_builtin_alloca): Likewise.
20695 (asan_emit_allocas_unpoison): Likewise.
20696 (get_mem_refs_of_builtin_call): Add new parameter, remove const
20697 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
20698 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
20699 (instrument_builtin_call): Pass gimple iterator to
20700 get_mem_refs_of_builtin_call.
20701 (last_alloca_addr): New global.
20702 * asan.h (asan_emit_allocas_unpoison): Declare.
20703 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
20704 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
20705 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
20706 if function calls alloca.
20707 * gimple-fold.c (replace_call_with_value): Remove static keyword.
20708 * gimple-fold.h (replace_call_with_value): Declare.
20709 * internal-fn.c: Include asan.h.
20710 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
20711 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
20713 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20715 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
20716 (C_SELFTEST_FLAGS): New.
20717 (CPP_SELFTEST_FLAGS): New.
20718 (SELFTEST_DEPS): New, from deps of s-selftest.
20719 (C_SELFTEST_DEPS): New, from deps of s-selftest.
20720 (CPP_SELFTEST_DEPS): New.
20721 (selftest): Add dependency on s-selftest-c++.
20722 (s-selftest): Rename to...
20723 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
20724 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
20725 than SELFTEST_FLAGS.
20726 (selftest-gdb): Rename to...
20727 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
20729 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
20730 (selftest-valgrind): Rename to...
20731 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
20733 (selftest-valgrind): Reintroduce as an alias for
20734 selftest-c-valgrind.
20735 (s-selftest-c++): New.
20736 (selftest-c++-gdb): New.
20737 (selftest-c++-valgrind): New.
20739 2017-07-06 Olivier Hainque <hainque@adacore.com>
20741 * gcc.c (process_command): When deciding if undefined variables
20742 should be ignored when processing specs, accept "gcc -v" as well.
20744 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20746 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
20747 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
20749 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20751 * config/arm/arm-cpus.in (armv8-r): Add new entry.
20752 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
20753 * config/arm/arm-tables.opt: Regenerate.
20754 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
20756 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
20758 2017-07-06 Carl Love <cel@us.ibm.com>
20760 * ChangeLog: Clean up from mid air collision
20762 2017-07-06 Carl Love <cel@us.ibm.com>
20764 * config/rs6000/rs6000-c.c: Add support for built-in functions
20765 vector signed int vec_subc (vector signed int, vector signed int);
20766 vector signed __int128 vec_subc (vector signed __int128,
20767 vector signed __int128);
20768 vector unsigned __int128 vec_subc (vector unsigned __int128,
20769 vector unsigned __int128);
20770 vector signed int vec_sube (vector signed int, vector signed int,
20771 vector signed int);
20772 vector unsigned int vec_sube (vector unsigned int,
20773 vector unsigned int,
20774 vector unsigned int);
20775 vector signed __int128 vec_sube (vector signed __int128,
20776 vector signed __int128,
20777 vector signed__int128);
20778 vector unsigned __int128 vec_sube (vector unsigned __int128,
20779 vector unsigned __int128,
20780 vector unsigned __int128);
20781 vector signed int vec_subec (vector signed int, vector signed int,
20782 vector signed int);
20783 vector unsigned int vec_subec (vector unsigned int,
20784 vector unsigned int,
20785 vector unsigned int);
20786 vector signed __int128 vec_subec (vector signed __int128,
20787 vector signed __int128,
20788 vector signed__int128);
20789 vector unsigned __int128 vec_subec (vector unsigned __int128,
20790 vector unsigned __int128,
20791 vector unsigned __int128);
20792 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
20793 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
20794 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
20795 BU_ALTIVEC_OVERLOAD_X definitions.
20796 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
20797 * doc/extend.texi: Update the built-in documentation file for the new
20798 built-in functions.
20800 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20803 * diagnostic-show-locus.c (layout::layout): Use start and finish
20804 spelling location for the start and finish of each range.
20805 * genmatch.c (linemap_client_expand_location_to_spelling_point):
20806 Add unused aspect param.
20807 * input.c (expand_location_1): Add "aspect" param, and use it
20808 to access the correct part of the location.
20809 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
20811 (expand_location_to_spelling_point): Likewise.
20812 (linemap_client_expand_location_to_spelling_point): Add "aspect"
20813 param, and pass it to expand_location_1.
20815 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
20817 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
20818 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
20819 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
20820 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
20821 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
20822 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
20823 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
20824 _mm_maskz_getmant_ss): New intrinsics.
20825 (__builtin_ia32_getexpss128_mask): Changed to ...
20826 __builtin_ia32_getexpss128_round ... this.
20827 (__builtin_ia32_getexpsd128_mask): Changed to ...
20828 __builtin_ia32_getexpsd128_round ... this.
20829 * config/i386/i386-builtin-types.def
20830 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
20831 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
20832 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
20833 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
20834 __builtin_ia32_getmantss_mask_round): New builtins.
20835 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
20836 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
20837 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
20838 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
20839 * config/i386/sse.md
20840 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
20841 avx512f_sgetexp<mode><mask_scalar_name>
20842 <round_saeonly_scalar_name> ... this.
20843 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
20844 %0, %1, %2<round_saeonly_op3>}): Changed to ...
20845 vgetexp<ssescalarmodesuffix>
20846 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20847 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
20848 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
20849 avx512f_vgetmant<mode><mask_scalar_name>
20850 <round_saeonly_scalar_name> ... this.
20851 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
20852 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
20853 vgetmant<ssescalarmodesuffix>
20854 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
20855 %0<mask_scalar_operand4>, %1, %2
20856 <round_saeonly_scalar_mask_op4>, %3} ... this.
20857 * config/i386/subst.md (mask_scalar_operand4,
20858 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
20859 round_saeonly_scalar_nimm_predicate): New subst attributes.
20861 2017-07-06 Julia Koval <julia.koval@intel.com>
20863 * config/i386/i386.c (ix86_erase_embedded_rounding):
20864 Remove code for old rounding pattern.
20866 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
20868 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
20870 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
20872 * doc/sourcebuild.texi (Test Directives, Variants of
20873 dg-require-support): Add documentation for dg-require-stack-check.
20875 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
20877 * config/i386/subst.md (mask_scalar, round_scalar,
20878 round_saeonly_scalar): New meta-templates.
20879 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
20880 round_scalar_mask_operand3, round_scalar_mask_op3,
20881 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
20882 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
20883 round_saeonly_scalar_constraint,
20884 round_saeonly_scalar_prefix): New subst attribute.
20885 * config/i386/sse.md
20886 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
20887 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
20888 <round_scalar_name> ... this.
20889 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
20890 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
20891 <round_scalar_name> ... this.
20892 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
20893 <sse>_vm<code><mode>3<mask_scalar_name>
20894 <round_saeonly_scalar_name> ... this.
20895 (v<plusminus_mnemonic><ssescalarmodesuffix>
20896 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20897 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20898 v<plusminus_mnemonic><ssescalarmodesuffix>
20899 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20900 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20901 (v<multdiv_mnemonic><ssescalarmodesuffix>
20902 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20903 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20904 v<multdiv_mnemonic><ssescalarmodesuffix>
20905 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20906 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20907 (v<maxmin_float><ssescalarmodesuffix>
20908 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
20909 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
20910 v<maxmin_float><ssescalarmodesuffix>
20911 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20912 %0<mask_scalar_operand3>, %1, %<iptr>2
20913 <round_saeonly_scalar_mask_op3>} ... this.
20915 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
20917 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
20918 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
20920 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
20921 Alan Hayward <alan.hayward@arm.com>
20922 David Sherwood <david.sherwood@arm.com>
20924 * combine.c (simplify_if_then_else): Remove "enum" before
20926 * compare-elim.c (can_eliminate_compare): Likewise.
20927 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20929 (aarch64_lookup_simd_builtin_type): Likewise.
20930 (aarch64_simd_builtin_type): Likewise.
20931 (aarch64_init_simd_builtin_types): Likewise.
20932 (aarch64_simd_expand_args): Likewise.
20933 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
20935 (aarch64_reverse_mask): Likewise.
20936 (aarch64_simd_emit_reg_reg_move): Likewise.
20937 (aarch64_gen_adjusted_ldpstp): Likewise.
20938 (aarch64_ccmp_mode_to_code): Likewise.
20939 (aarch64_operands_ok_for_ldpstp): Likewise.
20940 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20941 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20943 (aarch64_min_divisions_for_recip_mul): Likewise.
20944 (aarch64_reassociation_width): Likewise.
20945 (aarch64_get_condition_code_1): Likewise.
20946 (aarch64_simd_emit_reg_reg_move): Likewise.
20947 (aarch64_simd_attr_length_rglist): Likewise.
20948 (aarch64_reverse_mask): Likewise.
20949 (aarch64_operands_ok_for_ldpstp): Likewise.
20950 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20951 (aarch64_gen_adjusted_ldpstp): Likewise.
20952 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
20954 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
20955 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20956 (arm_lookup_simd_builtin_type): Likewise.
20957 (arm_simd_builtin_type): Likewise.
20958 (arm_init_simd_builtin_types): Likewise.
20959 (arm_expand_builtin_args): Likewise.
20960 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
20961 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
20962 (ft32_setup_incoming_varargs): Likewise.
20963 (ft32_function_arg): Likewise.
20964 (ft32_function_arg_advance): Likewise.
20965 (ft32_pass_by_reference): Likewise.
20966 (ft32_arg_partial_bytes): Likewise.
20967 (ft32_valid_pointer_mode): Likewise.
20968 (ft32_addr_space_pointer_mode): Likewise.
20969 (ft32_addr_space_legitimate_address_p): Likewise.
20970 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
20972 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
20973 (ix86_emit_outlined_ms2sysv_restore): Likewise.
20974 (iamcu_alignment): Likewise.
20975 (canonicalize_vector_int_perm): Likewise.
20976 (ix86_noce_conversion_profitable_p): Likewise.
20977 (ix86_mpx_bound_mode): Likewise.
20978 (ix86_operands_ok_for_move_multiple): Likewise.
20979 * config/microblaze/microblaze-protos.h
20980 (microblaze_expand_conditional_branch_reg): Likewise.
20981 * config/microblaze/microblaze.c
20982 (microblaze_expand_conditional_branch_reg): Likewise.
20983 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20985 (rs6000_reassociation_width): Likewise.
20986 (rs6000_invalid_binary_op): Likewise.
20987 (fusion_p9_p): Likewise.
20988 (emit_fusion_p9_load): Likewise.
20989 (emit_fusion_p9_store): Likewise.
20990 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
20992 (riscv_hard_regno_mode_ok_p): Likewise.
20993 (riscv_address_insns): Likewise.
20994 (riscv_split_symbol): Likewise.
20995 (riscv_legitimize_move): Likewise.
20996 (riscv_function_value): Likewise.
20997 (riscv_hard_regno_nregs): Likewise.
20998 (riscv_expand_builtin): Likewise.
20999 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21000 (riscv_build_integer): Likewise.
21001 (riscv_split_integer): Likewise.
21002 (riscv_legitimate_constant_p): Likewise.
21003 (riscv_cannot_force_const_mem): Likewise.
21004 (riscv_regno_mode_ok_for_base_p): Likewise.
21005 (riscv_valid_base_register_p): Likewise.
21006 (riscv_valid_offset_p): Likewise.
21007 (riscv_valid_lo_sum_p): Likewise.
21008 (riscv_classify_address): Likewise.
21009 (riscv_legitimate_address_p): Likewise.
21010 (riscv_address_insns): Likewise.
21011 (riscv_load_store_insns): Likewise.
21012 (riscv_force_binary): Likewise.
21013 (riscv_split_symbol): Likewise.
21014 (riscv_force_address): Likewise.
21015 (riscv_legitimize_address): Likewise.
21016 (riscv_move_integer): Likewise.
21017 (riscv_legitimize_const_move): Likewise.
21018 (riscv_legitimize_move): Likewise.
21019 (riscv_address_cost): Likewise.
21020 (riscv_subword): Likewise.
21021 (riscv_output_move): Likewise.
21022 (riscv_canonicalize_int_order_test): Likewise.
21023 (riscv_emit_int_order_test): Likewise.
21024 (riscv_function_arg_boundary): Likewise.
21025 (riscv_pass_mode_in_fpr_p): Likewise.
21026 (riscv_pass_fpr_single): Likewise.
21027 (riscv_pass_fpr_pair): Likewise.
21028 (riscv_get_arg_info): Likewise.
21029 (riscv_function_arg): Likewise.
21030 (riscv_function_arg_advance): Likewise.
21031 (riscv_arg_partial_bytes): Likewise.
21032 (riscv_function_value): Likewise.
21033 (riscv_pass_by_reference): Likewise.
21034 (riscv_setup_incoming_varargs): Likewise.
21035 (riscv_print_operand): Likewise.
21036 (riscv_elf_select_rtx_section): Likewise.
21037 (riscv_save_restore_reg): Likewise.
21038 (riscv_for_each_saved_reg): Likewise.
21039 (riscv_register_move_cost): Likewise.
21040 (riscv_hard_regno_mode_ok_p): Likewise.
21041 (riscv_hard_regno_nregs): Likewise.
21042 (riscv_class_max_nregs): Likewise.
21043 (riscv_memory_move_cost): Likewise.
21044 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
21045 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
21046 (rl78_addr_space_address_mode): Likewise.
21047 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21049 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21050 (rs6000_reassociation_width): Likewise.
21051 (rs6000_invalid_binary_op): Likewise.
21052 (fusion_p9_p): Likewise.
21053 (emit_fusion_p9_load): Likewise.
21054 (emit_fusion_p9_store): Likewise.
21055 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
21056 (ok_for_simple_move_operands): Likewise.
21057 (ok_for_simple_move_strict_operands): Likewise.
21058 (ok_for_simple_arith_logic_operands): Likewise.
21059 (visium_legitimize_reload_address): Likewise.
21060 (visium_select_cc_mode): Likewise.
21061 (output_cbranch): Likewise.
21062 (visium_split_double_move): Likewise.
21063 (visium_expand_copysign): Likewise.
21064 (visium_expand_int_cstore): Likewise.
21065 (visium_expand_fp_cstore): Likewise.
21066 * config/visium/visium.c (visium_pass_by_reference): Likewise.
21067 (visium_function_arg): Likewise.
21068 (visium_function_arg_advance): Likewise.
21069 (visium_libcall_value): Likewise.
21070 (visium_setup_incoming_varargs): Likewise.
21071 (visium_legitimate_constant_p): Likewise.
21072 (visium_legitimate_address_p): Likewise.
21073 (visium_legitimize_address): Likewise.
21074 (visium_secondary_reload): Likewise.
21075 (visium_register_move_cost): Likewise.
21076 (visium_memory_move_cost): Likewise.
21077 (prepare_move_operands): Likewise.
21078 (ok_for_simple_move_operands): Likewise.
21079 (ok_for_simple_move_strict_operands): Likewise.
21080 (ok_for_simple_arith_logic_operands): Likewise.
21081 (visium_function_value_1): Likewise.
21082 (rtx_ok_for_offset_p): Likewise.
21083 (visium_legitimize_reload_address): Likewise.
21084 (visium_split_double_move): Likewise.
21085 (visium_expand_copysign): Likewise.
21086 (visium_expand_int_cstore): Likewise.
21087 (visium_expand_fp_cstore): Likewise.
21088 (visium_split_cstore): Likewise.
21089 (visium_select_cc_mode): Likewise.
21090 (visium_split_cbranch): Likewise.
21091 (output_cbranch): Likewise.
21092 (visium_print_operand_address): Likewise.
21093 * expmed.c (flip_storage_order): Likewise.
21094 * expmed.h (emit_cstore): Likewise.
21095 (flip_storage_order): Likewise.
21096 * genrecog.c (validate_pattern): Likewise.
21097 * hsa-gen.c (gen_hsa_addr): Likewise.
21098 * internal-fn.c (expand_arith_overflow): Likewise.
21099 * ira-color.c (allocno_copy_cost_saving): Likewise.
21100 * lra-assigns.c (find_hard_regno_for_1): Likewise.
21101 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
21102 (process_invariant_for_inheritance): Likewise.
21103 * lra-eliminations.c (move_plus_up): Likewise.
21104 * omp-low.c (lower_oacc_reductions): Likewise.
21105 * simplify-rtx.c (simplify_subreg): Likewise.
21106 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
21107 (TARGET_CHKP_BOUND_MODE): Likewise..
21108 * targhooks.c (default_chkp_bound_mode): Likewise.
21109 (default_setup_incoming_vararg_bounds): Likewise.
21110 * targhooks.h (default_chkp_bound_mode): Likewise.
21111 (default_setup_incoming_vararg_bounds): Likewise.
21112 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
21113 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
21114 (have_whole_vector_shift): Likewise.
21115 * tree-vect-stmts.c (vectorizable_load): Likewise.
21116 * doc/tm.texi: Regenerate.
21118 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21120 Graceful degrade if Binutils PR21472 is not available.
21123 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
21124 .rodata in flash test fails.
21125 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
21126 * confgure: Regenerate.
21127 * config.in: Regenerate.
21128 * config/avr/avr.c (avr_asm_named_section)
21129 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
21130 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
21131 (avr_asm_init_sections): Same.
21133 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21135 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
21136 (fma<VH:mode>4_intrinsic): Likewise.
21137 (*fmsub<VCVTF:mode>4): Likewise.
21138 (*fmsub<VH:mode>4_intrinsic): Likewise.
21140 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21143 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
21144 Don't depend on "optimize > 0".
21145 (out_movhi_r_mr, out_movqi_mr_r): Same.
21146 (out_movhi_mr_r, out_movqi_r_mr): Same.
21147 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
21148 io_address_operand on "optimize > 0".
21150 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21152 * tree-loop-distribution.c: Add general explanantion on the pass.
21153 (generate_loops_for_partition): Mark distributed loop.
21154 (pg_add_dependence_edges): New parameter. Handle alias data
21155 dependence specially and record it in the parameter if asked.
21156 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
21157 (init_partition_graph_vertices, add_partition_graph_edge): New.
21158 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
21159 (free_partition_graph_vdata, build_partition_graph): New.
21160 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
21161 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
21162 (data_ref_segment_size, latch_dominated_by_data_ref): New.
21163 (compute_alias_check_pairs, version_loop_by_alias_check): New.
21164 (version_for_distribution_p, finalize_partitions): New.
21165 (distribute_loop): Handle alias data dependence specially. Factor
21166 out loop fusion code as functions and call these functions.
21168 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21170 * tree-loop-distribution.c (classify_partition): New parameter and
21171 better handle reduction statement.
21172 (rdg_build_partitions): Revise comment.
21173 (distribute_loop): Compute statements in all partitions and pass it
21174 to classify_partition.
21176 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21178 * tree-loop-distribution.c (enum partition_type): New.
21179 (struct partition): New field type.
21180 (partition_merge_into): Add parameter. Update partition type.
21181 (data_dep_in_cycle_p, update_type_for_merge): New functions.
21182 (build_rdg_partition_for_vertex): Compute partition type.
21183 (rdg_build_partitions): Dump partition type.
21184 (distribute_loop): Update calls to partition_merge_into.
21186 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21188 * tree-loop-distribution.c (struct ddr_hasher): New.
21189 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
21191 (classify_partition): Call get_data_dependence.
21192 (pg_add_dependence_edges): Ditto.
21193 (distribute_loop): Release data dependence hash table.
21195 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21197 * tree-loop-distribution.c (ref_base_address): Delete.
21198 (similar_memory_accesses): Rename ...
21199 (share_memory_accesses): ... to this. Check if partitions access
21200 the same memory reference.
21201 (distribute_loop): Call share_memory_accesses.
21203 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21205 * tree-loop-distribution.c (struct partition): New field recording
21206 its data reference.
21207 (partition_alloc, partition_free): Init and release data refs.
21208 (partition_merge_into): Merge data refs.
21209 (build_rdg_partition_for_vertex): Collect data refs for partition.
21210 (pg_add_dependence_edges): Change parameters from vector to bitmap.
21212 (distribute_loop): Remve data refs from vertice data of partition
21215 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21217 * tree-loop-distribution.c (params.h): Include header file.
21218 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
21219 (datarefs_vec): New global var.
21220 (create_rdg_vertices): Use datarefs_vec directly.
21221 (free_rdg): Don't free data references.
21222 (build_rdg): Update use. Don't free data references.
21223 (distribute_loop): Compute global variable for data references.
21224 Bail out if there are too many data references.
21226 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21228 * tree-loop-distribution.c (loop_nest): New global var.
21229 (build_rdg): Use loop directly, rather than loop nest.
21230 (pg_add_dependence_edges): Remove loop nest parameter. Use global
21232 (distribute_loop): Compute global variable loop nest. Update use.
21234 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21236 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
21237 (partition_merge_into): New parameter. Dump reason for fusion.
21238 (distribute_loop): Update use of partition_merge_into.
21240 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21242 * tree-loop-distribution.c (bb_top_order_index): New.
21243 (bb_top_order_index_size, bb_top_order_cmp): New.
21244 (stmts_from_loop): Use topological order.
21245 (pass_loop_distribution::execute): Compute and release topological
21246 order for basic blocks.
21248 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21250 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
21253 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21255 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
21256 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
21257 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
21258 * internal-fn.def (LOOP_DIST_ALIAS): New.
21259 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
21260 (fold_loop_internal_call): ... this.
21261 (vect_loop_dist_alias_call): New function.
21262 (set_uid_loop_bbs): Call fold_loop_internal_call.
21263 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
21266 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21269 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
21270 Require dead FLAGS_REG at the beginning of a peephole.
21272 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21275 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
21276 arguments in the call to __builtin_ia32_sbb_u32.
21277 (_subborrow_u64): Swap _X and _Y arguments in the call to
21278 __builtin_ia32_sbb_u64.
21280 2017-07-04 Jakub Jelinek <jakub@redhat.com>
21283 * tree-vrp.c (compare_assert_loc): Turn into a function template
21284 with stable template parameter. Only test if a->e is NULL,
21285 !a->e == !b->e has been verified already. Use e == NULL or
21286 e != NULL instead of e or ! e tests. If stable is true, don't use
21287 iterative_hash_expr, on the other side allow a or b or both NULL
21288 and sort the NULLs last.
21289 (process_assert_insertions): Sort using compare_assert_loc<false>
21290 instead of compare_assert_loc, later sort using
21291 compare_assert_loc<true> before calling process_assert_insertions_for
21292 in a loop. Use break instead of continue once seen NULL pointer.
21294 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21296 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21297 Cortex-R7 and Cortex-R8 processors.
21299 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21301 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
21302 uninitialized while src is not.
21304 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
21306 * common/config/arm/arm-common.c: Adjust include path for
21308 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
21309 (arm-cpu.h): Create in build directory. Adjust dependency rules.
21310 (arm-cpu-data.h): Likewise.
21311 (arm-cpu-cdata.h): Likewise.
21312 * config/arm/arm-cpu.h: Delete.
21313 * config/arm/arm-cpu-cdata.h: Delete.
21314 * config/arm/arm-cpu-data.h: Delete.
21316 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
21318 * config/arm/arm-cpus.in (cortex-a55): New.
21319 (cortex-a75): Likewise.
21320 (cortex-a75.cortex-a55): Likewise.
21321 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
21323 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
21324 * config/arm/arm-cpu-cdata.h: Regenerate.
21325 * config/arm/arm-cpu-data.h: Regenerate.
21326 * config/arm/arm-cpu.h: Regenerate.
21327 * config/arm/arm-tables.opt: Regenerate.
21328 * config/arm/arm-tune.md: Regenerate.
21330 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21332 * haifa-sched.c (sched_create_recovery_edges): Update profile.
21334 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21336 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
21339 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
21341 PR tree-optimization/81292
21342 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
21343 full_string_p, also call adjust_related_strinfos if the adjustment
21344 is simple, otherwise invalidate related strinfos.
21346 2017-07-04 Martin Liska <mliska@suse.cz>
21349 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
21350 newly created variable as DECL_IGNORED_P.
21352 2017-07-04 Martin Liska <mliska@suse.cz>
21355 * ipa-inline.c (inline_small_functions):
21356 Use xstrdup_for_dump.
21358 2017-07-04 Tom de Vries <tom@codesourcery.com>
21360 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
21362 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21365 * config/darwin.c (darwin_function_switched_text_sections):
21368 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
21370 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
21372 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21374 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
21376 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21378 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
21379 min_profitable_iters, and th as inclusive lower bounds.
21380 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
21381 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
21382 for min_profitable_iters and min_profitable_estimate.
21383 (vect_transform_loop): Treat th as an inclusive lower bound.
21384 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21386 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21389 * config/darwin.c (darwin_function_switched_text_sections):
21390 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
21391 in two pieces, and suppress the use of buf.
21393 2017-07-03 Nathan Sidwell <nathan@acm.org>
21395 * hash-table.h (hash_table_mod1): Fix indentation.
21397 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21399 PR middle-end/81290
21400 * predict.c (force_edge_cold): Be more careful about propagation
21402 * profile-count.h (profile_probability::guessed,
21403 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
21405 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
21407 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
21409 * doc/invoke.texi (rcpc architecture extension): Document it.
21411 2017-07-03 Richard Biener <rguenther@suse.de>
21413 PR tree-optimization/60510
21414 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
21415 the scalar reduction PHI and use it.
21416 (vectorizable_reduction): Properly guard the single_defuse_cycle
21417 path for non-SLP reduction chains where we cannot use it.
21418 Rework reduc_def/index and vector type deduction. Rework
21419 vector operand gathering during reduction op code-gen.
21420 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
21421 chains dissolve the chain and leave it to non-SLP reduction
21424 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21426 * tree-data-ref.h (dr_alignment): Declare.
21427 * tree-data-ref.c (dr_alignment): New function.
21428 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
21429 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
21431 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
21433 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21435 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
21436 and base_misalignment fields.
21437 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
21438 * tree-data-ref.c: Include builtins.h.
21439 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
21440 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
21441 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
21442 * tree-vect-data-refs.c: Include tree-cfg.h.
21443 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
21444 fields instead of calculating an alignment here.
21445 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
21446 innermost_loop_behavior fields.
21448 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21450 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
21452 (DR_STEP_ALIGNMENT): New macro.
21453 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
21454 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
21455 (create_data_ref): Print it.
21456 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
21457 to tell whether the step preserves vector (mis)alignment.
21458 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21459 Move the check for an integer step and generalise to all INTEGER_CST.
21460 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
21461 Print the outer step alignment.
21463 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21465 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
21466 with offset_alignment.
21467 (DR_ALIGNED_TO): Delete.
21468 (DR_OFFSET_ALIGNMENT): New macro.
21469 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
21470 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
21471 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
21472 (create_data_ref): Likewise.
21473 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21474 (vect_analyze_data_refs): Likewise.
21475 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
21476 creating dummy innermost behavior.
21478 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21480 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
21481 with a "innermost_loop_behavior *" and refeence tree.
21482 * tree-data-ref.c (dr_analyze_innermost): Likewise.
21483 (create_data_ref): Update call accordingly.
21484 * tree-predcom.c (find_looparound_phi): Likewise.
21486 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21488 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
21489 fields with dr_wrt_vec_loop.
21490 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
21491 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
21492 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
21493 (vect_dr_behavior): New function.
21494 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21495 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
21496 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
21497 track whether the step preserves the misalignment.
21498 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21499 Use vect_dr_behavior.
21500 (vect_setup_realignment): Update call accordingly.
21501 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
21502 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
21503 call to vect_create_addr_base_for_vector_ref.
21504 (vect_create_cond_for_align_checks): Likewise.
21505 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
21506 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
21507 (vect_recog_mask_conversion_pattern): Likewise.
21508 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
21509 (new_stmt_vec_info): Remove redundant zeroing.
21511 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21513 * common/config/arm/arm-common.c (arm_be8_option): New function.
21514 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
21515 (ISA_ARMv6): Add isa_bit_be8.
21516 * config/arm/arm.h (arm_be8_option): Add prototype.
21517 (BE8_SPEC_FUNCTION): New define.
21518 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
21519 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
21520 (mlittle-endian): Similarly.
21521 (mbe8, mbe32): New options.
21522 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
21523 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
21525 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21527 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
21529 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21531 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
21532 (cleanup_tree_cfg_bb): Use it.
21533 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
21535 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
21537 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21540 * loop-doloop.c (add_test): Update profile.
21542 2017-07-03 Martin Liska <mliska@suse.cz>
21545 * sanopt.c (rewrite_usage_of_param): New function.
21546 (sanitize_rewrite_addressable_params): Likewise.
21547 (pass_sanopt::execute): Call rewrite_usage_of_param.
21549 2017-07-03 Richard Biener <rguenther@suse.de>
21551 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
21552 back to using VIEW_CONVERT_EXPR.
21554 2017-07-03 Martin Liska <mliska@suse.cz>
21557 * doc/extend.texi: Document when a resolver function is
21558 generated for target_clones.
21560 2017-07-03 Martin Liska <mliska@suse.cz>
21562 * asan.c (asan_emit_stack_protection): Unpoison just red zones
21563 and shadow memory of auto variables which are subject of
21564 use-after-scope sanitization.
21565 (asan_expand_mark_ifn): Add do set only when is_poison.
21567 2016-07-03 Richard Biener <rguenther@suse.de>
21569 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
21571 (vect_force_simple_reduction): Record reduction def -> phi mapping.
21572 (vectorizable_reduction): Perform reduction PHI creation when
21573 visiting a reduction PHI and adjust and simplify code generation
21574 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
21575 (vect_transform_loop): Visit reduction PHIs.
21576 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
21577 defs into the SLP tree.
21578 (vect_build_slp_tree): Reduction defs terminate the recursion.
21579 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
21581 (vect_get_vec_defs_for_stmt_copy): Export.
21582 (vect_get_vec_defs): Likewise.
21583 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
21585 (vect_get_vec_defs_for_stmt_copy): Declare.
21586 (vect_get_vec_defs): Likewise.
21588 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21590 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
21591 parameter with a "loop" parameter and use it instead of the
21592 loop containing DR_STMT. Don't check simple_iv when doing
21593 BB analysis. Describe the two analysis modes in the comment.
21595 2017-07-03 Tom de Vries <tom@codesourcery.com>
21597 PR tree-optimization/69468
21598 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
21599 (find_same_succ_bb): Handle ignore_edge_flags.
21601 2017-07-03 Tom de Vries <tom@codesourcery.com>
21603 PR tree-optimization/81192
21604 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
21606 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
21608 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
21610 2017-07-03 Tom de Vries <tom@codesourcery.com>
21612 PR tree-optimization/81192
21613 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
21614 BB_SAME_SUCC (bb) == NULL.
21616 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21618 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
21621 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21623 * dumpfile.c: Include profile-count.h
21624 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
21626 (insert_cond_bb): Update profile.
21627 * tree-cfg.h (insert_cond_bb): Update prototype.
21628 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
21629 * tree-dump.c: Do not include tree-cfg.
21631 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21633 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
21635 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21637 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
21640 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21642 * tree-complex.c (expand_complex_div_wide): update profile.
21644 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21645 Alan Hayward <alan.hayward@arm.com>
21646 David Sherwood <david.sherwood@arm.com>
21648 * Makefile.in (MACHMODE_H): Remove insn-modes.h
21649 (CORETYPES_H): New define.
21650 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
21651 (insn-modes-inline.h, s-modes-inline-h): New rules.
21652 (generated_files): Add insn-modes-inline.h.
21653 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
21654 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
21655 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
21656 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
21657 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
21658 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
21659 (build/gencodes.o, build/genconditions.o): Likewise.
21660 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
21661 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
21662 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
21663 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
21664 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
21665 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
21666 * coretypes.h: Include everything up to real.h for generators.
21667 Include insn-modes.h first. Include wide-int-print.h after
21668 wide-int.h. Include insn-modes-inline.h and then machmode.h.
21669 * machmode.h: Don't include insn-modes.h here.
21670 * function-tests.c: Remove includes of signop.h, machmode.h,
21671 double-int.h and wide-int.h.
21673 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
21675 * optc-save-gen.awk: Likewise.
21676 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
21677 * godump.c: Remove include of wide-int-print.h.
21678 * pretty-print.h: Likewise.
21679 * wide-int-print.cc: Likewise.
21680 * wide-int.cc: Likewise.
21681 * hash-map-tests.c: Remove include of signop.h.
21682 * hash-set-tests.c: Likewise.
21683 * rtl-tests.c: Likewise.
21684 * mkconfig.sh: Remove include of machmode.h.
21685 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
21687 (emit_insn_modes_inline_h): ...this new function. Emit the code
21688 into an insn-modes-inline.h header file, adding appropriate
21689 include guards and end comments.
21690 (emit_insn_modes_c_header): Remove include of machmode.h.
21691 (emit_min_insn_modes_c_header): Include coretypes.h rather than
21693 (main): Handle -i flag and call emit_insn_modes_inline_h when
21696 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21698 * tree-ssa-strlen.c (strinfo): Rename the length field to
21699 nonzero_chars. Add a full_string_p field.
21700 (compare_nonzero_chars, zero_length_string_p): New functions.
21701 (get_addr_stridx): Add an offset_out parameter.
21702 Use compare_nonzero_chars.
21703 (get_stridx): Update accordingly. Use compare_nonzero_chars.
21704 (new_strinfo): Update after above changes to strinfo.
21705 (set_endptr_and_length): Set full_string_p.
21706 (get_string_length): Update after above changes to strinfo.
21707 (unshare_strinfo): Update call to new_strinfo.
21708 (maybe_invalidate): Likewise.
21709 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
21710 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
21711 as a uhwi instead of an shwi. Update after above changes to
21712 strinfo and new_strinfo.
21713 (zero_length_string): Assert that chainsi contains full strings.
21714 Use zero_length_string_p. Update call to new_strinfo.
21715 (adjust_related_strinfos): Update after above changes to strinfo.
21716 Copy full_string_p from origsi.
21717 (adjust_last_stmt): Use zero_length_string_p.
21718 (handle_builtin_strlen): Update after above changes to strinfo and
21719 new_strinfo. Install the lhs as the string length if the previous
21720 entry didn't describe a full string.
21721 (handle_builtin_strchr): Update after above changes to strinfo
21723 (handle_builtin_strcpy): Likewise.
21724 (handle_builtin_strcat): Likewise.
21725 (handle_builtin_malloc): Likewise.
21726 (handle_pointer_plus): Likewise.
21727 (handle_builtin_memcpy): Likewise. Track nonzero characters
21728 that aren't necessarily followed by a nul terminator.
21729 (handle_char_store): Likewise.
21731 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21733 PR tree-optimization/80769
21734 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
21735 for malloc and calloc. Document the new invariant that all related
21736 strinfos have delayed lengths or none do.
21737 (verify_related_strinfos): Move earlier in file.
21738 (set_endptr_and_length): New function, split out from...
21739 (get_string_length): ...here. Also set the lengths of related
21741 (zero_length_string): Assert that chainsi has known (rather than
21743 (adjust_related_strinfos): Likewise.
21745 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21747 PR tree-optimization/81136
21748 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
21749 assert that two references with the same misalignment have the same
21750 compile-time misalignment if those compile-time misalignments
21753 2017-07-01 Andi Kleen <ak@linux.intel.com>
21755 * print-tree.c (print_node): Print all attributes.
21757 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21759 * cfg.c (scale_bbs_frequencies): New function.
21760 * cfg.h (scale_bbs_frequencies): Declare it.
21761 * cfgloopanal.c (single_likely_exit): Cleanup.
21762 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
21764 (scale_loop_profile): Likewise.
21765 (loop_version): Likewise.
21766 (create_empty_loop_on_edge): Update.
21767 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
21768 scale_loop_frequencies, scale_loop_profile, loopify,
21769 loop_version): Update prototypes.
21770 * modulo-sched.c (sms_schedule): Update.
21771 * predict.c (unlikely_executed_edge_p): Also check probability.
21772 (probably_never_executed_edge_p): Fix typo.
21773 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21774 * tree-parloops.c (gen_parallel_loop): Update.
21775 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
21776 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21777 * tree-ssa-loop-split.c (split_loop): Update.
21778 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21779 * tree-vect-loop-manip.c (vect_do_peeling): Update.
21780 (vect_loop_versioning): Update.
21781 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21783 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21785 * trans-mem.c (split_bb_make_tm_edge): Update profile.
21787 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21789 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
21790 to keep profile consistent.
21792 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21794 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
21795 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
21796 * profile-count.h (max_safe_multiplier): Make unsigned.
21797 (profile_count::guessed_zero): New.
21799 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21801 * bb-reorder.c (fix_up_crossing_landing_pad,
21802 fix_crossing_conditional_branches): Use make_single_succ_edge
21803 to keep profile consistent.
21805 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21807 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
21810 2017-07-01 Jakub Jelinek <jakub@redhat.com>
21813 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
21814 the right scopes, make sure cond_jump isn't preserved between multiple
21815 iterations. Search for fallthru edge whenever there are 3+ edges and
21816 use find_fallthru_edge for it.
21818 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21820 Patch by Alexander Monakov <amonakov@ispras.ru>
21821 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
21822 probabilities consistently.
21824 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21826 * pa.c (pa_expand_compare_and_swap_loop): Update call of
21827 emit_cmp_and_jump_insns.
21829 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21832 * tree-inline.c (expand_call_inline): Combine profile statuses.
21834 2017-06-30 Andrew Pinski <apinski@cavium.com>
21836 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
21837 fold_stmt returned true.
21839 2017-06-30 Nathan Sidwell <nathan@acm.org>
21841 * ggc.h (empty_string): Delete.
21842 * cfgexpand.c (expand_asm_stmt): Use plain "".
21843 * optabs.c (expand_asm_memory_barrier): Likewise.
21844 * stringpool.c (empty_string): Delete.
21845 (digit_vector, digit_string): Delete.
21846 (ggc_alloc_string): Use plain "", don't optimize single digit
21847 strings. Use ggc_alloc_atomic.
21849 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
21851 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
21852 comparison set and one other set, use the cost of the non-comparison
21855 2017-06-30 Nathan Sidwell <nathan@acm.org>
21857 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
21860 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
21862 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
21863 loops. Remove now unneeded calls to gimple_switch_set_label() that
21864 just set removed labels to NULL_TREE.
21866 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
21868 * tree-ssanames.c (set_range_info_raw): Abstract from ...
21869 (set_range_info): ...here. Only call set_range_info_raw if domain
21871 (set_nonzero_bits): Call set_range_info_raw.
21872 * tree-ssanames.h (set_range_info_raw): New.
21874 2017-06-30 Jakub Jelinek <jakub@redhat.com>
21877 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
21878 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
21879 of nonimmediate_operand and <store_mask_constraint> instead of m
21880 for the input operand. For V8FI iterator, always split if input
21881 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
21882 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
21883 <store_mask_predicate> instead of register_operand and
21884 <store_mask_constraint> instead of v for the input operand. Make
21885 sure both operands aren't MEMs for if not <mask_applied>.
21887 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
21889 * lto-wrapper.c (copy_file) Close both file descriptors before
21892 2017-06-30 Martin Liska <mliska@suse.cz>
21895 * multiple_target.c (create_dispatcher_calls): Make ifunc
21896 also for function that don't have calls or are not referenced.
21898 2017-06-30 Richard Biener <rguenther@suse.de>
21900 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
21901 analyze the first scalar stmt. Move vector type computation
21902 for the BB case here from ...
21903 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
21904 live operation processing in the SLP case properly.
21906 2017-06-30 Richard Biener <rguenther@suse.de>
21908 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
21910 2017-06-30 Martin Liska <mliska@suse.cz>
21913 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
21914 before BUILT_IN_UNWIND_RESUME when ASAN is used.
21916 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
21918 * doc/invoke.texi (AArch64): Add missing options and remove redundant
21921 2017-06-30 Richard Biener <rguenther@suse.de>
21923 PR tree-optimization/81249
21924 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
21925 condition reduction result to original scalar type.
21927 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21929 * profile-count.h (enum profile_quality): Fix typos and whitespace
21932 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21934 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
21935 type for branch probabilities.
21937 2017-06-29 Julian Brown <julian@codesourcery.com>
21938 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21940 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
21941 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
21942 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
21943 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
21945 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21947 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
21948 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
21949 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
21950 CC usage from generic code to here.
21951 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
21952 CC usage into the target macros.
21954 2017-06-29 Maya Rashish <coypu@sdf.org>
21956 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
21959 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21961 * arm/arm-builtins.c: Include profile-count.h
21962 * except.c (sjlj_emit_function_enter): Use
21963 profile_probability::unlikely.
21965 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21967 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
21968 and tocrel_offset be pointer args rather than implicitly using
21970 (legitimate_constant_pool_address_p, rs6000_emit_move,
21971 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
21972 tocrel_offset and use in toc_relative_expr_p call.
21973 (print_operand, print_operand_address): Use static tocrel_base_oac
21974 and tocrel_offset_oac.
21975 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
21978 2017-06-29 Maya Rashish <coypu@sdf.org>
21980 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
21982 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
21984 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
21985 objects, take into account only the alignment of 'op0' and 'mode1' if
21988 2017-06-29 Steve Ellcey <sellcey@cavium.com>
21990 * ccmp.c (ccmp_tree_comparison_p): New function.
21991 (ccmp_candidate_p): Update to use above function.
21992 (get_compare_parts): New function.
21993 (expand_ccmp_next): Update to use new functions.
21994 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
21996 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
21997 take mode as argument.
21998 * ccmp.h (expand_ccmp_expr): Add mode as argument.
21999 * expr.c (expand_expr_real_1): Pass mode as argument.
22001 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22003 * combine.c (combine_instructions): Print insns to dump_file, together
22006 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22008 * asan.c (asan_emit_stack_protection): Update.
22009 (create_cond_insert_point): Update.
22010 * auto-profile.c (afdo_propagate_circuit): Update.
22011 * basic-block.h (struct edge_def): Turn probability to
22012 profile_probability.
22013 (EDGE_FREQUENCY): Update.
22014 * bb-reorder.c (find_traces_1_round): Update.
22015 (better_edge_p): Update.
22016 (sanitize_hot_paths): Update.
22017 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22018 (make_single_succ_edge): Update.
22019 (check_bb_profile): Update.
22020 (dump_edge_info): Update.
22021 (update_bb_profile_for_threading): Update.
22022 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
22023 probabilitycount to 0.
22024 * cfgbuild.c (compute_outgoing_frequencies): Update.
22025 * cfgcleanup.c (try_forward_edges): Update.
22026 (outgoing_edges_match): Update.
22027 (try_crossjump_to_edge): Update.
22028 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
22029 (expand_gimple_tailcall): Update.
22030 (construct_init_block): Use make_single_succ_edge.
22031 (construct_exit_block): Use make_single_succ_edge.
22032 * cfghooks.c (verify_flow_info): Update.
22033 (redirect_edge_succ_nodup): Update.
22034 (split_edge): Update.
22035 (account_profile_record): Update.
22036 * cfgloopanal.c (single_likely_exit): Update.
22037 * cfgloopmanip.c (scale_loop_profile): Update.
22038 (set_zero_probability): Remove.
22039 (duplicate_loop_to_header_edge): Update.
22040 * cfgloopmanip.h (loop_version): Update prototype.
22041 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
22042 (force_nonfallthru_and_redirect): Update.
22043 (update_br_prob_note): Update.
22044 (rtl_verify_edges): Update.
22045 (purge_dead_edges): Update.
22046 (rtl_lv_add_condition_to_bb): Update.
22047 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
22048 * cgraphunit.c (init_lowered_empty_function): Update.
22049 (cgraph_node::expand_thunk): Update.
22050 * cilk-common.c: Include profile-count.h
22051 * dojump.c (inv): Remove.
22052 (jumpifnot): Update.
22053 (jumpifnot_1): Update.
22054 (do_jump_1): Update.
22056 (do_jump_by_parts_greater_rtx): Update.
22057 (do_compare_rtx_and_jump): Update.
22058 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
22059 do_jump_1. do_compare_rtx_and_jump): Update prototype.
22060 * dwarf2cfi.c: Include profile-count.h
22061 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
22062 (sjlj_emit_dispatch_table): Likewise.
22063 * explow.c: Include profile-count.h
22064 * expmed.c (emit_store_flag_force): Update.
22065 (do_cmp_and_jump): Update.
22066 * expr.c (compare_by_pieces_d::generate): Update.
22067 (compare_by_pieces_d::finish_mode): Update.
22068 (emit_block_move_via_loop): Update.
22069 (store_expr_with_bounds): Update.
22070 (store_constructor): Update.
22071 (expand_expr_real_2): Update.
22072 (expand_expr_real_1): Update.
22073 * expr.h (try_casesi, try_tablejump): Update prototypes.
22074 * gimple-pretty-print.c (dump_probability): Update.
22075 (dump_profile): New.
22076 (dump_gimple_label): Update.
22077 (dump_gimple_bb_header): Update.
22078 * graph.c (draw_cfg_node_succ_edges): Update.
22079 * hsa-gen.c (convert_switch_statements): Update.
22080 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
22081 (find_if_case_1): Update.
22082 (find_if_case_2): Update.
22083 * internal-fn.c (expand_arith_overflow_result_store): Update.
22084 (expand_addsub_overflow): Update.
22085 (expand_neg_overflow): Update.
22086 (expand_mul_overflow): Update.
22087 (expand_vector_ubsan_overflow): Update.
22088 * ipa-cp.c (good_cloning_opportunity_p): Update.
22089 * ipa-split.c (split_function): Use make_single_succ_edge.
22090 * ipa-utils.c (ipa_merge_profiles): Update.
22091 * loop-doloop.c (add_test): Update.
22092 (doloop_modify): Update.
22093 * loop-unroll.c (compare_and_jump_seq): Update.
22094 (unroll_loop_runtime_iterations): Update.
22095 * lra-constraints.c (lra_inheritance): Update.
22096 * lto-streamer-in.c (input_cfg): Update.
22097 * lto-streamer-out.c (output_cfg): Update.
22098 * mcf.c (adjust_cfg_counts): Update.
22099 * modulo-sched.c (sms_schedule): Update.
22100 * omp-expand.c (expand_omp_for_init_counts): Update.
22101 (extract_omp_for_update_vars): Update.
22102 (expand_omp_ordered_sink): Update.
22103 (expand_omp_for_ordered_loops): Update.
22104 (expand_omp_for_generic): Update.
22105 (expand_omp_for_static_nochunk): Update.
22106 (expand_omp_for_static_chunk): Update.
22107 (expand_cilk_for): Update.
22108 (expand_omp_simd): Update.
22109 (expand_omp_taskloop_for_outer): Update.
22110 (expand_omp_taskloop_for_inner): Update.
22111 * omp-simd-clone.c (simd_clone_adjust): Update.
22112 * optabs.c (expand_doubleword_shift): Update.
22113 (expand_abs): Update.
22114 (emit_cmp_and_jump_insn_1): Update.
22115 (expand_compare_and_swap_loop): Update.
22116 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
22117 * predict.c (predictable_edge_p): Update.
22118 (edge_probability_reliable_p): Update.
22119 (set_even_probabilities): Update.
22120 (combine_predictions_for_insn): Update.
22121 (combine_predictions_for_bb): Update.
22122 (propagate_freq): Update.
22123 (estimate_bb_frequencies): Update.
22124 (force_edge_cold): Update.
22125 * profile-count.c (profile_count::dump): Add missing space into dump.
22126 (profile_count::debug): Add newline.
22127 (profile_count::differs_from_p): Explicitly convert to unsigned.
22128 (profile_count::stream_in): Update.
22129 (profile_probability::dump): New member function.
22130 (profile_probability::debug): New member function.
22131 (profile_probability::differs_from_p): New member function.
22132 (profile_probability::differs_lot_from_p): New member function.
22133 (profile_probability::stream_in): New member function.
22134 (profile_probability::stream_out): New member function.
22135 * profile-count.h (profile_count_quality): Rename to ...
22136 (profile_quality): ... this one.
22137 (profile_probability): New.
22138 (profile_count): Update.
22139 * profile.c (compute_branch_probabilities): Update.
22140 * recog.c (peep2_attempt): Update.
22141 * sched-ebb.c (schedule_ebbs): Update.
22142 * sched-rgn.c (find_single_block_region): Update.
22143 (compute_dom_prob_ps): Update.
22144 (schedule_region): Update.
22145 * sel-sched-ir.c (compute_succs_info): Update.
22146 * stmt.c (struct case_node): Update.
22147 (do_jump_if_equal): Update.
22148 (get_outgoing_edge_probs): Update.
22149 (conditional_probability): Update.
22150 (emit_case_dispatch_table): Update.
22151 (expand_case): Update.
22152 (expand_sjlj_dispatch_table): Update.
22153 (emit_case_nodes): Update.
22154 * targhooks.c: Update.
22155 * tracer.c (better_p): Update.
22156 (find_best_successor): Update.
22157 * trans-mem.c (expand_transaction): Update.
22158 * tree-call-cdce.c: Update.
22159 * tree-cfg.c (gimple_split_edge): Upate.
22160 (move_sese_region_to_fn): Upate.
22161 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
22162 * tree-eh.c (lower_resx): Upate.
22163 (cleanup_empty_eh_move_lp): Upate.
22164 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22165 * tree-inline.c (copy_edges_for_bb): Update.
22166 (copy_cfg_body): Update.
22167 * tree-parloops.c (gen_parallel_loop): Update.
22168 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
22169 (gimple_gen_time_profiler): Update.
22170 * tree-ssa-dce.c (remove_dead_stmt): Update.
22171 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22172 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
22173 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
22174 (unloop_loops): Update.
22175 (try_peel_loop): Update.
22176 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22177 * tree-ssa-loop-split.c (connect_loops): Update.
22178 (split_loop): Update.
22179 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22180 (hoist_guard): Update.
22181 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
22182 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
22183 (value_replacement): Update.
22184 * tree-ssa-reassoc.c (branch_fixup): Update.
22185 * tree-ssa-tail-merge.c (replace_block_by): Update.
22186 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
22187 (create_edge_and_update_destination_phis): Update.
22188 (compute_path_counts): Update.
22189 (recompute_probabilities): Update.
22190 (update_joiner_offpath_counts): Update.
22191 (freqs_to_counts_path): Update.
22192 (duplicate_thread_path): Update.
22193 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
22194 (struct switch_conv_info): Update.
22195 (gen_inbound_check): Update.
22196 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22197 (vect_do_peeling): Update.
22198 (vect_loop_versioning): Update.
22199 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22200 (optimize_mask_stores): Update.
22201 * ubsan.c (ubsan_expand_null_ifn): Update.
22202 * value-prof.c (gimple_divmod_fixed_value): Update.
22203 (gimple_divmod_fixed_value_transform): Update.
22204 (gimple_mod_pow2): Update.
22205 (gimple_mod_pow2_value_transform): Update.
22206 (gimple_mod_subtract): Update.
22207 (gimple_mod_subtract_transform): Update.
22208 (gimple_ic): Update.
22209 (gimple_stringop_fixed_value): Update.
22210 (gimple_stringops_transform): Update.
22211 * value-prof.h: Update.
22213 2017-06-29 Carl Love <cel@us.ibm.com>
22215 * config/rs6000/rs6000-c.c: Add support for built-in functions
22216 vector signed int vec_signed (vector float);
22217 vector signed long long vec_signed (vector double);
22218 vector signed int vec_signed2 (vector double, vector double);
22219 vector signed int vec_signede (vector double);
22220 vector signed int vec_signedo (vector double);
22221 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
22222 instruction generator.
22223 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22224 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
22225 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
22227 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
22228 vunsignede_v2df): Add define_expands.
22229 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
22230 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
22231 VEC_UNSIGNEDO): Add definitions.
22232 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22233 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
22234 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
22235 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
22236 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
22237 * config/rs6000/altivec.h (vec_signed, vec_signed2,
22238 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
22239 vec_unsignede, vec_unsignedo): Add builtin defines.
22240 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
22242 * doc/extend.texi: Update the built-in documentation file for the
22243 new built-in functions.
22245 2017-06-29 Richard Biener <rguenther@suse.de>
22247 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
22248 reduction chains to LOOP_VINFO_REDUCTIONS.
22249 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
22250 SLP reductions after processing reduction chains.
22252 2017-06-29 Nathan Sidwell <nathan@acm.org>
22254 * builtins.c (fold_builtin_FUNCTION): Use
22255 lang_hooks.decl_printable_name.
22257 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
22259 PR middle-end/81194
22260 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
22261 with only one label.
22262 * stmt.c (expand_case): Assert NCASES is greater than one.
22264 2017-06-29 Richard Biener <rguenther@suse.de>
22266 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
22268 (group_case_labels): Likewise.
22269 (find_taken_edge): Push sanity checking on val to workers...
22270 (find_taken_edge_cond_expr): ... here
22271 (find_taken_edge_switch_expr): ... and here, handle cases
22272 with just a default label.
22273 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
22274 (group_case_labels): Likewise.
22275 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
22276 group_case_labels does anything cleanup the CFG again.
22278 2017-06-29 Bin Cheng <bin.cheng@arm.com>
22280 PR tree-optimization/81196
22281 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
22282 exit condition comparing two IVs.
22284 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
22286 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
22287 profile to the dummy entry at the end of the list of architectures.
22288 * config/arm/arm-cpu-cdata.h: Regenerated.
22290 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22291 Michael Collison <michael.collison@arm.com>
22294 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
22296 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
22297 (*aarch64_reg_<mode>3_minus_mask): New pattern.
22298 (*aarch64_<optab>_reg_di3_mask2): New pattern.
22299 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
22300 of shift when the shift amount is masked with constant equal to
22301 the size of the mode.
22302 * config/aarch64/predicates.md (subreg_lowpart_operator): New
22305 2017-06-29 Martin Liska <mliska@suse.cz>
22307 * config/i386/i386.opt: Change range from [1,5] to [0,5].
22309 2017-06-29 Yury Gribov <tetra2005@gmail.com>
22312 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
22315 (edge_growth_cache_entry::edge_growth_cache_entry): New
22317 (reset_edge_growth_cache): Update to use constructor.
22319 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22321 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
22322 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22323 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22325 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
22327 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
22328 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
22330 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
22332 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
22333 (*-linux-uclibc*): Add t-uclibc tmake_file.
22334 * config/t-musl: New.
22335 * config/t-uclibc: New.
22337 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
22339 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
22341 (gen_comm_data): Emit architectural setting of arch_prof.
22342 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
22344 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
22345 (armv8-m.base, armv8-m.main): Likewise.
22346 * arm-protos.h (arm_build_target): Add profile field.
22347 (arch_option): Likewise.
22348 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
22350 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
22351 arm_active_target.profile.
22353 2017-06-28 Richard Biener <rguenther@suse.de>
22355 PR middle-end/81227
22356 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
22357 TYPE_OVERFLOW_WRAPS.
22358 * match.pd (negate_expr_p): Likewise.
22359 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
22360 fold_build2, not fold_binary.
22362 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22364 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
22365 Convert memory address to Pmode.
22366 (aarch64_print_operand): Assert MEM operands are always Pmode.
22368 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22371 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
22372 Remove redundant if.
22373 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
22374 * config/arm/aarch-common-protos.h
22375 (aarch_forward_to_shift_is_not_shifted_re): Remove.
22376 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
22378 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
22381 * multiple_target.c (create_dispatcher_calls): Set the default
22382 clone to be static, not public.
22384 2017-06-28 Richard Biener <rguenther@suse.de>
22386 * tree-vect-loop.c (vectorizable_reduction): Move special
22387 cond reduction IV var creation ...
22388 (vect_create_epilog_for_reduction): ... here. Remove induction_index
22389 parameter. Use STMT_VINFO_VECTYPE.
22390 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
22393 2017-06-28 Martin Liska <mliska@suse.cz>
22396 * ipa-visibility.c (non_local_p): Handle visibility.
22398 2017-06-28 Martin Liska <mliska@suse.cz>
22401 * common.opt: Add IntegerRange to various options.
22402 * opt-functions.awk (integer_range_info): New function.
22403 * optc-gen.awk: Add integer_range_info to cl_options struct.
22404 * opts-common.c (decode_cmdline_option): Handle
22405 CL_ERR_INT_RANGE_ARG.
22406 (cmdline_handle_error): Likewise.
22407 * opts.c (print_filtered_help): Show valid interval in
22408 when --help is provided.
22409 * opts.h (struct cl_option): Add range_min and range_max fields.
22410 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
22412 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
22414 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
22415 (x * C EQ/NE y * C): New transformation.
22417 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
22419 * genmultilib (combination_space): Accept '+' in option names.
22421 2017-06-28 Martin Liska <mliska@suse.cz>
22424 * asan.c (instrument_derefs): Bail out inner references
22425 that are hard register variables.
22427 2017-06-28 Jakub Jelinek <jakub@redhat.com>
22430 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
22431 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
22433 2017-06-28 Richard Biener <rguenther@suse.de>
22435 * tree-vectorizer.h (vect_get_vec_defs): Remove.
22436 (vect_get_slp_defs): Adjust.
22437 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
22439 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
22441 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
22442 get_initial_defs_for_reduction instead of vect_get_vec_defs.
22443 (vectorizable_reduction): Adjust.
22444 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
22446 (vect_get_slp_defs): Likewise.
22447 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
22448 (vectorizable_bswap): Adjust.
22449 (vectorizable_call): Likewise.
22450 (vectorizable_conversion): Likewise.
22451 (vectorizable_assignment): Likewise.
22452 (vectorizable_shift): Likewise.
22453 (vectorizable_operation): Likewise.
22454 (vectorizable_store): Likewise.
22455 (vectorizable_condition): Likewise.
22456 (vectorizable_comparison): Likewise.
22458 2017-06-28 Michael Collison <michael.collison@arm.com>
22461 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
22463 (arm_gen_movmemqi): Removed unused variable 'i'.
22464 Convert 'for' loop into 'while' loop.
22465 (arm_expand_prologue): Remove last unnecessary set of insn.
22466 (thumb_pop): Remove unused variable 'pushed_words'.
22467 (thumb_exit): Remove last unnecessary set of regs_to_pop.
22469 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22471 * config/s390/predicates.md: Use s390_rel_address_ok_p.
22472 * config/s390/s390-protos.h: Add prototype of
22473 s390_rel_address_ok_p.
22474 * config/s390/s390.c (s390_got_symbol): New function.
22475 (s390_rel_address_ok_p): New function.
22476 (legitimize_pic_address): Use s390_rel_address_ok_p.
22477 (s390_load_got): Use s390_got_symbol.
22478 (s390_option_override): Issue error if
22479 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
22480 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
22482 * config/s390/s390.opt: New option mpic-data-is-text-relative.
22484 2017-06-27 Andrew Pinski <apinski@cavium.com>
22486 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
22487 (X * copysign (1.0, X)): New pattern.
22488 (X * copysign (1.0, -X)): New pattern.
22489 (copysign (-1.0, CST)): New pattern.
22491 2017-06-27 Joseph Myers <joseph@codesourcery.com>
22493 * genmultilib (combination_space): Remove variable.
22494 Validate reuse rules against regular expression for any sequence
22495 of multilib options in any order.
22497 2017-06-27 Michael Collison <michael.collison@arm.com>
22499 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
22500 call aarch64_split_simd_combine.
22501 * (aarch64_combine_internal<mode>): Delete pattern.
22502 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
22503 Allow register and subreg operands.
22505 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22507 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
22508 specific need, just fallback on defaults.
22509 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
22511 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22512 Olivier Hainque <hainque@adacore.com>
22514 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
22516 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
22517 targets. Pick a default if no particular attempt applied.
22518 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
22521 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22523 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
22524 (x86_64-wrs-vxworks7): Likewise.
22526 2017-06-27 Marek Polacek <polacek@redhat.com>
22529 * ubsan.c (instrument_null): Check get_base_address's result for null.
22531 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22533 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
22535 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22537 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
22538 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
22539 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
22540 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
22541 New function types.
22542 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
22543 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
22544 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
22545 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
22546 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
22547 BUILT_IN_FEUPDATEENV): New builtins.
22548 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
22549 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
22550 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
22551 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
22553 (builtin_structptr_types): Adjust size.
22554 * tree.c (builtin_structptr_types): Add four entries.
22556 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22557 Olivier Hainque <hainque@adacore.com>
22559 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
22560 (TLS_SYM): New local macro, forcing reference to __tls__ on
22561 link command lines for VxWorks 7 RTPs, triggering initialization
22563 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
22564 OS features TLS support, true for RTPs on VxWorks 7.
22565 * config/vxworks.c (vxworks_override_options): Setup emutls
22568 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22570 * predict.c (test_prediction_value_range): Use -1U instead of -1
22571 to avoid narrowing conversion warning.
22572 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
22573 to avoid narrowing conversion warning.
22574 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
22576 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
22578 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22580 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
22581 64bit configurations.
22582 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
22583 (SIZE_TYPE): Likewise.
22584 * config/vxworks.c (vxworks_emutls_var_fields): Use
22585 long_unsigned_type_node instead of unsigned_type_node as the offset
22586 field type, which is "pointer" mode in emutls.c.
22588 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22591 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
22593 PR middle-end/81207
22594 * gimple-fold.c (replace_call_with_call_and_fold): Handle
22595 gimple_vuse copying separately from gimple_vdef copying.
22597 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22599 * value-prof.c (free_hist): Remove call to memset and the enclosing if
22602 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22603 Olivier Hainque <hainque@adacore.com>
22605 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
22606 for all vxworks7 targets.
22607 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
22608 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
22609 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
22610 variations for VX6/VX7 and 32/64bits later on in ...
22611 (VXWORKS_LIB_SPEC): Leverage new macros.
22612 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
22613 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
22615 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22617 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
22618 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
22620 2017-06-26 Carl Love <cel@us.ibm.com>
22622 * config/rs6000/rs6000-c.c: Add support for built-in functions
22623 vector bool char vec_reve (vector bool char);
22624 vector signed char vec_reve (vector signed char);
22625 vector unsigned char vec_reve (vector unsigned char);
22626 vector bool int vec_reve (vector bool int);
22627 vector signed int vec_reve (vector signed int);
22628 vector unsigned int vec_reve (vector unsigned int);
22629 vector bool long long vec_reve (vector bool long long);
22630 vector signed long long vec_reve (vector signed long long);
22631 vector unsigned long long vec_reve (vector unsigned long long);
22632 vector bool short vec_reve (vector bool short);
22633 vector signed short vec_reve (vector signed short);
22634 vector double vec_reve (vector double);
22635 vector float vec_reve (vector float);
22636 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
22637 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
22638 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
22639 (altivec_vreve): New pattern.
22640 * config/rs6000/altivec.h (vec_reve): New define.
22641 * doc/extend.texi (vec_rev): Update the built-in documentation file
22642 for the new built-in functions.
22644 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22646 PR tree-optimization/71815
22647 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
22649 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
22651 (slsr_process_phi): Likewise.
22652 (replace_uncond_cands_and_profitable_phis): Don't replace a
22653 multiply candidate with a stride of 1 (copy or cast).
22654 (phi_incr_cost): Call uses_consumed_by_stmt rather than
22656 (lowest_cost_path): Likewise.
22657 (total_savings): Likewise.
22659 2017-06-26 Richard Biener <rguenther@suse.de>
22662 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
22663 Use def_builtin_pure for all gather builtins.
22665 2017-06-26 Richard Biener <rguenther@suse.de>
22667 PR tree-optimization/81203
22668 * tree-tailcall.c (find_tail_calls): Do not move stmts into
22669 non-dominating BBs.
22671 2017-06-26 Marek Polacek <polacek@redhat.com>
22674 * doc/invoke.texi: Document -Wmultistatement-macros.
22676 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
22678 * doc/sourcebuild.texi (ARM-specific attributes): Document new
22679 arm_neon_ok_no_float_abi effective target.
22681 2017-06-26 Richard Biener <rguenther@suse.de>
22683 PR tree-optimization/80928
22684 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
22685 (copy_bbs): Set BB_DUPLICATED flag early.
22686 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
22688 (execute_on_shrinking_pred): Likewise.
22689 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
22690 BB_DUPLICATED blocks.
22691 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
22692 iterate over all PHIs considering removal of *gsi.
22694 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
22696 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
22699 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22701 * config/rs6000/rs6000-string.c: (expand_block_clear,
22702 do_load_for_compare, select_block_compare_mode,
22703 compute_current_alignment, expand_block_compare,
22704 expand_strncmp_align_check, expand_strn_compare,
22705 expand_block_move, rs6000_output_load_multiple)
22706 Move functions related to string/block move/compare
22707 to a separate file.
22708 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
22709 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
22710 for this function which is now used in two files.
22711 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
22712 * config.gcc: Add rs6000-string.o to extra_objs for
22713 targets powerpc*-*-* and rs6000*-*-*.
22715 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
22718 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
22719 32-bit, since indexed is not valid for DImode.
22720 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
22721 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
22722 (define_peephole2 for Altivec d-form load): Add 32-bit support.
22723 (define_peephole2 for Altivec d-form store): Likewise.
22726 * multiple_target.c (create_dispatcher_calls): Only create the
22727 dispatcher call if the function is the default clone of a
22728 versioned function.
22730 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
22732 PR middle-end/80902
22733 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
22734 a call, force the call to not be a tail call.
22736 2017-06-23 Jeff Law <law@redhat.com>
22738 * doc/contrib.texi: Add entry for Steven Pemberton's work on
22741 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
22743 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
22744 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
22745 handling for early expansion of vector shifts (sl,sr,sra,rl).
22746 (builtin_function_type): Add vector shift right instructions
22747 to the unsigned argument list.
22749 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
22751 rtl-optimizatoin/79286
22752 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
22753 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
22754 trap. PIC register plus a const unspec without offset can never trap.
22756 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
22758 * tree.h (builtin_structptr_type): New type.
22759 (builtin_structptr_types): Declare new array.
22760 * tree.c (builtin_structptr_types): New array.
22761 (free_lang_data, build_common_tree_nodes): Use it.
22763 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
22766 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
22769 2017-06-22 Matt Turner <mattst88@gmail.com>
22771 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
22772 Lake models to skylake case. Assume skylake for unknown
22773 models with clflushopt.
22775 2017-06-22 Jeff Law <law@redhat.com>
22777 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
22778 frame sizes that do not satisfy aarch64_uimm12_shift.
22780 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
22782 * profile-count.h (apply_probability,
22783 apply_scale, probability_in): Fix checks for zero.
22785 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22787 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
22788 * doc/cppdiropts.texi (-I @var{dir}): Document it.
22790 2016-06-22 Richard Biener <rguenther@suse.de>
22792 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22793 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
22794 REDUC_MAX_EXPR support.
22795 (vectorizable_reduction): Likewise.
22796 (vect_create_epilog_for_reduction): Likewise.
22798 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
22800 * match.pd (A / (1 << B) -> A >> B): New.
22801 * generic-match-head.c: Include optabs-tree.h.
22802 * gimple-match-head.c: Likewise.
22803 * optabs-tree.h (target_supports_op_p): New.
22804 * optabs-tree.c (target_supports_op_p): New.
22806 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22808 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
22809 $gcc_cv_ld --help output.
22810 (gcc_cv_ld_demangle): Likewise.
22811 (gcc_cv_ld_eh_frame_hdr): Likewise.
22812 (gcc_cv_ld_pie): Likewise.
22813 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
22814 (gcc_cv_ld_buildid): Likewise.
22815 (gcc_cv_ld_sysroot): Likewise.
22816 (ld_bndplt_support): Likewise.
22817 (ld_pushpopstate_support): Likewise.
22818 * configure: Regenerate.
22819 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
22821 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22824 * config/i386/sse.md (round<mode>2): Renumber match_dup and
22825 operands indexes to avoid gap between operands and match_dups.
22827 2017-06-21 Andrew Pinski <apinski@cavium.com>
22829 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
22830 Increment Arith_shift and Arith_shift_reg by 1.
22831 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
22833 * config/aarch64/aarch64.c (thunderx_tunings): Enable
22834 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
22835 (aarch64_strip_extend): Add new argument and test for it.
22836 (aarch64_cheap_mult_shift_p): New function.
22837 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
22838 add a cost if it is true.
22839 Update calls to aarch64_strip_extend.
22840 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
22842 2017-06-21 Andrew Pinski <apinski@cavium.com>
22844 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
22846 (thunderxt88): Likewise.
22847 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
22848 (thunderx_prefetch_tune): New variable.
22849 (thunderx2t99_prefetch_tune): Update for the correct values.
22850 (thunderxt88_tunings): New variable.
22851 (thunderx_tunings): Use thunderx_prefetch_tune instead of
22852 generic_prefetch_tune.
22853 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
22855 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22857 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
22858 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
22859 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
22860 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
22861 (aarch64_atomic_cas<mode>, GPI): Likewise.
22863 2017-06-21 Martin Liska <mliska@suse.cz>
22865 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
22866 statements on cold and hot labels.
22867 * predict.c (tree_estimate_probability_bb): Remove the
22868 prediction from this place.
22870 2017-06-21 Martin Liska <mliska@suse.cz>
22872 PR tree-optimization/79489
22873 * gimplify.c (maybe_add_early_return_predict_stmt): New
22875 (gimplify_return_expr): Call the function.
22876 * predict.c (tree_estimate_probability_bb): Remove handling
22878 * predict.def: Update comment about early return predictor.
22879 * gimple-predict.h (is_gimple_predict): New function.
22880 * predict.def: Change default value of early return to 66.
22881 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
22883 * passes.def: Put pass_strip_predict_hints to the beginning of
22886 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
22888 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
22889 FUNCTION_DECL declarations.
22890 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
22892 (dwaf2out_decl): Likewise.
22893 * godump.c (go_early_global_decl): Skip call to the real debug hook
22894 for FUNCTION_DECL declarations.
22895 * passes.c (rest_of_decl_compilation): Skip call to the
22896 early_global_decl debug hook for FUNCTION_DECL declarations, unless
22897 -fdump-go-spec is passed.
22899 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22901 * config/i386/i386.c (struct builtin_isa): New field pure_p.
22902 Reorder for compactness.
22903 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
22904 (def_builtin_pure, def_builtin_pure2): New functions.
22905 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
22907 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22909 * match.pd (nop_convert): New predicate.
22910 ((A +- CST1) +- CST2): Allow some NOP conversions.
22912 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22915 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
22916 with ctors/dtors if GOVD_SHARED is set.
22918 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22920 * config/aarch64/aarch64.md (movti_aarch64):
22921 Emit mov rather than orr.
22922 (movtf_aarch64): Likewise.
22923 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
22924 Emit mov rather than orr.
22926 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22928 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
22929 Swap alternatives, make integer dup more expensive.
22931 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22933 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
22934 Return true for non-tls symbols.
22936 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
22938 * config/aarch64/aarch64-cores.def (cortex-a55): New.
22939 (cortex-a75): Likewise.
22940 (cortex-a75.cortex-a55): Likewise.
22941 * config/aarch64/aarch64-tune.md: Regenerate.
22942 * doc/invoke.texi (-mtune): Document new values for -mtune.
22944 2017-06-21 Tom de Vries <tom@codesourcery.com>
22946 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
22947 stack_size feature.
22948 (Effective-Target Keywords, Other attributes): Suggest using
22949 dg-add-options stack_size feature to get stack limit in stack_size
22950 effective target documentation.
22952 2017-06-21 Julian Brown <julian@codesourcery.com>
22953 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22955 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
22956 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
22957 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
22959 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
22960 attribute type list for neon_multiply.
22961 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
22962 attribute type list for neon_multiply.
22963 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
22964 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
22965 attribute type list for neon_multiply.
22966 * config/arm/types.md (crypto_pmull): Add.
22967 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
22968 attribute type list.
22970 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
22972 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
22975 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22977 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
22978 to make sure not to dereference a NULL cost_classes_ptr pointer.
22980 2017-06-20 Carl Love <cel@us.ibm.com>
22982 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22983 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22984 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22985 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22986 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22987 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22988 VMULOSW): New enum "unspec" values.
22989 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22990 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22991 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22992 altivec_vmulosw): New patterns.
22993 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22994 VMULOSW): Add definitions.
22996 2017-06-20 Julia Koval <julia.koval@intel.com>
22998 * config/i386/i386.c: Fix rounding expand for new pattern.
22999 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23001 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23003 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23004 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23006 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23008 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23011 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23013 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23014 architecture, then by implementer ID.
23015 * config/aarch64/aarch64-tune.md: Regenerate.
23017 2017-06-20 Richard Biener <rguenther@suse.de>
23019 PR middle-end/81097
23020 * fold-const.c (split_tree): Fold to type before negating.
23022 2017-06-20 David Malcolm <dmalcolm@redhat.com>
23024 * diagnostic-show-locus.c
23025 (selftest::test_fixit_deletion_affecting_newline): New function.
23026 (selftest::diagnostic_show_locus_c_tests): Call it.
23028 2017-06-20 Andreas Schwab <schwab@suse.de>
23031 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
23034 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23036 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
23037 __ARM_FEATURE_COPROC according to support.
23039 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23041 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23042 Rewritten to avoid overflow for > 32-bit pointers.
23045 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
23046 by removing enum keyword.
23047 (ubsan_type_descriptor): Likewise. Formatting fix.
23050 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
23051 splitter): Require TARGET_SSE2 in the condition.
23053 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
23056 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
23057 for doing vector set of SFmode on ISA 3.0.
23058 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
23059 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
23061 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
23062 SFmode value into a V4SF variable that was extracted from another
23063 V4SF variable without converting the element to double precision
23064 and back to single precision vector format.
23065 (vsx_insert_extract_v4sf_p9_2): Likewise.
23067 2017-06-19 Jakub Jelinek <jakub@redhat.com>
23069 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
23070 in UWHI to avoid undefined overflow.
23073 * ubsan.h (enum ubsan_encode_value_phase): New.
23074 (ubsan_encode_value): Change second argument to
23075 enum ubsan_encode_value_phase with default value of
23076 UBSAN_ENCODE_VALUE_GENERIC.
23077 * ubsan.c (ubsan_encode_value): Change second argument to
23078 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
23079 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
23080 create_tmp_var_raw instead of create_tmp_var and use a
23082 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
23083 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
23084 ubsan_encode_value callers.
23087 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
23088 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
23089 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
23091 2017-06-19 Richard Biener <rguenther@suse.de>
23093 PR middle-end/81118
23094 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
23095 estimates if we changed anything.
23097 2017-06-19 Richard Biener <rguenther@suse.de>
23099 PR tree-optimization/80887
23100 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
23101 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
23102 simplified lookups, then reset mprts_hook.
23103 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
23105 (try_to_simplify): Likewise.
23107 2017-06-19 Martin Liska <mliska@suse.cz>
23110 * gimplify.c (gimplify_switch_expr):
23111 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
23113 2017-06-19 Martin Liska <mliska@suse.cz>
23115 * doc/install.texi: Document that PGO runs in 4 stages.
23117 2017-06-19 Martin Liska <mliska@suse.cz>
23120 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
23121 to dispatcher function name.
23122 * multiple_target.c (replace_function_decl): New function.
23123 (create_dispatcher_calls): Redirect both edges and references.
23125 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
23127 * profile-count.c (profile_count::dump): Dump quality.
23128 (profile_count::differs_from_p): Update for unsigned val.
23129 * profile-count.h (profile_count_quality): New enum.
23130 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
23132 2017-06-19 Richard Biener <rguenther@suse.de>
23134 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
23135 struct function as arg.
23136 (estimate_numbers_of_iterations): Export overload with loop arg.
23137 (free_numbers_of_iterations_estimates_loop): Use an overload of
23138 free_numbers_of_iterations_estimates instead.
23139 * tree-cfg.c (remove_bb): Adjust.
23140 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
23141 * tree-parloops.c (gen_parallel_loop): Likewise.
23142 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
23144 (tree_unroll_loops_completely): Likewise.
23145 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
23146 Use an overload instead and export.
23147 (estimated_loop_iterations): Adjust.
23148 (max_loop_iterations): Likewise.
23149 (likely_max_loop_iterations): Likewise.
23150 (estimate_numbers_of_iterations): Take struct function as arg
23152 (loop_exits_before_overflow): Adjust.
23153 (free_numbers_of_iterations_estimates_loop): Use an overload.
23154 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
23155 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
23157 2017-06-19 Richard Biener <rguenther@suse.de>
23160 * ipa-prop.c (find_constructor_constant_at_offset): Handle
23161 RANGE_EXPR conservatively.
23163 2017-06-16 Carl Love <cel@us.ibm.com>
23165 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23166 definitions for vec_float, vec_float2, vec_floato,
23167 vec_floate built-ins.
23168 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
23169 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
23171 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
23172 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
23173 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
23174 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
23175 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
23176 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
23177 vec_floato): Add builtin defines.
23178 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
23179 Update the built-in documentation file for the new built-in
23182 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23184 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
23185 (mthumb): Mark as the negative of -marm.
23187 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23189 * doc/invoke.texi (ARM Options, -mcpu): Document supported
23191 (ARM Options, -mtune): Document that this accepts the same
23192 extension options as -mcpu.
23193 (ARM Options, -mfpu): Document addition of -mfpu=auto.
23195 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23197 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
23198 permitted extensions.
23200 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23202 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
23203 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
23204 (armv8-m.main): Add option +nodsp.
23205 * config/arm/arm-cpu-cdata.h: Regenerated.
23207 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23209 * config/arm/t-fuchsia: New file.
23210 * config.gcc (arm*-*-fuchsia*): Use it.
23212 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23214 * config/arm/t-symbian: Rewrite for new option infrastructure.
23216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23218 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
23219 (MULTILIB_REQUIRED): Likewise.
23221 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23223 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
23224 (MULTILIB_RESUE): Likewise.
23225 (MULTILIB_MATCHES): Likewise.
23226 (MULTLIB_REQUIRED): Likewise.
23228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23230 * config/arm/t-rtems: Rewrite for new option framework.
23232 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23234 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
23235 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
23236 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
23237 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
23238 * config/arm/t-multilib: ... here.
23239 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
23240 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
23241 armv7-a and armv8*-a when A-profile libraries have not been built.
23242 * config/arm/t-rmprofile: Rewrite.
23244 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23246 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
23247 with a backslash. Remove the backslash after substituting unescaped
23249 * doc/fragments.texi (MULTILIB_REUSE): Document it.
23251 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23253 * config.gcc: (arm*-*-*): When building a-profile libraries, force
23254 the driver to pass through the default setting of -mfloat-abi.
23255 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
23257 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
23258 (all_feat_combs): New rule.
23259 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
23261 * config/arm/t-aprofile: Rewrite.
23263 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23265 * config/arm/arm.h (FPUTYPE_AUTO): Define.
23266 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
23267 fpu is not specified by the user/command-line.
23268 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
23269 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
23270 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
23271 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
23272 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
23273 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
23275 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23277 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
23278 * config/arm/t-arm-elf: Rewritten.
23280 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23282 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
23283 have some floating-point instructions.
23284 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
23285 (TARGET_MAYBE_HARD_FLOAT): New macro.
23286 * config/arm/arm-builtins.c (arm_init_builtins): Use
23287 TARGET_MAYBE_HARD_FLOAT.
23288 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
23290 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23292 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
23293 (configargs.h): Include it.
23294 (arm_print_hint_for_fpu_option): New function.
23295 (arm_parse_fpu_option): New function.
23296 (candidate_extension): New class.
23297 (arm_canon_for_multilib): New function.
23298 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
23299 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
23300 (ARCH_CANONICAL_SPECS): New macro.
23301 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
23303 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23305 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
23306 are set after handling multilib fragments. Set target_cpu_default2
23309 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23311 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
23313 (arm*-*-*): Set target_cpu_default2 to a quoted string.
23314 * config/arm/parsecpu.awk (check_cpu): Validate any extension
23316 (check_arch): Likewise.
23317 * config/arm/arm.c (arm_configure_build_target): Handle
23318 TARGET_CPU_DEFAULT being a string constant. Scan any feature
23319 options in the default.
23321 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23323 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
23324 when an option is an alias of another.
23325 * config/arm/parsecpu.awk (optalias): New parser token.
23326 (gen_comm_data): Mark non-alias options as such. Emit entries
23327 for extension aliases.
23328 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
23329 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
23330 (armv6kz, armv6zk, armv6t2): Likewise.
23331 (armv7): Make vfpv3-d16 an alias.
23332 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
23334 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
23335 Sort in canonical order.
23336 (armv8-a): Sort in canonical order.
23337 (armv8.1-a, armv8.2-a): Likewise.
23338 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
23340 (cortex-a9): Sort in canonical order.
23341 * config/arm/arm.c (selftests.h): Include it.
23342 (arm_test_cpu_arch_data): New function.
23343 (arm_run_self_tests): New function.
23344 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
23345 (targetm): Move declaration to the end of the file.
23346 * arm-cpu-cdata.h: Regenerated.
23348 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23350 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
23351 call to target_mode_check describing the type of option passed.
23352 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
23353 (arm_target_thumb_only): Use arm_parse_arch_option_name or
23354 arm_parse_cpu_option_name to match parameters against list of
23356 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
23357 arm_arch_core_flags data structure.
23358 * config/arm/arm-cpu_cdata.h: Regenerated.
23360 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23362 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
23364 (arm_print_hint_for_cpu_option): Likewise.
23365 (arm_print_hint_for_arch_option): Likewise.
23366 (arm_parse_cpu_option_name): Likewise.
23367 (arm_parse_arch_option_name): Likewise.
23368 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
23369 of entries in the all_fpus list.
23370 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
23371 (arm_parse_cpu_option_name): Declare.
23372 (arm_parse_arch_option_name): Declare.
23373 (arm_parse_option_features): Declare.
23374 (arm_intialize_isa): Declare.
23375 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
23377 (gen_comm_data): ... here. Make definitions non-static.
23378 * config/arm/arm-cpu-data.h: Regenerated.
23379 * config/arm/arm-cpu-cdata.h: Regenerated.
23381 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23383 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
23384 (cpu_arch_extension): New structure.
23385 (cpu_arch_option, arch_option, cpu_option): New structures.
23386 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
23387 architecture types.
23388 (gen_data): Generate new format data tables.
23389 * config/arm/arm.c (cpu_tune): New structure.
23390 (cpu_option, processors): Delete.
23391 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
23392 (arm_print_hint_for_cpu_option): ... this and ...
23393 (arm_print_hint_for_arch_option): ... this.
23394 (arm_parse_arch_cpu_name): Delete. Replace with ...
23395 (arm_parse_cpu_option_name): ... this and ...
23396 (arm_parse_arch_option_name): ... this.
23397 (arm_unrecognized_feature): Change type of target parameter to
23399 (arm_parse_arch_cpu_features): Delete. Replace with ...
23400 (arm_parse_option_features): ... this.
23401 (arm_configure_build_target): Rework to use new configuration data
23403 (arm_print_tune_info): Rework for new configuration data tables.
23404 * config/arm/arm-cpu-data.h: Regenerated.
23405 * config/arm/arm-cpu.h: Regenerated.
23407 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23409 * Makefile.in (OBJS): Move sbitmap.o from here ...
23410 (OBJS-libcommon): ... to here.
23412 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23414 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
23415 (ISA_ALL_CRYPTO): New macro.
23416 (ISA_ALL_SIMD): New macro
23417 (ISA_ALL_FP): New macro.
23418 * config/arm/arm.c (fpu_bitlist): Update initializer.
23419 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
23421 (arm9e): Add fpu. Add option for nofp
23422 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
23423 (arm926ej-s, arm1026ej-s): Likewise.
23424 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
23425 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
23426 neon-fp16, neon-vfpv4, nofp and nosimd.
23427 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
23428 (cortex-a8): Add fpu. Add option for nofp.
23429 (cortex-a9): Add fpu. Add options for nosimd and nofp.
23430 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
23431 (cortex-r4f): Add fpu.
23432 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
23433 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
23435 (cortex-r8): Likewise.
23436 (cortex-m4): Add fpu. Add option for nofp.
23437 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
23438 (cortex-a17.cortex-a7): Likewise.
23439 (cortex-a32): Add fpu. Add options for crypto and nofp.
23440 (cortex-a35, cortex-a53): Likewise.
23441 (cortex-a57): Add fpu. Add option for crypto.
23442 (cortex-a72, cortex-a73): Likewise.
23443 (exynos-m1): Likewise.
23444 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
23445 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
23446 (cortex-m33): Add fpu. Add option for nofp.
23447 * config/arm/arm-cpu-cdata.h: Regenerated
23448 * config/arm/arm-cpu-data.h: Regenerated.
23450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23452 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
23453 (armv5te, armv5tej): Likewise.
23454 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
23455 (armv7): Add options fp and vfpv3-d16.
23456 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
23457 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
23459 (armv7ve): Likewise.
23460 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
23461 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
23462 (armv8-a): Add nocrypto option.
23463 (armv8.1-a, armv8.2-a): Likewise.
23464 (armv8-m.main): add options fp, fp.dp and nofp.
23466 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23468 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
23470 (armv8-a+crc): Delete.
23471 (armv8.1-a): Add options simd, crypto and nofp.
23472 (armv8.2-a): Add options fp16, simd, crypto and nofp.
23473 (armv8.2-a+fp16): Delete.
23474 (armv8-m.main): Add option dsp.
23475 (armv8-m.main+dsp): Delete.
23476 (cortex-a8): Add fpu. Add nofp option.
23477 (cortex-a9): Add fpu. Add nofp and nosimd options.
23478 * config/arm/parsecpu.awk (gen_data): Generate option tables and
23479 link to main cpu and architecture data structures.
23480 (gen_comm_data): Only put isa attributes from the main architecture
23482 (option): New statement for architecture and CPU entries.
23483 * arm.c (struct cpu_option): New structure.
23484 (struct processors): Add entry for options.
23485 (arm_unrecognized_feature): New function.
23486 (arm_parse_arch_cpu_name): Ignore any characters after the first
23488 (arm_parse_arch_cpu_feature): New function.
23489 (arm_configure_build_target): Separate out any CPU and architecture
23490 features and parse separately. Don't error out if -mfpu=auto is
23491 used with only an architecture string.
23492 (arm_print_asm_arch_directives): New function.
23493 (arm_file_start): Call it.
23494 * config/arm/arm-cpu-cdata.h: Regenerated.
23495 * config/arm/arm-cpu-data.h: Likewise.
23496 * config/arm/arm-tables.opt: Likewise.
23498 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23500 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
23501 assembler when it is not -mfpu=auto.
23503 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23505 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
23506 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
23507 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
23508 (ASM_CPU_SPEC): Rewrite.
23509 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
23510 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
23511 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
23512 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
23513 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
23514 copied string is NUL-terminated. Also strip any characters prefixed
23516 (arm_rewrite_selected_arch): New function.
23517 (arm_rewrite_march): New function.
23519 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23521 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
23522 (x_arm_cpu_string, x_arm_tune_string): Likewise.
23523 (march, mcpu, mtune): Convert to string-based options.
23524 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
23525 (arm_parse_arch_cpu_name): New function.
23526 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
23527 identify selected architecture or CPU.
23528 (arm_option_save): New function.
23529 (TARGET_OPTION_SAVE): Redefine.
23530 (arm_option_restore): Restore string options.
23531 (arm_option_print): Print string options.
23533 2017-06-16 Martin Sebor <msebor@redhat.com>
23535 PR tree-optimization/80933
23536 PR tree-optimization/80934
23537 * builtins.c (fold_builtin_3): Do not handle bcmp here.
23538 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
23539 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
23540 (gimple_fold_builtin): Call them.
23542 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23544 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
23545 as unlikely; update profile.
23547 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23549 * predict.c (force_edge_cold): Handle declaring edges impossible
23552 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23554 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
23556 (try_unroll_loop_completely): Fix reporting.
23558 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23560 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
23562 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
23565 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
23566 if given a non-constant argument for an intrinsic which requires a
23569 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23571 * profile.c (compare_freqs): New function.
23572 (branch_prob): Sort edge list.
23573 (find_spanning_tree): Assume that the list is priority sorted.
23575 2017-06-16 Richard Biener <rguenther@suse.de>
23577 PR tree-optimization/81090
23578 * passes.def (pass_record_bounds): Remove.
23579 * tree-pass.h (make_pass_record_bounds): Likewise.
23580 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
23581 make_pass_record_bounds): Likewise.
23582 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
23583 not free niter estimates at the beginning but at the end.
23584 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
23586 2017-06-16 Richard Biener <rguenther@suse.de>
23588 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
23589 initializer to workaround ICE in host GCC 4.8.
23591 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23593 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
23595 (clone_inlined_nodes): Update.
23597 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23599 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
23600 prefetch settings, and enable prefetching by default at -O3.
23602 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23604 * config/aarch64/aarch64.c (aarch64_override_options_internal):
23605 Set flag_prefetch_loop_arrays according to tuning data.
23607 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23609 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
23610 New tune structure.
23611 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
23612 [Unrelated to main purpose of the patch] Place the pointer field last
23613 to enable type checking errors when tune structure are wrongly merged.
23614 * config/aarch64/aarch64.c (generic_prefetch_tune,)
23615 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
23616 (thunderx2t99_prefetch_tune): New tune constants.
23617 (tune_params *_tunings): Update all tunings (no functional change).
23618 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
23619 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
23620 from tunings structures.
23622 2017-06-16 Jakub Jelinek <jakub@redhat.com>
23625 * ubsan.c (instrument_null): Add T argument, use it instead
23626 of computing it based on IS_LHS.
23627 (instrument_object_size): Likewise.
23628 (pass_ubsan::execute): Adjust instrument_null and
23629 instrument_object_size callers to pass gimple_get_lhs or
23630 gimple_assign_rhs1 result to it. Use instrument_null instead of
23631 calling get_base_address and instrument_mem_ref. Handle
23632 aggregate call arguments for object-size sanitization.
23634 2017-06-16 Yury Gribov <tetra2005@gmail.com>
23636 PR tree-optimization/81089
23637 * tree-vrp.c (is_masked_range_test): Validate operands of
23640 2017-06-15 Martin Sebor <msebor@redhat.com>
23643 * dumpfile.c (dump_register): Avoid calling memset to initialize
23644 a class with a default ctor.
23645 * gcc.c (struct compiler): Remove const qualification.
23646 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
23647 * hash-table.h: Ditto.
23648 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
23650 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
23651 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
23653 * params.h (struct param_info): Make struct members non-const.
23654 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
23655 with default initialization.
23656 * vec.h (vec_copy_construct, vec_default_construct): New helper
23658 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
23659 with vec_copy_construct.
23660 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
23661 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
23662 * doc/invoke.texi (-Wclass-memaccess): Document.
23664 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23666 * emit-rtl.h (is_leaf): Update comment about local
23667 register allocator.
23669 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23672 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
23673 for a variable to have a section before checking if the section has a
23675 Set section to.persistent if persistent attribute is set.
23676 Warn if .persistent attribute is used on an automatic variable.
23678 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
23680 PR rtl-optimization/80474
23681 * reorg.c (update_block): Do not ignore instructions in a delay slot.
23683 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
23685 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
23688 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
23690 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
23691 (casesi): Emit bounds checking as RTL.
23692 (casesi_internal_mips16_<mode>): Remove bounds checking.
23694 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
23696 * config/xtensa/xtensa.c (xtensa_option_override): Append
23697 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
23698 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
23699 xtensa_doloop_hooks): Define unconditionally.
23700 (xtensa_reorg_loops): Only call reorg_loops in the presence of
23702 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
23703 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
23704 for it in xtensa_option_override.
23705 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
23706 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
23708 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
23710 * doc/cppopts.texi: Document '-' special value to -MF.
23712 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
23714 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
23715 (cortex_a53_fconst): Likewise.
23716 (cortex_a53_fpmul): Likewise.
23717 (cortex_a53_f_load_64): Likewise.
23718 (cortex_a53_f_load_many): Likewise.
23719 (cortex_a53_advsimd_alu): Likewise.
23720 (cortex_a53_advsimd_alu_q): Likewise.
23721 (cortex_a53_advsimd_mul): Likewise.
23722 (cortex_a53_advsimd_mul_q): Likewise.
23723 (fpmac bypass): Add new bypass for fpmac-fpmac case.
23724 Add missing fmul, r2f_cvt and fconst cases.
23726 2017-06-14 Richard Biener <rguenther@suse.de>
23728 PR middle-end/81088
23729 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
23731 (fold_binary_loc): When associating do not treat pre-existing
23732 TREE_OVERFLOW on literal constants as a reason to allow
23733 TREE_OVERFLOW on associated literal constants.
23735 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23737 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
23738 (MASK_FEATURES): New macro.
23739 * config/sparc/sparc.c (sparc_option_override): Remove the special
23740 handling of -mfpu and generalize it to all MASK_FEATURES switches.
23742 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23744 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
23745 a division of 0 if non-call exceptions are enabled.
23747 2017-06-14 Andrew Pinski <apinski@cavium.com>
23748 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23751 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
23752 Improve vector initialization code gen for only variable case.
23754 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23756 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
23758 2017-06-14 Richard Biener <rguenther@suse.de>
23760 PR tree-optimization/81083
23761 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
23764 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23766 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
23767 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
23768 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
23769 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
23770 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
23771 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
23773 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23775 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
23776 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
23778 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23780 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
23782 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23784 * config/rs6000/t-rtems: Don't handle SPE.
23786 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23788 * config/rs6000/t-linux: Don't handle SPE.
23790 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23792 * config/rs6000/eabispe.h: Delete file.
23794 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23796 * config/rs6000/t-spe: Delete file.
23798 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23800 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
23801 (rs6000_legitimate_offset_address_p): Return false for anything in
23802 V2SImode or V2SFmode.
23804 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23806 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
23807 except V2SF and V2SI. Rearrange the vector modes, and add comments.
23808 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
23810 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
23811 (rs6000_legitimate_offset_address_p): Ditto.
23812 (rs6000_emit_move): Ditto.
23813 (rs6000_init_builtins): Remove V4HI_type_node.
23815 2017-06-13 Martin Liska <mliska@suse.cz>
23818 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
23819 (gate_asan): Likewise.
23820 * asan.h (asan_no_sanitize_address_p): Remove the function.
23821 (sanitize_flags_p): New function.
23822 * builtins.def: Fix coding style.
23823 * common.opt: Use renamed enum value.
23824 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
23825 * doc/extend.texi: Document no_sanitize attribute.
23826 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
23827 to SANITIZE_UNDEFINED_NONDEFAULT.
23828 * gcc.c (sanitize_spec_function): Use the renamed enum value.
23829 * gimple-fold.c (optimize_atomic_compare_exchange_p):
23830 Use sanitize_flags_p.
23831 * gimplify.c (gimplify_function_tree): Likewise.
23832 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
23833 * opts.c (parse_no_sanitize_attribute): New function.
23834 (common_handle_option): Use renamed enum value.
23835 * opts.h (parse_no_sanitize_attribute): Declare.
23836 * tree.c (sanitize_flags_p): New function.
23837 * tree.h: Declared here.
23838 * tsan.c: Use sanitize_flags_p.
23839 * ubsan.c (ubsan_expand_null_ifn): Likewise.
23840 (instrument_mem_ref): Likewise.
23841 (instrument_bool_enum_load): Likewise.
23842 (do_ubsan_in_current_function): Remove the function.
23843 (pass_ubsan::execute): Use sanitize_flags_p.
23844 * ubsan.h: Remove do_ubsan_in_current_function
23845 * tree-cfg.c (print_no_sanitize_attr_value): New function.
23846 (dump_function_to_file): Use it here.
23848 2017-06-13 Martin Jambor <mjambor@suse.cz>
23850 PR tree-optimization/80803
23851 PR tree-optimization/81063
23852 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
23853 (propagate_subaccesses_across_link): Enqueue subtree whenever
23854 necessary instead of relying on the caller.
23856 2017-06-13 Martin Jambor <mjambor@suse.cz>
23858 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
23859 that have a first_link.
23860 (sort_and_splice_var_accesses): Do not check first_link before
23862 (subtree_mark_written_and_enqueue): Likewise.
23863 (propagate_all_subaccesses): Likewise and do not stop at first
23864 parent with a first_link.
23866 2017-06-13 Martin Jambor <mjambor@suse.cz>
23868 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
23871 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23873 * match.pd: New pattern.
23875 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23877 * tree-vrp.c (is_masked_range_test): New function.
23878 (register_edge_assert_for): Determine ranges for
23881 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23883 PR tree-optimization/67328
23884 * fold-const.c (maskable_range_p): New function.
23885 (build_range_check): Generate bittests if possible.
23887 2017-06-13 Martin Liska <mliska@suse.cz>
23889 * gimple-pretty-print.c (dump_probability): Add new argument.
23890 (dump_edge_probability): Dump both probability and count.
23891 (dump_gimple_label): Likewise.
23892 (dump_gimple_bb_header): Likewise.
23894 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
23897 * config/avr/avr-devices.c: Fix indentation.
23898 * config/avr/gen-avr-mmcu-specs.c: Dito.
23900 2017-06-13 Richard Biener <rguenther@suse.de>
23902 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
23903 instead get vector type from stmt_info.
23904 (vectorizable_reduction): Adjust. Remove dead code.
23906 2017-06-13 Richard Biener <rguenther@suse.de>
23908 PR middle-end/81065
23909 * fold-const.c (extract_muldiv_1): Remove bogus distribution
23910 case of C * (x * C2 + C3).
23911 (fold_addr_of_array_ref_difference): Properly fold index difference.
23913 2017-06-12 David S. Miller <davem@davemloft.net>
23916 * config/sparc/sparc.md (return expander): Emit frame blockage if
23917 function uses alloca.
23919 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
23921 * combine.c (make_field_assignment): Check len rather than the mode
23922 precision when calling force_to_mode.
23924 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
23926 Support multilibs and devices that see flash in RAM address range.
23929 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
23930 (avr_mcu_t) <flash_pm_offset>: New field.
23931 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
23932 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
23933 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
23934 (AVR_TINY_PM_OFFSET): Remove macro.
23935 * config/avr/avr.opt (-mshort-calls): New option.
23936 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23937 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
23938 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
23939 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
23940 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
23941 instead of avr_arch->have_jmp_call.
23942 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
23943 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
23944 avr_arch->flash_pm_offset to define.
23945 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
23946 new field flash_pm_offset. Add entry for avrxmega3.
23947 (avr_texinfo): Add entry for avrxmega3.
23948 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
23949 attiny212, attiny214,
23950 attiny412, attiny414, attiny416, attiny417,
23951 attiny814, attiny816, attiny817,
23952 attiny1614, attiny1616, attiny1617,
23953 attiny3214, attiny3216, attiny3217.
23954 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
23955 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
23956 (avr_print_operand_address) [AVR_TINY]: Same.
23957 (avr_asm_init_sections) <readonly_data_section>: Only patch
23958 callback if avr_arch->flash_pm_offset = 0.
23959 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
23960 for rodata if avr_arch->flash_pm_offset != 0.
23961 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
23962 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
23963 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
23964 (m_options): Append opt_rcall.
23965 (m_dirnames): Append dir_rcall.
23966 * config/avr/t-multilib: Regenerate.
23968 * configure.ac [target=avr]: Check whether avrxmega3 default
23969 linker description file works as needed.
23970 * configure: Regenerate.
23971 * doc/avr-mmcu.texi: Regenerate.
23972 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
23973 <__AVR_ARCH__>: Document avrxmega3 and 103.
23974 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
23975 <__AVR_SHORT_CALLS__>: Document it.
23976 <__AVR_PM_BASE_ADDRESS__>: Document it.
23977 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
23978 (AVR Variable Attributes) <progmem>: Document this is
23979 not needed for avrxmega3.
23980 (AVR Named Address Spaces) <__flash>: Dito.
23982 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
23984 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
23986 2017-06-12 Doug Rupp <rupp@adacore.com>
23988 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
23989 Append vxworks-stdint.h to the tm_file list.
23990 * config/vxworks-stdint.h: New file.
23992 2017-06-12 Martin Liska <mliska@suse.cz>
23994 PR tree-optimization/81041
23995 * tree-profile.c (gimple_gen_ic_func_profiler):
23996 Create an extra BB in profile-generate
23997 (gimple_gen_time_profiler): Likewise.
23999 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24001 PR tree-optimization/81003
24002 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24003 (update_range_test): Use it instead of force_gimple_operand_gsi.
24005 2017-06-12 Richard Biener <rguenther@suse.de>
24007 PR tree-optimization/81053
24008 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24009 with backedge value not defined in loop. Simplify def stmt
24012 2017-06-11 Tom de Vries <tom@codesourcery.com>
24015 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24017 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24018 nvptx_cannot_force_const_mem.
24020 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24022 * opts.c (finish_options): Move test for flag_split_stack after
24023 it has been initialized.
24025 2017-06-11 Jason Merrill <jason@redhat.com>
24027 * tree.h (id_equal): New.
24028 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
24029 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
24030 instead of strcmp of IDENTIFIER_POINTER.
24032 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24034 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
24035 (mark_all_inlined_calls_cdtor): Fix formating.
24036 (inline_transform): Rescale profile before inlining.
24038 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24040 * cgraph.h (cgraph_edge::clone): Update prototype.
24041 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
24042 (cgraph_node::create_clone): Update.
24043 (cgraph_node::create_version_clone): Update.
24044 * tree-inline.c (copy_bb): Update.
24045 (expand_call_inline): Update.
24047 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24049 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
24050 factored out from ...
24051 (rs6000_emit_prologue): ... here.
24053 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24055 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
24056 factored out from ...
24057 (rs6000_emit_prologue): ... here.
24059 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24061 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
24063 (handle_missing_profiles): Fix computation of tp_first_run.
24064 (counts_to_freqs): Do not touch freqs when count is 0.
24066 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24068 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
24071 2017-06-10 Tom de Vries <tom@codesourcery.com>
24073 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24074 attributes): Document signal effective target.
24076 2017-06-10 Tom de Vries <tom@codesourcery.com>
24078 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24079 Document effective target stack_size.
24081 2017-06-09 David Malcolm <dmalcolm@redhat.com>
24083 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
24084 to the edit_context if they can be auto-applied.
24086 2017-06-9 Ian Lance Taylor <iant@golang.org>
24088 * opts.c (finish_options): If -fsplit-stack, disable implicit
24089 -forder-blocks-and-partition.
24090 * doc/invoke.texi (Optimize Options): Document that when using
24091 -fsplit-stack -forder-blocks-and-partition is not implicitly
24094 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24096 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
24097 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
24098 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
24099 * builtins.def (abort, trap, unreachable): Declare cold.
24100 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
24101 * tree-core.h (ECF_COLD): New.
24102 * tree.c (set_call_expr_flags): Handle ECF_COLD.
24103 (build_common_builtin_nodes): Mark unreachable and abort as cold.
24105 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24107 * predict.c (unlikely_executed_stmt_p): Cleanup.
24109 2017-06-09 Richard Biener <rguenther@suse.de>
24111 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
24112 model if the ref is always written to.
24114 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24116 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
24118 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24120 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
24123 2017-06-09 Tom de Vries <tom@codesourcery.com>
24126 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
24127 "target cannot support label values" when encountering LABEL_REF.
24129 2017-06-09 Martin Liska <mliska@suse.cz>
24131 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
24132 (gimple_gen_ic_func_profiler): Emit direct comparison
24133 of __gcov_indirect_call_callee with NULL.
24134 (gimple_gen_time_profiler): Change probability from
24135 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
24137 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24139 * profile.c (edge_gcov_counts): Turn to pointer.
24140 (compute_branch_probabilities, compute_branch_probabilities): Update.
24141 (branch_prob): Do not clear edge_gcov_count.
24142 * profile.h (edge_gcov_counts): Turn to pointer.
24143 (edge_gcov_count): Update.
24145 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24147 * gimple.h (gimple_check_failed): Mark cold.
24149 2017-06-09 Richard Biener <rguenther@suse.de>
24151 PR tree-optimization/66623
24152 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
24153 refactor check_reduction into two parts, properly computing
24154 whether we have to check reduction validity for outer loop
24157 2017-06-09 Richard Biener <rguenther@suse.de>
24159 PR tree-optimization/79483
24160 * graphite-scop-detection.c (order): New global.
24161 (get_order): Compute bb to order mapping that satisfies code
24162 generation constraints.
24163 (cmp_pbbs): New helper.
24164 (build_scops): Start domwalk at entry block, sort generated
24167 2017-06-09 Richard Biener <rguenther@suse.de>
24169 PR middle-end/81007
24170 * ipa-polymorphic-call.c
24171 (ipa_polymorphic_call_context::restrict_to_inner_class):
24172 Skip FIELD_DECLs with error_mark_node type.
24173 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
24176 2017-06-09 Martin Liska <mliska@suse.cz>
24178 * predict.c (struct branch_predictor): New struct.
24179 (test_prediction_value_range): New test.
24180 (predict_c_tests): New function.
24181 * selftest-run-tests.c (selftest::run_tests): Run the function.
24182 * selftest.h: Declare new tests.
24184 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
24187 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
24188 gen_add3_insn did not fail.
24189 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
24190 r0, construct that number in a temporary reg and add that reg to r0.
24191 If asked to put the result in r0 as well, fail.
24193 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
24195 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24196 for early expansion of vec_eqv.
24198 2017-06-08 Jakub Jelinek <jakub@redhat.com>
24200 PR middle-end/81005
24201 * ubsan.c (instrument_null): Avoid pointless code temporary.
24202 (pass_ubsan::execute): Instrument aggregate arguments of calls.
24204 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
24208 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
24211 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
24214 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24216 * predict.c (unlikely_executed_edge_p): Move ahead.
24217 (probably_never_executed_edge_p): Use it.
24219 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24221 PR middle-end/79988
24222 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
24223 gimple_call_builtin_p call.
24225 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24227 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
24228 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
24229 rtl_check_failed_type2, rtl_check_failed_code1,
24230 rtl_check_failed_code2, rtl_check_failed_code_mode,
24231 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
24232 rtvec_check_failed_bounds, rtl_check_failed_flag,
24233 _fatal_insn_not_found, _fatal_insn): Likewise.
24234 * tree.h (tree_contains_struct_check_failed,
24235 tree_check_failed, tree_not_check_failed,
24236 tree_class_check_failed, tree_range_check_failed,
24237 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
24238 tree_vec_elt_check_failed, phi_node_elt_check_failed,
24239 tree_operand_check_failed, omp_clause_check_failed,
24240 omp_clause_operand_check_failed, omp_clause_range_check_failed):
24243 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24245 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
24246 flag_branch_probabilities.
24247 * ipa-inline.c (edge_badness): Likewise.
24248 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
24249 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
24250 * predict.c (maybe_hot_frequency_p): Likewise.
24251 (probably_never_executed): Likewise.
24252 * sched-ebb.c (schedule_ebbs): Likewise.
24253 * sched-rgn.c (find_single_block_region): Likewise.
24254 * tracer.c (tail_duplicate): Likewise.
24256 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24258 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
24259 longer requires x_flag_profile_use.
24261 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24263 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
24264 instead of flag_reorder_blocks_and_partition.
24265 * dbxout.c (dbxout_function_end): Likewise.
24266 * dwarf2out.c (gen_subprogram_die): Likewise.
24267 * haifa-sched.c (sched_create_recovery_edges): Likewise.
24268 * hw-doloop.c (reorg_loops): Likewise.
24269 * varasm.c (assemble_start_function,
24270 assemble_end_function): Likewise.
24271 (decide_function_section): Do not check for
24272 flag_reorder_blocks_and_partition.
24274 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24276 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24278 (chkp_get_hard_register_fake_addr_expr): Ditto.
24279 (chkp_build_addr_expr): Add check for hard reg case.
24280 (chkp_parse_array_and_component_ref): Ditto.
24281 (chkp_find_bounds_1): Ditto.
24282 (chkp_process_stmt): Don't generate bounds store for
24285 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24287 * predict.c (maybe_hot_bb_p): Do not check profile status.
24288 (maybe_hot_edge_p): Likewise.
24289 (probably_never_executed): Check for zero counts even if profile
24291 (unlikely_executed_edge_p): New function.
24292 (unlikely_executed_stmt_p): New function.
24293 (unlikely_executed_bb_p): New function.
24294 (set_even_probabilities): Use unlikely predicates.
24295 (combine_predictions_for_bb): Likewise.
24296 (predict_paths_for_bb): Likewise.
24297 (predict_paths_leading_to_edge): Likewise.
24298 (determine_unlikely_bbs): New function.
24299 (estimate_bb_frequencies): Use it.
24300 (compute_function_frequency): Use zero counts even if profile is
24302 * profile-count.h: Fix typo.
24304 2017-08-08 Julia Koval <julia.koval@intel.com>
24306 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
24307 _mm512_mask_cvtsepi16_storeu_epi8,
24308 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
24309 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
24310 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
24311 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
24312 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
24313 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
24314 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
24315 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
24316 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
24317 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
24318 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
24319 __builtin_ia32_pmovuswb256mem_mask,
24320 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
24321 __builtin_ia32_pmovwb512mem_mask): New builtins.
24323 2017-08-08 Julia Koval <julia.koval@intel.com>
24325 PR target/73350,80862
24326 * config/i386/subst.md (round): Fix round pattern.
24327 * config/i386/i386.c (ix86_erase_embedded_rounding):
24328 Fix erasing rounding for the fixed pattern.
24330 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24332 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
24334 2017-06-08 Martin Liska <mliska@suse.cz>
24336 PR gcov-profile/80911
24337 * gcov.c (block_info::block_info): New constructor.
24339 2017-06-07 Carl Love <cel@us.ibm.com>
24341 * config/rs6000/rs6000-c: The return type of the following
24342 built-in functions was implemented as int not long long. Fix sign
24343 of return value for the unsigned version of vec_mulo and vec_mule.
24344 vector unsigned long long vec_bperm (vector unsigned long long,
24345 vector unsigned char)
24346 vector signed long long vec_mule (vector signed int,
24348 vector unsigned long long vec_mule (vector unsigned int,
24349 vector unsigned int)
24350 vector signed long long vec_mulo (vector signed int,
24352 vector unsigned long long vec_mulo (vector unsigned int,
24353 vector unsigned int)
24354 * doc/extend.texi: Fix the documentation for the built-in
24357 2017-06-07 Carl Love <cel@us.ibm.com>
24360 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
24363 2017-06-07 Carl Love <cel@us.ibm.com>
24365 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
24366 support, Generate doublehv for signed int/float for BE case only.
24368 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
24370 * doc/invoke.texi (mcx16): Rewrite.
24372 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24374 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
24375 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
24376 *mov<mode>_softfloat, and an anonymous splitter): Use
24377 nonimmediate_operand instead of rs6000_nonimmediate_operand.
24379 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24381 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
24383 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24384 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
24385 (rs6000_debug_reg_global): Adjust.
24386 (rs6000_init_hard_regno_mode_ok): Adjust.
24387 (rs6000_dbx_register_number): Adjust.
24388 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
24389 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24390 Remove SPE_ACC and SPEFSCR.
24391 (REG_ALLOC_ORDER): Ditto.
24392 (FRAME_POINTER_REGNUM): Change to 111.
24393 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
24394 (REG_CLASS_NAMES): Ditto.
24395 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
24396 (REGISTER_NAMES): Ditto.
24397 (ADDITIONAL_REG_NAMES): Ditto.
24398 (rs6000_reg_names): Ditto.
24399 * config/rs6000/rs6000.md: Renumber some register number
24402 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24404 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
24406 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24407 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
24409 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
24410 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24411 Delete the SPE high registers.
24412 (REG_ALLOC_ORDER): Ditto.
24413 (enum reg_class): Remove SPE_HIGH_REGS.
24414 (REG_CLASS_NAMES): Ditto.
24415 (REG_CLASS_CONTENTS): Delete the SPE high registers.
24416 (REGISTER_NAMES): Ditto.
24417 (rs6000_reg_names): Ditto.
24418 * doc/tm.texi.in: Remove SPE as example.
24419 * doc/tm.texi: Regenerate.
24421 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24423 * config/rs6000/8540.md (ppc8540_brinc): Delete.
24424 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
24425 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
24426 * config/rs6000/rs6000.md (type): Remove "brinc".
24428 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24430 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
24431 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
24432 * config/rs6000/linuxspe.h: Delete file.
24433 * config/rs6000/rs6000.md: Don't include spe.md.
24434 * config/rs6000/spe.h: Delete file.
24435 * config/rs6000/spe.md: Delete file.
24436 * config/rs6000/t-rs6000: Remove spe.md.
24438 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24440 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
24441 (reg_or_none500mem_operand): Delete.
24442 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
24443 instead of reg_or_none500mem_operand.
24445 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24447 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24448 handling of SPE flags.
24449 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
24451 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24453 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
24455 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
24456 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
24457 paired_divv2sf3): Similar.
24458 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
24459 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
24460 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
24462 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
24463 Rename the paired_* instruction patterns.
24464 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
24466 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
24467 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
24468 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
24469 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
24470 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
24471 PAIRED_VECTOR_MODE.
24472 (struct machine_function): Delete field spe_insn_chain_scanned_p.
24473 (spe_func_has_64bit_regs_p): Delete.
24474 (spe_expand_predicate_builtin): Delete.
24475 (spe_expand_evsel_builtin): Delete.
24476 (TARGET_DWARF_REGISTER_SPAN): Do not define.
24477 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
24478 (invalid_e500_subreg): Delete.
24479 (rs6000_legitimize_address): Always force_reg op2 as well, for
24480 paired single memory accesses.
24481 (rs6000_member_type_forces_blk): Delete.
24482 (rs6000_spe_function_arg): Delete.
24483 (rs6000_expand_unop_builtin): Delete SPE handling.
24484 (rs6000_expand_binop_builtin): Ditto.
24485 (spe_expand_stv_builtin): Delete.
24486 (bdesc_2arg_spe): Delete.
24487 (spe_expand_builtin): Delete.
24488 (spe_expand_predicate_builtin): Delete.
24489 (spe_expand_evsel_builtin): Delete.
24490 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
24491 (spe_init_builtins): Delete.
24492 (spe_func_has_64bit_regs_p): Delete.
24493 (savres_routine_name): Delete "info" parameter. Adjust callers.
24494 (rs6000_emit_stack_reset): Ditto.
24495 (rs6000_dwarf_register_span): Delete.
24496 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
24497 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
24498 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
24500 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
24502 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
24503 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
24504 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
24505 mulv2sf3, divv2sf3): Delete expanders.
24507 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24509 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
24511 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24513 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
24514 * config/rs6000/rs6000.c: Ditto.
24516 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24518 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
24519 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
24520 comparison_operator.
24522 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24524 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
24525 * config/rs6000/rs6000.opt: Ditto.
24526 * config/rs6000/t-rtems: Ditto.
24528 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24530 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
24531 TARGET_E500_SINGLE by 0, simplify.
24532 * config/rs6000/rs6000.c: Ditto.
24533 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
24534 (spe_build_register_parallel): Delete.
24535 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
24536 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
24537 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
24538 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
24539 (E500_CONVERT): Delete.
24540 * config/rs6000/spe.md: Remove many patterns and all define_constants.
24542 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24544 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
24545 * config/rs6000/dfp.md: Ditto.
24546 (negdd2, *negdd2_fpr): Merge.
24547 (absdd2, *absdd2_fpr): Merge.
24548 (negtd2, *negtd2_fpr): Merge.
24549 (abstd2, *abstd2_fpr): Merge.
24550 * config/rs6000/e500.h: Delete file.
24551 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
24552 TARGET_FPRS by 1 and simplify.
24553 * config/rs6000/rs6000-c.c: Ditto.
24554 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
24555 TARGET_DF_SPE by 0.
24556 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
24558 * config/rs6000/rs6000.md: Ditto.
24559 (floatdidf2, *floatdidf2_fpr): Merge.
24560 (move_from_CR_gt_bit): Delete.
24561 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
24562 (E500_CR_IOR_COMPARE): Delete.
24563 (All patterns that require !TARGET_FPRS): Delete.
24564 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
24566 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24568 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
24570 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24572 * graphds.c (add_edge): Intitialize edge's attached data.
24573 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
24574 pointer parameter. Call pointed function on each edge during
24575 graph traversing. Skip traversing the edge when the function
24577 (graphds_dfs, graphds_scc): Ditto.
24578 (for_each_edge): New parameter. Pass the new parameter to callback
24580 * graphds.h (skip_edge_callback): New function pointer type.
24581 (graphds_dfs, graphds_scc): New function pointer parameter.
24582 (graphds_edge_callback, for_each_edge): New parameter.
24584 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24586 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
24587 out code checking if runtime alias check is possible to below ...
24588 Call the new function.
24589 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
24590 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
24592 2017-06-07 Marek Polacek <polacek@redhat.com>
24595 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
24596 TYPE_OVERFLOW_WRAPS check.
24598 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24600 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
24601 if versioning is required.
24602 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
24603 peeling with the check for versioning.
24605 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24607 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
24608 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
24609 Set true to new parameter if new ssa variable is defined.
24610 (vect_gen_vector_loop_niters): Refactor. Set range information
24611 for the new vector loop bound variable.
24612 (vect_do_peeling): Ditto.
24614 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24616 * tree-affine.c (ssa.h): Include header file.
24617 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
24618 has wrapping overflow behavior.
24620 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24622 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
24624 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24626 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
24627 (tree_to_aff_combination): ... here.
24629 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24631 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
24632 reg_pressure model function.
24633 (ivopts_global_cost_for_size): Delete.
24634 (determine_set_costs, iv_ca_recount_cost): Call new model function
24635 ivopts_estimate_reg_pressure.
24637 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24639 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
24640 expensive than udiv. Remove floating point cases from mod.
24642 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24644 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
24645 Increase idiv cost.
24647 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24649 * config/aarch64/aarch64.md
24650 (copysignsf3): Fix mask generation.
24652 2017-06-07 Jakub Jelinek <jakub@redhat.com>
24654 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
24656 (class dump_manager): Add register_dumps method.
24657 * dumpfile.c: Include langhooks.h.
24658 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
24659 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
24660 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
24661 (dump_manager::dump_register): Start with 512 entries instead of 32.
24662 (dump_manager::register_dumps): New method.
24663 * toplev.c (general_init): Instead of invoking register_dumps
24664 langhook, invoke register_dumps method on the dump manager.
24665 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
24668 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
24670 * doc/md.texi: Clarify the restrictions on a define_insn condition.
24671 Say that # requires an associated define_split to exist, and that
24672 the define_split must be suitable for use after register allocation.
24674 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24676 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
24677 (compute_outgoing_frequencies): Also initialize zero counts.
24678 (find_many_sub_basic_blocks): Do not produce uninitialized profile
24679 around loops; preserve more of profile when nothing changes.
24681 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
24683 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
24685 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
24686 * config/arm/arm-cpu-cdata.h: Regenerate.
24687 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
24688 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
24689 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
24690 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
24691 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
24693 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
24695 * config/arm/t-rmprofile: Likewise.
24696 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
24698 2017-06-06 David S. Miller <davem@davemloft.net>
24701 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
24702 blockage if function uses alloca.
24704 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24706 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
24707 New "uid" fields to hold pretty-print IDs of group and ref.
24708 Memory references are now identified as <group_id>:<ref_id>
24709 instead of using [random] addresses.
24710 (dump_mem_details): Simplify, no functional change.
24711 (dump_mem_ref): Simplify and make output more concise.
24712 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
24713 (find_or_create_group): Initialize group uid.
24714 (record_ref): Initialize ref uid. Improve debug output.
24715 (prune_group_by_reuse, should_issue_prefetch_p,)
24716 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
24717 (mark_nontemporal_store, determine_loop_nest_reuse):
24718 Improve debug output.
24720 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24722 * dbgcnt.def (prefetch): New debug counter.
24723 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
24724 (schedule_prefetches): Stop issueing prefetches if debug counter
24727 2017-06-06 Tom de Vries <tom@codesourcery.com>
24729 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
24730 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
24732 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24734 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
24735 Use aarch64_reg_or_zero predicate for operand 4.
24736 (aarch64_compare_and_swap<mode> define_insn_and_split):
24737 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
24738 (aarch64_store_exclusive<mode>): Likewise for operand 2.
24740 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24742 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
24743 (arm_compute_save_core_reg_mask): This.
24744 (thumb1_compute_save_reg_mask): Rename into ...
24745 (thumb1_compute_save_core_reg_mask): This.
24746 (arm_compute_save_reg0_reg12_mask): Adapt comment.
24747 (arm_compute_frame_layout): Likewise.
24749 2017-06-06 Richard Biener <rguenther@suse.de>
24751 PR tree-optimization/80974
24752 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
24753 keep or clear leaders SSA info.
24755 2017-06-06 Tom de Vries <tom@codesourcery.com>
24757 * config/nvptx/nvptx.c (split_mode_p): New function.
24758 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
24760 2017-06-06 Tom de Vries <tom@codesourcery.com>
24762 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
24764 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24767 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
24770 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24772 * shrink-wrap.c (handle_simple_exit): Update profile.
24773 (try_shrink_wrapping): Upate profile.
24775 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24777 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
24778 (tree_guess_outgoing_edge_probabilities): New.
24779 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
24780 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
24782 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24784 * ipa-split.c (split_function): Initialize return bb profile.
24786 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24788 * profile.c (compute_branch_probabilities): Also initialize
24789 EXIT_BLOCK profile.
24791 2017-06-06 Richard Biener <rguenther@suse.de>
24793 PR tree-optimization/80928
24794 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
24795 (vect_analyze_loop_operations): Properly guard analysis for
24797 (vect_transform_loop): Likewise.
24798 (vect_analyze_loop_2): Also reset SLP type on PHIs.
24799 (vect_model_induction_cost): Do not cost for pure SLP.
24800 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
24801 of induction in inner loop vectorization.
24802 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
24803 (vect_get_and_check_slp_defs): Handle vect_induction_def.
24804 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
24806 (vect_analyze_slp_cost_1): Cost induction.
24807 (vect_detect_hybrid_slp_stmts): Handle PHIs.
24808 (vect_get_slp_vect_defs): Likewise.
24809 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
24810 (vect_transform_stmt): Handle SLP reductions.
24811 * tree-vectorizer.h (vectorizable_induction): Adjust.
24813 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24815 * config/rs6000/rs6000.c (make_resolver_func): Update
24816 init_lowered_empty_function call.
24818 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
24820 * doc/invoke.texi: Document the -fprofile-abs-path option.
24821 * common.opt (fprofile-abs-path): New option.
24822 * gcov-io.h (gcov_write_filename): Declare.
24823 * gcov-io.c (gcov_write_filename): New function.
24824 * coverage.c (coverage_begin_function): Use gcov_write_filename.
24825 * profile.c (output_location): Likewise.
24827 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24829 * shring-wrap.c: Revert accidental commit.
24831 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
24833 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
24835 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24837 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
24839 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
24840 profile in callgraph edge.
24841 * profile-count.h (apply_probability): If THIS is 0, then result is 0
24842 (apply_scale): Likewise.
24843 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
24844 Also scale profile when inlining function with zero profile.
24845 (initialize_cfun): Update exit block profile even when it is zero.
24846 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
24847 when profile is read.
24849 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24851 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
24852 (CLONE_*): New constants to define the processors we can generate
24853 code for with the target_clone attribute.
24854 (rs6000_clone_map): New array to identify which clone processors
24855 the current program is running on.
24856 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
24857 target_clone attribute.
24858 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
24859 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
24860 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
24861 (cpu_expand_builtin): Add support for target_clone attribute.
24862 (rs6000_valid_attribute_p): Allow "default" attribute.
24863 (get_decl_name): New debug function to simplify printing the
24864 current function name in debugging statements.
24865 (rs6000_clone_priority): New functions to support the target_clone
24866 attribute, and be able to generate code to switch between ISA 2.05
24867 through ISA 3.0 (power6 through power9).
24868 (rs6000_compare_version_priority): Likewise.
24869 (rs6000_get_function_versions_dispatcher): Likewise.
24870 (make_resolver_func): Likewise.
24871 (add_condition_to_bb): Likewise.
24872 (dispatch_function_versions): Likewise.
24873 (rs6000_generate_version_dispatcher_body): Likewise.
24874 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
24875 (fusion_gpr_load_p): Fix a spacing issue.
24876 * doc/extend.texi (Common Function Attributes): Document that the
24877 PowerPC supports the target_clone attribute.
24879 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24881 * config/arm/arm.h: explain F symbol found in description of ARM
24882 register allocation in its legend.
24884 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24886 * config/mips/frame-header-opt.c: Include profile-count.h.
24887 * config/riscv/riscv.c: Include profile-count.h
24889 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24891 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
24893 (sm_set_flag_if_changed): Add bbs field.
24894 (execute_sm_if_changed_flag_set): Pass BBS.
24895 (execute_sm): Update.
24897 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24899 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
24902 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24904 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
24905 (peephole2): New peephole2 to emit the above.
24906 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
24908 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24910 * config/aarch64/aarch64.c (define_peephole2 above
24911 *sub_<shift>_<mode>): New peephole.
24913 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24915 * config/i386/i386.c (make_resolver_func): Update.
24916 * Makefile.in: Add profile-count.h and profile-count.o
24917 * auto-profile.c (afdo_indirect_call): Update to new API.
24918 (afdo_set_bb_count): Update.
24919 (afdo_propagate_edge): Update.
24920 (afdo_propagate_circuit): Update.
24921 (afdo_calculate_branch_prob): Update.
24922 (afdo_annotate_cfg): Update.
24923 * basic-block.h: Include profile-count.h
24924 (struct edge_def): Turn count to profile_count.
24925 (struct basic_block_def): Likewie.
24926 (REG_BR_PROB_BASE): Move to profile-count.h
24927 (RDIV): Move to profile-count.h
24928 * bb-reorder.c (max_entry_count): Turn to profile_count.
24929 (find_traces): Update.
24930 (rotate_loop):Update.
24931 (connect_traces):Update.
24932 (sanitize_hot_paths):Update.
24933 * bt-load.c (migrate_btr_defs): Update.
24934 * cfg.c (RDIV): Remove.
24935 (init_flow): Use alloc_block.
24936 (alloc_block): Uninitialize count.
24937 (unchecked_make_edge): Uninitialize count.
24938 (check_bb_profile): Update.
24939 (dump_edge_info): Update.
24940 (dump_bb_info): Update.
24941 (update_bb_profile_for_threading): Update.
24942 (scale_bbs_frequencies_int): Update.
24943 (scale_bbs_frequencies_gcov_type): Update.
24944 (scale_bbs_frequencies_profile_count): New.
24945 * cfg.h (update_bb_profile_for_threading): Update.
24946 (scale_bbs_frequencies_profile_count): Declare.
24947 * cfgbuild.c (compute_outgoing_frequencies): Update.
24948 (find_many_sub_basic_blocks): Update.
24949 * cfgcleanup.c (try_forward_edges): Update.
24950 (try_crossjump_to_edge): Update.
24951 * cfgexpand.c (expand_gimple_tailcall): Update.
24952 (construct_exit_block): Update.
24953 * cfghooks.c (verify_flow_info): Update.
24954 (dump_bb_for_graph): Update.
24955 (split_edge): Update.
24956 (make_forwarder_block): Update.
24957 (duplicate_block): Update.
24958 (account_profile_record): Update.
24959 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
24960 (get_estimated_loop_iterations): Update.
24961 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
24962 (single_likely_exit): Update.
24963 * cfgloopmanip.c (scale_loop_profile): Update.
24965 (set_zero_probability): Update.
24966 (lv_adjust_loop_entry_edge): Update.
24967 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24968 (purge_dead_edges): Update.
24969 (rtl_account_profile_record): Update.
24970 * cgraph.c (cgraph_node::create): Uninitialize count.
24971 (symbol_table::create_edge): Uninitialize count.
24972 (cgraph_update_edges_for_call_stmt_node): Update.
24973 (cgraph_edge::dump_edge_flags): Update.
24974 (cgraph_node::dump): Update.
24975 (cgraph_edge::maybe_hot_p): Update.
24976 * cgraph.h: Include profile-count.h
24977 (create_clone), create_edge, create_indirect_edge): Update.
24978 (cgraph_node): Turn count to profile_count.
24979 (cgraph_edge0: Likewise.
24980 (make_speculative, clone): Update.
24981 (create_edge): Update.
24982 (init_lowered_empty_function): Update.
24983 * cgraphclones.c (cgraph_edge::clone): Update.
24984 (duplicate_thunk_for_node): Update.
24985 (cgraph_node::create_clone): Update.
24986 * cgraphunit.c (cgraph_node::analyze): Update.
24987 (cgraph_node::expand_thunk): Update.
24988 * final.c (dump_basic_block_info): Update.
24989 * gimple-streamer-in.c (input_bb): Update.
24990 * gimple-streamer-out.c (output_bb): Update.
24991 * graphite.c (print_global_statistics): Update.
24992 (print_graphite_scop_statistics): Update.
24993 * hsa-brig.c: Include basic-block.h.
24994 * hsa-dump.c: Include basic-block.h.
24995 * hsa-gen.c (T sum_slice): Update.
24996 (convert_switch_statements):Update.
24997 * hsa-regalloc.c: Include basic-block.h.
24998 * ipa-chkp.c (chkp_produce_thunks): Update.
24999 * ipa-cp.c (struct caller_statistics): Update.
25000 (init_caller_stats): Update.
25001 (gather_caller_stats): Update.
25002 (ipcp_cloning_candidate_p): Update.
25003 (good_cloning_opportunity_p): Update.
25004 (get_info_about_necessary_edges): Update.
25005 (dump_profile_updates): Update.
25006 (update_profiling_info): Update.
25007 (update_specialized_profile): Update.
25008 (perhaps_add_new_callers): Update.
25009 (decide_about_value): Update.
25010 (ipa_cp_c_finalize): Update.
25011 * ipa-devirt.c (struct odr_type_warn_count): Update.
25012 (struct decl_warn_count): Update.
25013 (struct final_warning_record): Update.
25014 (possible_polymorphic_call_targets): Update.
25015 (ipa_devirt): Update.
25016 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25017 * ipa-icf.c (sem_function::merge): Update.
25018 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
25019 * ipa-inline.c (compute_uninlined_call_time): Update.
25020 (compute_inlined_call_time): Update.
25021 (want_inline_small_function_p): Update.
25022 (want_inline_self_recursive_call_p): Update.
25023 (edge_badness): Update.
25024 (lookup_recursive_calls): Update.
25025 (recursive_inlining): Update.
25026 (inline_small_functions): Update.
25027 (dump_overall_stats): Update.
25028 (dump_inline_stats): Update.
25029 * ipa-profile.c (ipa_profile_generate_summary): Update.
25030 (ipa_propagate_frequency): Update.
25031 (ipa_profile): Update.
25032 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25033 * ipa-utils.c (ipa_merge_profiles): Update.
25034 * loop-doloop.c (doloop_modify): Update.
25035 * loop-unroll.c (report_unroll): Update.
25036 (unroll_loop_runtime_iterations): Update.
25037 * lto-cgraph.c (lto_output_edge): Update.
25038 (lto_output_node): Update.
25039 (input_node): Update.
25040 (input_edge): Update.
25041 (merge_profile_summaries): Update.
25042 * lto-streamer-in.c (input_cfg): Update.
25043 * lto-streamer-out.c (output_cfg): Update.
25044 * mcf.c (create_fixup_graph): Update.
25045 (adjust_cfg_counts): Update.
25046 (sum_edge_counts): Update.
25047 * modulo-sched.c (sms_schedule): Update.
25048 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
25049 * predict.c (maybe_hot_count_p): Update.
25050 (probably_never_executed): Update.
25051 (dump_prediction): Update.
25052 (combine_predictions_for_bb): Update.
25053 (propagate_freq): Update.
25054 (handle_missing_profiles): Update.
25055 (counts_to_freqs): Update.
25056 (rebuild_frequencies): Update.
25057 (force_edge_cold): Update.
25058 * predict.h: Include profile-count.h
25059 (maybe_hot_count_p, counts_to_freqs): UPdate.
25060 * print-rtl-function.c: Do not include cfg.h
25061 * print-rtl.c: Include basic-block.h
25062 * profile-count.c: New file.
25063 * profile-count.h: New file.
25064 * profile.c (is_edge_inconsistent): Update.
25065 (correct_negative_edge_counts): Update.
25066 (is_inconsistent): Update.
25067 (set_bb_counts): Update.
25068 (read_profile_edge_counts): Update.
25069 (compute_frequency_overlap): Update.
25070 (compute_branch_probabilities): Update; Initialize and deinitialize
25072 (branch_prob): Update.
25073 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
25074 (edge_gcov_count): New.
25075 (bb_gcov_count): New.
25076 * shrink-wrap.c (try_shrink_wrapping): Update.
25077 * tracer.c (better_p): Update.
25078 * trans-mem.c (expand_transaction): Update.
25079 (ipa_tm_insert_irr_call): Update.
25080 (ipa_tm_insert_gettmclone_call): Update.
25081 * tree-call-cdce.c: Update.
25082 * tree-cfg.c (gimple_duplicate_sese_region): Update.
25083 (gimple_duplicate_sese_tail): Update.
25084 (gimple_account_profile_record): Update.
25085 (execute_fixup_cfg): Update.
25086 * tree-inline.c (copy_bb): Update.
25087 (copy_edges_for_bb): Update.
25088 (initialize_cfun): Update.
25089 (freqs_to_counts): Update.
25090 (copy_cfg_body): Update.
25091 (expand_call_inline): Update.
25092 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25093 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
25094 (try_unroll_loop_completely): Update.
25095 (try_peel_loop): Update.
25096 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25097 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
25098 * tree-ssa-loop-split.c (connect_loops): Update.
25099 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
25100 * tree-ssa-reassoc.c (branch_fixup): Update.
25101 * tree-ssa-tail-merge.c (replace_block_by): Update.
25102 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
25103 (compute_path_counts): Update.
25104 (update_profile): Update.
25105 (recompute_probabilities): Update.
25106 (update_joiner_offpath_counts): Update.
25107 (estimated_freqs_path): Update.
25108 (freqs_to_counts_path): Update.
25109 (clear_counts_path): Update.
25110 (ssa_fix_duplicate_block_edges): Update.
25111 (duplicate_thread_path): Update.
25112 * tree-switch-conversion.c (case_bit_test_cmp): Update.
25113 (struct switch_conv_info): Update.
25114 * tree-tailcall.c (decrease_profile): Update.
25115 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25116 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25117 * value-prof.c (check_counter): Update.
25118 (gimple_divmod_fixed_value): Update.
25119 (gimple_mod_pow2): Update.
25120 (gimple_mod_subtract): Update.
25121 (gimple_ic_transform): Update.
25122 (gimple_stringop_fixed_value): Update.
25123 * value-prof.h (gimple_ic): Update.
25125 2017-06-02 Carl Love <cel@us.ibm.com>
25127 * config/rs6000/rs6000-c: Add support for built-in functions
25128 vector double vec_doublee (vector signed int);
25129 vector double vec_doublee (vector unsigned int);
25130 vector double vec_doublee (vector float);
25131 vector double vec_doubleh (vector signed int);
25132 vector double vec_doubleh (vector unsigned int);
25133 vector double vec_doubleh (vector float);
25134 vector double vec_doublel (vector signed int);
25135 vector double vec_doublel (vector unsigned int);
25136 vector double vec_doublel (vector float);
25137 vector double vec_doubleo (vector signed int);
25138 vector double vec_doubleo (vector unsigned int);
25139 vector double vec_doubleo (vector float);.
25140 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
25141 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
25143 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
25144 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
25145 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
25147 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
25148 vec_doublel, vec_doubleh.
25149 * doc/extend.texi: Update the built-in documentation file for the
25150 new built-in functions.
25152 2017-06-02 David Malcolm <dmalcolm@redhat.com>
25155 * ipa-inline-analysis.c (free_growth_caches): Set
25156 edge_removal_hook_holder to NULL after removing it.
25158 2017-06-02 Sudakshina Das <sudi.das@arm.com>
25160 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
25161 comparision with zero.
25163 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25164 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25165 for early expansion of vec_min and vec_max builtins.
25166 (builtin_function_type): Add min/max unsigned variants to those
25167 identified as having unsigned arguments.
25169 2017-06-02 Olivier Hainque <hainque@adacore.com>
25171 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
25173 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25175 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
25176 Use VALL_F16 iterator rather than VALL.
25178 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25180 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
25181 Emit CBNZ inside loop when doing a strong exchange and comparing
25182 against zero. Generate the CC flags after the loop.
25184 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
25186 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
25187 (dl_section_ref): New.
25188 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
25189 On AIX, append an expression to subtract the size of the
25190 section length to dl_section_ref.
25192 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25194 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25195 for early expansion of vector absolute builtins.
25197 2017-06-02 Richard Biener <rguenther@suse.de>
25199 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
25200 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
25202 2017-06-02 Richard Biener <rguenther@suse.de>
25204 PR tree-optimization/80948
25205 * tree-tailcall.c (find_tail_calls): Track stmts to move in
25206 stmt order as well.
25208 2017-06-02 Richard Biener <rguenther@suse.de>
25210 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
25212 * tree-vect-stmts.c (process_use): Do not mark backedge defs
25213 for inductions as relevant.
25215 2017-06-02 Richard Biener <rguenther@suse.de>
25217 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
25218 (vectorizable_induction): ... this. Remove dead code.
25220 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
25222 * builtins. (expand_builtin_alloca): Remove second parameter and
25223 infer its value from the first parameter instead.
25224 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
25226 2017-06-02 Jakub Jelinek <jakub@redhat.com>
25228 PR rtl-optimization/80903
25229 * loop-doloop.c (add_test): Unshare sequence.
25231 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
25233 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
25235 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
25237 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
25239 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
25240 xlogue_layout::get_instance, logue_layout::xlogue_layout,
25241 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
25242 (xlogue_layout::get_stub_rtx): Make static.
25243 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
25244 (xlogue_layout::compute_stub_managed_regs): Rename to...
25245 (xlogue_layout::count_stub_managed_regs): ...this.
25246 (xlogue_layout::is_stub_managed_reg): New function.
25247 (xlogue_layout::m_stub_names): Rename to...
25248 (xlogue_layout::s_stub_names): ...this, make static.
25249 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
25250 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
25251 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
25252 xlogue_layout::s_stub_names): Instantiate statics.
25253 (stub_managed_regs): Remove.
25254 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
25255 (disable_call_ms2sysv_xlogues): Rename to...
25256 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
25257 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
25259 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
25260 change after reload_completed.
25261 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
25263 (ix86_expand_prologue): Likewise.
25264 (ix86_expand_epilogue): Likewise.
25265 (ix86_expand_split_stack_prologue): Likewise.
25266 (ix86_compute_frame_layout): Remove frame parameter ...
25267 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
25268 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
25270 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
25271 (ix86_frame): Move from here ...
25272 * config/i386/i386.h (ix86_frame): ... to here.
25273 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
25274 complete ix86_frame data structure instead. Remove some_ld_name.
25276 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25278 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
25279 symbols that hold a DECL_VALUE_EXPR.
25281 2017-06-01 Martin Jambor <mjambor@suse.cz>
25283 PR tree-optimization/80898
25284 * tree-sra.c (process_subtree_disqualification): Removed.
25285 (disqualify_candidate): Do not acll
25286 process_subtree_disqualification.
25287 (subtree_mark_written_and_enqueue): New function.
25288 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
25289 RHS has been disqualified and re-queue LHS if necessary. Apart
25290 from that, ignore disqualified RHS.
25292 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25294 * config/s390/s390.c (s390_emit_epilogue): Disable early return
25295 address fetch for z10 or later.
25297 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25299 * config/arc/arc.md (tst_movb): Add guard when splitting.
25301 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25303 * config/arc/arc.c (arc_can_eliminate): Test against
25304 arc_frame_pointer_needed.
25306 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25308 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
25309 to prevent store reordering.
25310 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
25311 (type): Add block type.
25312 (stack_tie): Define special instruction to be used in
25315 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25317 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
25318 constraint. It is not valid for the pattern.
25319 (noncommutative_binary_comparison): Likewise.
25321 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25323 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
25326 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25328 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
25329 be used by the reg-alloc.
25331 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25333 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
25334 reg-alloc when having mul64 or mul32x16 instructions.
25335 (mulsidi3): Likewise.
25336 (umulsidi3): Likewise.
25337 (mulsi32x16): New pattern.
25338 (mulsi64): Likewise.
25339 (mulsidi64): Likewise.
25340 (umulsidi64): Likewise.
25341 (MUL32x16_REG): Define.
25342 (mul64_600): Use MUL32x16_REG.
25343 (mac64_600): Likewise.
25344 (umul64_600): Likewise.
25345 (umac64_600): Likewise.
25347 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25349 * config/arc/arc.md (mulsi3_700): Make it commutative.
25351 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
25353 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
25355 (*sign_extendsidi2_insn): Likewise for movstosw.
25357 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25359 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
25360 the type of the input discriminant value. Convert the
25361 discriminant value of signedness vary.
25363 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
25365 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
25366 Add to -Wall section.
25368 2017-06-01 Richard Biener <rguenther@suse.de>
25370 PR middle-end/66313
25371 * fold-const.c (fold_plusminus_mult_expr): If the factored
25372 factor may be zero use a wrapping type for the inner operation.
25373 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
25374 and handle moved defs.
25375 (process_assignment): Properly guard the unary op case. Return a
25376 tri-state indicating that moving the stmt before the call may allow
25377 to continue. Pass through to_move.
25378 (find_tail_calls): Handle moving unrelated defs before
25381 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
25384 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
25385 splitter result in the canonical way.
25387 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
25389 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
25390 also for 32bit target. Update insn attributes.
25391 (zero-extendsidi2 splitter): Allow all registers for operand 1.
25393 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
25395 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
25396 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
25397 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
25398 (_mm_maskz_min_ss): New intrinsics.
25400 2017-05-31 Martin Liska <mliska@suse.cz>
25402 * tree-vect-loop.c (vect_create_epilog_for_reduction):
25403 Change comment style to one we normally use.
25404 (vectorizable_reduction): Likewise.
25405 (vectorizable_induction): Likewise.
25406 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
25407 (vectorizable_call): Likewise.
25408 (vectorizable_simd_clone_call): Likewise.
25409 (vectorizable_conversion): Likewise.
25410 (vectorizable_assignment): Likewise.
25411 (vectorizable_shift): Likewise.
25412 (vectorizable_operation): Likewise.
25413 (vectorizable_store): Likewise.
25414 (vectorizable_load): Likewise.
25415 * tree-vectorizer.h: Likewise.
25417 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
25419 * passes.c (emergency_dump_function): New.
25420 * tree-pass.h (emergency_dump_function): Declare.
25421 * plugin.c (plugins_internal_error_function): Remove.
25422 * plugin.h (plugins_internal_error_function): Remove declaration.
25423 * toplev.c (internal_error_function): New static function. Use it...
25424 (general_init): ...here.
25426 2017-05-31 Graham Markall <graham.markall@embecosm.com>
25428 * config/arc/arc.c (arc_print_operand): Handle constant operands.
25429 (arc_rtx_costs): Add costs for new patterns.
25430 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
25431 * config/arc/predicates.md: Add _1_2_3_operand predicate.
25433 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25435 * tree-ssa-strlen.c (get_next_strinfo): New function.
25436 (get_stridx_plus_constant): Use it.
25437 (zero_length_string): Likewise.
25438 (adjust_related_strinfos): Likewise.
25439 (adjust_last_stmt): Likewise.
25441 2017-05-31 Richard Biener <rguenther@suse.de>
25444 * config/i386/i386.c (ix86_expand_builtin): Remove assert
25445 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
25447 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25449 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
25450 loop_vinfo argument and use of dependence distance vectors.
25451 Check instead whether the two references differ only in their
25452 initial value and assume that they have the same alignment if the
25453 difference is a multiple of the vector alignment.
25454 (vect_analyze_data_refs_alignment): Update call accordingly.
25456 2017-05-31 Martin Liska <mliska@suse.cz>
25459 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
25461 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25463 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
25464 (create_intersect_range_checks): Move from ...
25465 * tree-data-ref.c (create_intersect_range_checks_index)
25466 (create_intersect_range_checks): ... to here.
25467 (create_runtime_alias_checks): New function factored from ...
25468 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
25469 here. Call above function.
25470 * tree-data-ref.h (create_runtime_alias_checks): New function.
25472 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25474 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
25475 segment length for dr_b and compute it in wide_int.
25477 2017-05-31 Richard Biener <rguenther@suse.de>
25479 PR tree-optimization/80906
25480 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
25481 and pass through iv_map.
25482 (copy_bb_and_scalar_dependences): Adjust.
25483 (translate_pending_phi_nodes): Likewise.
25484 (copy_loop_close_phi_args): Handle code-generating IVs instead
25487 2017-05-30 David Malcolm <dmalcolm@redhat.com>
25489 * diagnostic-color.c (color_dict): Add "type-diff".
25490 (parse_gcc_colors): Update comment.
25491 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
25492 -fdiagnostics-show-template-tree and -fno-elide-type.
25493 (GCC_COLORS): Add type-diff to example.
25495 (-fdiagnostics-show-template-tree): New.
25496 (-fno-elide-type): New.
25497 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
25498 the pp_format_decoder callback. Call any m_format_postprocessor's
25500 (pretty_printer::pretty_printer): Initialize
25501 m_format_postprocessor.
25502 (pretty_printer::~pretty_printer): Delete any
25503 m_format_postprocessor.
25504 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
25505 (class format_postprocessor): New class.
25506 (struct pretty_printer::format_decoder): Document the new parameters.
25507 (struct pretty_printer::m_format_postprocessor): New field.
25508 * tree-diagnostic.c (default_tree_printer): Update for new
25509 bool and const char ** params.
25510 * tree-diagnostic.h (default_tree_printer): Likewise.
25512 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
25514 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
25515 (lwa_operand): Delete rs6000_gen_cell_microcode test.
25516 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25517 rs6000_gen_cell_microcode code.
25518 (rs6000_final_prescan_insn): Delete.
25519 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
25520 "warn-cell-microcode" entries.
25521 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
25522 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
25523 throughout. Change cc_reg_not_micro_cr0_operand to
25524 cc_reg_not_cr0_operand throughout.
25525 (*extendhi<mode>2_noload): Delete.
25526 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
25527 (mwarn-cell-microcode): Delete.
25528 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
25529 -mgen-cell-microcode and -mwarn-cell-microcode.
25531 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
25534 * config/i386/constraints.md (Yd): New constraint.
25536 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
25537 and (?Yd, r) alternatives. Update insn attributes.
25538 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
25539 and (?*Yd, r) alternatives. Update insn attributes.
25540 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
25542 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
25544 * gimplify.c (gimplify_modify_expr): Don't create a
25545 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
25548 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
25550 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
25552 2017-05-30 Richard Biener <rguenther@suse.de>
25554 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
25555 and reduc_def fields.
25556 (STMT_VINFO_REDUC_TYPE): New define.
25557 (STMT_VINFO_REDUC_DEF): Likewise.
25558 (vect_force_simple_reduction): Adjust prototype.
25559 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
25560 (vect_is_simple_reduction): Remove check_reduction argument.
25561 (vect_force_simple_reduction): Adjust and set
25562 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25563 (vectorizable_reduction): Do not re-do reduction analysis
25564 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25565 * tree-parloops.c (gather_scalar_reductions): Adjust.
25567 2017-05-30 Richard Biener <rguenther@suse.de>
25569 PR middle-end/80901
25570 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
25573 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
25575 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
25576 Introduce unknown_misalignment parameter and remove vf.
25577 (vect_peeling_hash_get_lowest_cost):
25578 Pass unknown_misalignment parameter.
25579 (vect_enhance_data_refs_alignment):
25580 Fix unsupportable data ref treatment.
25582 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25584 * tree-vect-data-refs.c (vect_get_data_access_cost):
25585 Workaround for SLP handling.
25586 (vect_enhance_data_refs_alignment):
25587 Compute costs for doing no peeling at all, compare to the best
25588 peeling costs so far and avoid peeling if cheaper.
25590 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25592 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
25593 Return peeling info and set costs to zero for unlimited cost
25595 (vect_enhance_data_refs_alignment): Also inspect all datarefs
25596 with unknown misalignment. Compute and costs for unknown
25597 misalignment, compare them to the costs for known misalignment
25598 and choose the cheapest for peeling.
25600 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25602 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
25603 (vect_get_peeling_costs_all_drs): Create function.
25604 (vect_peeling_hash_get_lowest_cost):
25605 Use vect_get_peeling_costs_all_drs.
25606 (vect_peeling_supportable): Create function.
25607 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
25609 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25611 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
25612 DR_HAS_NEGATIVE_STEP.
25613 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
25614 (vect_enhance_data_refs_alignment): Use.
25615 (vect_duplicate_ssa_name_ptr_info): Use.
25616 * tree-vectorizer.h (dr_misalignment): Use.
25617 (known_alignment_for_access_p): Use.
25619 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25622 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
25624 (has_section_name): New function.
25626 2017-05-30 Martin Liska <mliska@suse.cz>
25629 * auto-profile.c (get_function_decl_from_block): Fix
25632 2017-05-30 Richard Biener <rguenther@suse.de>
25634 PR middle-end/80876
25635 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
25637 2017-05-30 Martin Liska <mliska@suse.cz>
25639 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
25640 * dumpfile.h (struct dump_file_info): Remove ctors.
25642 2017-05-30 Martin Liska <mliska@suse.cz>
25644 * predict.def: Fix GNU coding style.
25646 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25648 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25649 Mark 'to' argument with ATTRIBUTE_UNUSED.
25651 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25653 * config/xtensa/xtensa.c (xtensa_emit_call): Use
25654 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
25655 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
25658 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
25660 * doc/install.texi (Options specification): Restore entry of
25661 --enable-sjlj-exceptions.
25663 2017-05-27 Michael Eager <eager@eagercon.com>
25666 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
25668 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
25670 * config/microblaze/microblaze.h
25671 (FIXED_REGISTERS): Update in macro.
25672 (CALL_USED_REGISTERS): Update in macro.
25674 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
25676 * doc/install.texi: Add links to macOS binary distributions.
25678 2017-05-27 Jakub Jelinek <jakub@redhat.com>
25682 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25684 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25686 2017-05-26 Martin Liska <mliska@suse.cz>
25688 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
25690 2017-05-26 Martin Liska <mliska@suse.cz>
25692 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
25693 always leading ';; '.
25694 (dump_bb_info): Likewise.
25695 (brief_dump_cfg): Likewise.
25696 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
25697 * dumpfile.c: Remove usage of TDF_VERBOSE.
25698 * dumpfile.h (enum dump_kind): Likewise.
25699 (dump_gimple_bb_header): Do not use TDF_COMMENT.
25700 * print-tree.c (debug_verbose): Remove.
25701 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
25702 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
25703 * tree-diagnostic.c (default_tree_printer): Replace
25704 TDF_DIAGNOSTIC with TDF_SLIM.
25706 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25708 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
25709 in parameter loop, rather than loop_vinfo.
25710 (create_intersect_range_checks): Ditto.
25711 (vect_create_cond_for_alias_checks): Update call to above functions.
25713 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25715 PR tree-optimization/80815
25716 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
25717 for merging runtime alias checks. Handle negative DR_STEPs.
25719 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25721 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
25723 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
25724 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
25725 out code pruning runtime alias checks.
25726 * tree-data-ref.c (prune_runtime_alias_test_list): New function
25727 factored out from above.
25728 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25730 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25732 (prune_runtime_alias_test_list): New decalaration.
25734 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25736 * tree-vect-data-refs.c (compare_tree): Rename and move ...
25737 * tree-data-ref.c (data_ref_compare_tree): ... to here.
25738 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
25739 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
25740 (operator==, comp_dr_with_seg_len_pair): Ditto.
25741 (vect_prune_runtime_alias_test_list): Ditto.
25743 2017-05-26 Martin Liska <mliska@suse.cz>
25746 * params.def: Bound partial-inlining-entry-probability param.
25748 2017-05-26 Marek Polacek <polacek@redhat.com>
25751 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
25754 2017-05-26 Richard Biener <rguenther@suse.de>
25756 PR tree-optimization/80842
25757 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
25760 2017-05-26 Richard Biener <rguenther@suse.de>
25762 PR tree-optimization/80844
25763 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
25765 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
25767 * doc/md.texi (Machine Constraints): Update x86 family
25768 machine constraints section to match 'config/i386/constraints.md'.
25770 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
25772 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
25774 2017-05-25 Nathan Sidwell <nathan@acm.org>
25776 * doc/invoke.texi (--enable-languages): Update documentation.
25778 2017-05-25 Martin Liska <mliska@suse.cz>
25780 * dumpfile.c: Add TDF_FOLDING.
25781 * dumpfile.h (enum dump_kind): Likewise.
25782 * genmatch.c (dt_simplify::gen_1): Use it.
25784 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25786 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
25788 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25790 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25791 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
25793 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25795 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
25796 * match.pd (X == C): Rewrite it here.
25797 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
25798 with_certain_nonzero_bits2): New predicates.
25799 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
25801 2017-05-24 Nathan Sidwell <nathan@acm.org>
25803 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
25806 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
25809 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25811 * config/powerpcspe: New port. Files are copied from the rs6000
25812 port, with "rs6000" in filenames replaced by "powerpcspe".
25814 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
25816 PR rtl-optimization/80754
25817 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
25819 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
25821 * config/sparc/sparc.md (length): Return the correct value for -mflat
25822 sibcalls to match output_sibcall.
25824 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25828 * config/rs6000/rs6000.c (struct machine_function): Add new field
25830 (rs6000_get_separate_components): Init that field, use it.
25831 (rs6000_components_for_bb): Use the field.
25833 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
25835 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
25837 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
25839 PR middle-end/80823
25840 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
25842 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25845 * config/s390/s390.c (s390_check_qrst_address): Check incoming
25846 address against address_operand predicate.
25847 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
25849 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
25851 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
25852 parameters passed indirectly.
25854 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
25856 * config/i386/i386.md (*movdi_internal): Remove SSE4
25857 alternative 18 (?r, *v). Update insn attributes.
25858 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
25859 Update insn attributes.
25860 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
25861 Update insn attributes.
25862 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
25863 alternative 1 (r, v). Remove isa attribute.
25864 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
25865 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
25866 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
25868 2017-05-23 Tom de Vries <tom@codesourcery.com>
25870 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
25873 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25875 * cgraphunit.c (symbol_table::process_new_functions): Update.
25876 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
25877 (inline_generate_summary): Rename to ...
25878 (ipa_fn_summary_generate): ... this one.
25879 (inline_read_summary): Rename to ...
25880 (ipa_fn_summary_read): ... this one.
25881 (inline_write_summary): Rename to ...
25882 (ipa_fn_summary_write): ... this one.
25883 (inline_free_summary): Rename to ...
25884 (ipa_free_fn_summary): ... this one.
25885 (pass_data_local_fn_summary, pass_local_fn_summary,
25886 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
25887 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
25888 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
25889 make_pass_ipa_fn_summary): New.
25890 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
25891 inline_write_summary, inline_free_summary): Remove.
25892 (ipa_free_fn_summary) : New.
25893 * ipa-inline.c (ipa_inline): Update.
25894 (pass_ipa_inline): Do not generate summaries.
25895 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
25897 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
25898 and add pass_ipa_fn_summary.
25899 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
25901 (make_pass_inline_parameters): Remove.
25903 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
25905 * omp-low.c (struct omp_context): Remove "default_kind" member.
25908 * omp-offload.c (execute_oacc_device_lower): Remove the
25909 parallelism dimensions function attributes for unparallelized
25910 OpenACC kernels constructs.
25912 2017-05-23 Martin Liska <mliska@suse.cz>
25914 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
25916 (cgraph_edge::make_speculative): Likewise.
25917 (cgraph_edge::resolve_speculation): Likewise.
25918 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25919 (cgraph_node::dump): Likewise.
25920 * cgraph.h: Likewise.
25921 * cgraphunit.c (analyze_functions): Likewise.
25922 (symbol_table::compile): Likewise.
25923 * ipa-cp.c (print_all_lattices): Likewise.
25924 (determine_versionability): Likewise.
25925 (initialize_node_lattices): Likewise.
25926 (ipcp_verify_propagated_values): Likewise.
25927 (estimate_local_effects): Likewise.
25928 (update_profiling_info): Likewise.
25929 (create_specialized_node): Likewise.
25930 (perhaps_add_new_callers): Likewise.
25931 (decide_about_value): Likewise.
25932 (decide_whether_version_node): Likewise.
25933 (identify_dead_nodes): Likewise.
25934 (ipcp_store_bits_results): Likewise.
25935 * ipa-devirt.c (dump_targets): Likewise.
25936 (ipa_devirt): Likewise.
25937 * ipa-icf.c (sem_item::dump): Likewise.
25938 (sem_function::equals): Likewise.
25939 (sem_variable::equals): Likewise.
25940 (sem_item_optimizer::read_section): Likewise.
25941 (sem_item_optimizer::execute): Likewise.
25942 (congruence_class::dump): Likewise.
25943 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
25944 (dump_inline_summary): Likewise.
25945 (estimate_node_size_and_time): Likewise.
25946 (inline_analyze_function): Likewise.
25947 * ipa-inline-transform.c (inline_call): Likewise.
25948 * ipa-inline.c (report_inline_failed_reason): Likewise.
25949 (want_early_inline_function_p): Likewise.
25950 (edge_badness): Likewise.
25951 (update_edge_key): Likewise.
25952 (inline_small_functions): Likewise.
25953 * ipa-profile.c (ipa_profile): Likewise.
25954 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
25955 (ipa_make_edge_direct_to_target): Likewise.
25956 (remove_described_reference): Likewise.
25957 (ipa_impossible_devirt_target): Likewise.
25958 (propagate_controlled_uses): Likewise.
25959 (ipa_print_node_params): Likewise.
25960 (ipcp_transform_function): Likewise.
25961 * ipa-pure-const.c (pure_const_read_summary): Likewise.
25962 (propagate_pure_const): Likewise.
25963 * ipa-reference.c (generate_summary): Likewise.
25964 (read_write_all_from_decl): Likewise.
25965 (propagate): Likewise.
25966 (ipa_reference_read_optimization_summary): Likewise.
25967 * ipa-utils.c (ipa_merge_profiles): Likewise.
25968 * ipa.c (walk_polymorphic_call_targets): Likewise.
25969 (symbol_table::remove_unreachable_nodes): Likewise.
25970 (ipa_single_use): Likewise.
25971 * passes.c (execute_todo): Likewise.
25972 * predict.c (drop_profile): Likewise.
25973 * symtab.c (symtab_node::get_dump_name): New function.
25974 (symtab_node::dump_name): Likewise.
25975 (symtab_node::dump_asm_name): Likewise.
25976 (symtab_node::dump_references): Likewise.
25977 (symtab_node::dump_referring): Likewise.
25978 (symtab_node::dump_base): Likewise.
25979 (symtab_node::debug_symtab): Likewise.
25980 * tree-sra.c (convert_callers_for_node): Likewise.
25981 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
25982 * value-prof.c (init_node_map): Likewise.
25984 2017-05-23 Martin Liska <mliska@suse.cz>
25986 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
25987 and symtab_node::debug_symtab to symbol_table::debug.
25988 * cgraphunit.c (analyze_functions): Use the renamed function.
25989 (symbol_table::compile): Likewise.
25990 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
25991 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
25992 * passes.c (execute_todo): Likewise.
25993 * symtab.c (symbol_table::dump): New function.
25994 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
25996 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25998 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
25999 that nonconst implies exec.
26001 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26003 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26004 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26005 (inline_edge_summary_vec): Turn into ...
26006 (ipa_call_summaries): ... this one.
26007 (redirect_to_unreachable, edge_set_predicate,
26008 evaluate_properties_for_edge, inline_summary_alloc,
26009 reset_ipa_call_summary, reset_inline_summary,
26010 inline_summary_t::duplicate): Update.
26011 (inline_edge_duplication_hook): Turn to ...
26012 (ipa_call_summary_t::duplicate): ... this one.
26013 (inline_edge_removal_hook): Turn to ...
26014 (ipa_call_summary_t::remove): ... this one.
26015 (dump_inline_edge_summary): Turn to ...
26016 (dump_ipa_call_summary): ... this one.
26017 (estimate_function_body_sizes): Update.
26018 (inline_update_callee_summaries): Update.
26019 (remap_edge_change_prob): Update.
26020 (remap_edge_summaries): Update.
26021 (inline_merge_summary): Update.
26022 (do_estimate_edge_time): Update.
26023 (inline_generate_summary): Update.
26024 (inline_read_section): Update.
26025 (inline_read_summary): Update.
26026 (inline_free_summary): Update.
26027 * ipa-inline.c (can_inline_edge_p): Update.
26028 (compute_inlined_call_time): Update.
26029 (want_inline_small_function_p): Update.
26030 (edge_badness): Update.
26031 (early_inliner): Update.
26032 * ipa-inline.h (inline_edge_summary): Turn to ...
26033 (ipa_call_summary): ... this one.
26034 (ipa_call_summary_t): New class.
26035 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26036 (ipa_call_summaries): New.
26037 (inline_edge_summary): Remove.
26038 (estimate_edge_growth): Update.
26039 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26040 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26041 * ipa-split.c (execute_split_functions): Update.
26042 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26044 2017-05-23 Tom de Vries <tom@codesourcery.com>
26046 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26047 attributes): Document rdrand effective target.
26049 2017-05-23 Tom de Vries <tom@codesourcery.com>
26051 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26052 attributes): Sort alphabetically.
26054 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
26056 * config/avr/genmultilib.awk: Use gsub instead of gensub.
26058 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26061 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
26062 V2DF/V2DI splat into two separate patterns, one that handles
26063 registers, and the other that only handles memory. Drop support
26064 for splatting from a GPR on ISA 2.07 and then splitting the
26065 splat into direct move and splat.
26066 (vsx_splat_<mode>_reg): Likewise.
26067 (vsx_splat_<mode>_mem): Likewise.
26069 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
26071 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
26073 2017-05-22 Jakub Jelinek <jakub@redhat.com>
26075 PR middle-end/80809
26076 * omp-low.c (finish_taskreg_remap): New function.
26077 (finish_taskreg_scan): If unit size of ctx->record_type
26078 is non-constant, unshare the size expression and replace
26079 decls in it with possible outer var refs.
26081 PR middle-end/80809
26082 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
26083 GOVD_SHARED rather than GOVD_PRIVATE with it.
26084 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
26085 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
26087 PR middle-end/80853
26088 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
26089 as last argument to build_outer_var_ref for pointer bases of array
26090 section reductions.
26092 2017-05-19 Martin Sebor <msebor@redhat.com>
26094 * print-tree.c (print_node): Print DECL_READ_P flag.
26096 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26098 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
26099 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
26100 * cgraph.c: Likewise.
26101 * cgraphunit.c: Likewise.
26102 * gengtype.c: Likewise.
26103 * ipa-cp.c: Likewise.
26104 * ipa-devirt.c: Likewise.
26105 * ipa-icf.c: Likewise.
26106 * ipa-predicate.c: Likewise.
26107 * ipa-profile.c: Likewise.
26108 * ipa-prop.c: Likewise.
26109 * ipa-split.c: Likewise.
26111 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
26112 edge_predicate_pool, dump_inline_hints,
26113 inline_summary::account_size_time, redirect_to_unreachable,
26114 edge_set_predicate, set_hint_predicate,
26115 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
26116 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
26117 inline_summary_t::remove, remap_hint_predicate_after_duplication,
26118 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
26119 ipa_call_summary_t::remove, initialize_growth_caches,
26120 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
26121 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
26122 mark_modified, unmodified_parm_1, unmodified_parm,
26123 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
26124 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
26125 compute_bb_predicates, will_be_nonconstant_expr_predicate,
26126 will_be_nonconstant_predicate, record_modified_bb_info,
26127 get_minimal_bb, record_modified, param_change_prob,
26128 phi_result_unknown_predicate, predicate_for_phi_result,
26129 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
26130 estimate_function_body_sizes, compute_inline_parameters,
26131 compute_inline_parameters_for_curren, pass_data_inline_parameters,
26132 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
26133 inline_update_callee_summaries, remap_edge_change_prob,
26134 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
26135 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
26136 inline_analyze_function, inline_summary_t::insert,
26137 inline_generate_summary, read_ipa_call_summary, inline_read_section,
26138 inline_read_summary, write_ipa_call_summary, inline_write_summary,
26139 inline_free_summary): Move to ipa-fnsummary.h
26140 (predicate_t): Remove.
26141 * ipa-fnsummary.c: New file.
26142 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
26143 (enum inline_hints_vals, inline_hints, agg_position_info,
26144 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
26145 inline_summaries, ipa_call_summary, ipa_call_summary_t,
26146 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
26147 dump_inline_summary, dump_inline_hints, inline_generate_summary,
26148 inline_read_summary, inline_write_summary, inline_free_summary,
26149 inline_analyze_function, initialize_inline_failed,
26150 inline_merge_summary, inline_update_overall_summary,
26151 compute_inline_parameters): Move to ipa-fnsummary.h
26152 * ipa-fnsummary.h: New file.
26153 * ipa-inline-transform.h: Include ipa-inline.h.
26154 * ipa-inline.c: LIkewise.
26156 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26158 * ipa-inline.c (edge_badness): Use inlined_time instead of
26159 inline_summaries->get.
26161 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26163 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
26165 2017-05-22 Nathan Sidwell <nathan@acm.org>
26167 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
26168 (fdump-lang): Document 'raw' option.
26169 * dumpfile.h (TDI_tu): Delete.
26170 * dumpfile.c (dump_files): Remove translation-unit.
26171 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
26173 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
26175 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
26176 command option from $(AWK) call.
26177 * config/avr/genmultilib.awk: Simplify and rewrite so that it
26178 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
26179 [FORMAT]: Remove handling of variable.
26180 * config/avr/t-multilib: Regenerate.
26182 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26184 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
26186 (dump_inline_summary): Do not print self_time.
26187 (estimate_function_body_sizes): Do not set self_time.
26188 (compute_inline_parameters): Likewise.
26189 (inline_read_section, inline_write_summary): Do not stream self_time.
26190 * ipa-inline.h (inline_summary): Drop self_time.
26192 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26194 * ipa-inline-analysis.c (account_size_time): Rename to ...
26195 (inline_summary::account_size_time): ... this one.
26196 (reset_ipa_call_summary): Turn to ...
26197 (ipa_call_summary::reset): ... this one.
26198 (reset_inline_summary): Turn to ...
26199 (inline_summary::reset): ... this one.
26200 (inline_summary_t::remove): Update.
26201 (inline_summary_t::duplicate): Update.
26202 (ipa_call_summary_t::remove): Update.
26203 (dump_inline_summary): Update.
26204 (estimate_function_body_sizes): Update.
26205 (compute_inline_parameters): Update.
26206 (estimate_node_size_and_time): Update.
26207 (inline_merge_summary): Update.
26208 (inline_update_overall_summary): Update.
26209 (inline_read_section): Update.
26210 (inline_write_summary): Update.
26211 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
26212 add account_size_time and reset member functions.
26213 (ipa_call_summary): Add reset function.
26214 * ipa-predicate.h (predicate::operator &): Constify.
26216 2017-05-22 Richard Biener <rguenther@suse.de>
26218 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
26220 2017-05-19 Jason Merrill <jason@redhat.com>
26222 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
26224 2017-05-19 Marek Polacek <polacek@redhat.com>
26227 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
26228 TYPE_OVERFLOW_WRAPS checks.
26230 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
26232 * tree-core.h (enum omp_clause_default_kind): Add
26233 "OMP_CLAUSE_DEFAULT_PRESENT".
26234 * tree-pretty-print.c (dump_omp_clause): Handle it.
26235 * gimplify.c (enum gimplify_omp_var_data): Add
26236 "GOVD_MAP_FORCE_PRESENT".
26237 (gimplify_adjust_omp_clauses_1): Map it to
26238 "GOMP_MAP_FORCE_PRESENT".
26239 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
26241 * gimplify.c (oacc_default_clause): Clarify.
26243 2017-05-19 Nathan Sidwell <nathan@acm.org>
26245 LANG_HOOK_REGISTER_DUMPS
26246 * toplev.c (general_init): Call register dump lang hook.
26247 * doc/invoke.texi: Document -fdump-lang option family.
26248 * dumpfile.c (dump_files): Remove class dump here.
26249 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
26250 * dumpfile.h (tree_dump_index): Remove TDI_class.
26251 * langhooks-def.h (lhd_register_dumps): Declare.
26252 (LANG_HOOKS_REGISTER_DUMPS): Define.
26253 (LANG_HOOKS_INITIALIZER): Add it.
26254 * langhooks.c (lhd_register_dumps): Define.
26255 * langhooks.h (struct lang_hooks): Add register_dumps.
26257 2017-05-19 Nathan Sidwell <nathan@acm.org>
26259 * context.h (context::set_passes): New.
26260 * context.c (context::context): Do not create pass manager.
26261 * toplev.c (general_init): Create pass manager here.
26263 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
26265 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
26266 use this splitter if two add or or instructions would also work for
26267 the constant we want to generate.
26269 2017-05-19 Richard Biener <rguenther@suse.de>
26272 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
26273 predicate evaluation.
26275 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26277 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
26279 * ipa-inline.c (want_inline_small_function_p): Do not cast to
26282 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26284 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26285 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26286 (inline_edge_summary_vec): Turn into ...
26287 (ipa_call_summaries): ... this one.
26288 (redirect_to_unreachable, edge_set_predicate,
26289 evaluate_properties_for_edge, inline_summary_alloc,
26290 reset_ipa_call_summary, reset_inline_summary,
26291 inline_summary_t::duplicate): Update.
26292 (inline_edge_duplication_hook): Turn to ...
26293 (ipa_call_summary_t::duplicate): ... this one.
26294 (inline_edge_removal_hook): Turn to ...
26295 (ipa_call_summary_t::remove): ... this one.
26296 (dump_inline_edge_summary): Turn to ...
26297 (dump_ipa_call_summary): ... this one.
26298 (estimate_function_body_sizes): Update.
26299 (inline_update_callee_summaries): Update.
26300 (remap_edge_change_prob): Update.
26301 (remap_edge_summaries): Update.
26302 (inline_merge_summary): Update.
26303 (do_estimate_edge_time): Update.
26304 (inline_generate_summary): Update.
26305 (inline_read_section): Update.
26306 (inline_read_summary): Update.
26307 (inline_free_summary): Update.
26308 * ipa-inline.c (can_inline_edge_p): Update.
26309 (compute_inlined_call_time): Update.
26310 (want_inline_small_function_p): Update.
26311 (edge_badness): Update.
26312 (early_inliner): Update.
26313 * ipa-inline.h (inline_edge_summary): Turn to ...
26314 (ipa_call_summary): ... this one.
26315 (ipa_call_summary_t): New class.
26316 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26317 (ipa_call_summaries): New.
26318 (inline_edge_summary): Remove.
26319 (estimate_edge_growth): Update.
26320 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26321 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26322 * ipa-split.c (execute_split_functions): Update.
26323 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26325 2017-05-19 Richard Biener <rguenther@suse.de>
26327 PR middle-end/80764
26328 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
26330 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
26332 * config/rs6000/rs6000.c (struct machine_function): Add field
26333 fpr_is_wrapped_separately.
26334 (rs6000_get_separate_components): Use 64 components. Handle the
26335 new FPR components.
26336 (rs6000_components_for_bb): Handle the FPR components.
26337 (rs6000_emit_prologue_components): Handle the FPR components.
26338 (rs6000_emit_epilogue_components): Handle the FPR components.
26339 (rs6000_set_handled_components): Handle the FPR components.
26340 (rs6000_emit_prologue): Don't output prologue code for those FPRs
26341 that are already separately shrink-wrapped.
26342 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
26343 that are already separately shrink-wrapped.
26345 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26348 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
26351 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
26352 (define_peephole2 for Altivec d-form load): Add peepholes to catch
26353 cases where the register allocator uses a move and an offsettable
26354 memory operation to/from a FPR register on ISA 2.06/2.07.
26355 (define_peephole2 for Altivec d-form store): Likewise.
26357 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
26360 * config/i386/mmx.md (*mov<mode>_internal): Enable
26361 alternatives 11, 12, 13 and 14 also for 32bit targets.
26362 Remove alternatives 15, 16, 17 and 18.
26363 * config/i386/sse.md (vec_concatv2di): Change
26364 alternative (!x, *y) to (x, ?!*Yn).
26366 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
26368 * dumpfile.h (enum dump_kind): Remove stray comma.
26370 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26372 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
26373 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
26374 predicate::num_conditions
26375 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
26376 (CHANGED): turn into predicate::changed.
26377 (agg_position_info): Move to ipa-predicate.h
26378 (add_condition, predicate::add_clause, predicate::operator &=,
26379 predicate::or_with, predicate::evaluate, predicate::probability,
26380 dump_condition, dump_clause, predicate::dump,
26381 predicate::remap_after_duplication, predicate::remap_after_inlining,
26382 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
26383 (evaluate_conditions_for_known_args): Update.
26384 (set_cond_stmt_execution_predicate): Update.
26385 * ipa-inline.h: Include ipa-predicate.h
26386 (condition, inline_param_summary, conditions, agg_position_info,
26387 predicate): Move to ipa-predicate.h
26388 * ipa-predicate.c: New file.
26389 * ipa-predicate.h: New file.
26391 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
26393 * final.c (leaf_function_p): Check we are not in a sequence.
26395 2017-05-18 Martin Liska <mliska@suse.cz>
26397 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
26398 * dumpfile.c (dump_register): Use new enum dump_kind.
26399 (get_dump_file_name): Likewise.
26400 (dump_enable_all): Likewise.
26401 (dump_switch_p_1): Likewise.
26402 (enable_rtl_dump_file): Remove usage of TDF_RTL.
26403 * dumpfile.h (enum dump_kind): New enum type.
26404 (struct dump_file_info): Create constructor and
26405 format fields and comments.
26406 * passes.c (pass_manager::register_one_dump_file):
26408 * statistics.c (statistics_early_init): Likewise.
26409 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
26410 TDF_TREE with TDF_SLIM.
26411 (gather_memory_references_ref): Likewise.
26413 2017-05-18 Martin Liska <mliska@suse.cz>
26415 * vec.h (struct vnull): Use it.
26417 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26419 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
26420 (true_predicate, false_predicate, true_predicate_p,
26421 false_predicate_p): Remove.
26422 (single_cond_predicate, not_inlined_predicate): Turn to member function
26424 (add_condition): Update.
26425 (add_clause): Turn to...
26426 (predicate::add_clause): ... this one; update; allow passing NULL
26428 (and_predicates): Turn to ...
26429 (predicate::operator &=): ... this one.
26430 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
26431 (or_predicates): Turn to ...
26432 (predicate::or_with): ... this one.
26433 (evaluate_predicate): Turn to ...
26434 (predicate::evaluate): ... this one.
26435 (predicate_probability): Turn to ...
26436 (predicate::probability): ... this one.
26437 (dump_condition): Update.
26438 (dump_predicate): Turn to ...
26439 (predicate::dump): ... this one.
26440 (account_size_time): Update.
26441 (edge_set_predicate): Update.
26442 (set_hint_predicate): UPdate.
26443 (evaluate_conditions_for_known_args): Update.
26444 (evaluate_properties_for_edge): Update.
26445 (remap_predicate_after_duplication): Turn to...
26446 (predicate::remap_after_duplication): ... this one.
26447 (remap_hint_predicate_after_duplication): Update.
26448 (inline_summary_t::duplicate): UPdate.
26449 (dump_inline_edge_summary): Update.
26450 (dump_inline_summary): Update.
26451 (set_cond_stmt_execution_predicate): Update.
26452 (set_switch_stmt_execution_predicate): Update.
26453 (compute_bb_predicates): Update.
26454 (will_be_nonconstant_expr_predicate): Update.
26455 (will_be_nonconstant_predicate): Update.
26456 (phi_result_unknown_predicate): Update.
26457 (predicate_for_phi_result): Update.
26458 (array_index_predicate): Update.
26459 (estimate_function_body_sizes): Update.
26460 (estimate_node_size_and_time): Update.
26461 (estimate_ipcp_clone_size_and_time): Update.
26462 (remap_predicate): Rename to ...
26463 (predicate::remap_after_inlining): ... this one.
26464 (remap_hint_predicate): Update.
26465 (inline_merge_summary): Update.
26466 (inline_update_overall_summary): Update.
26467 (estimate_size_after_inlining): Update.
26468 (read_predicate): Rename to ...
26469 (predicate::stream_in): ... this one.
26470 (read_inline_edge_summary): Update.
26471 (write_predicate): Rename to ...
26472 (predicate::stream_out): ... this one.
26473 (write_inline_edge_summary): Update.
26474 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
26475 (clause_t): Turn to uint32_t
26476 (predicate): Turn to class; implement constructor and operators
26478 (size_time_entry): Update.
26479 (inline_summary): Update.
26480 (inline_edge_summary): Update.
26482 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
26484 * fold-const.c (fold_binary_loc): Move transformation...
26485 * match.pd (C - X CMP X): ... here.
26487 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
26489 * config/sparc/sparc.c (sparc_option_override): Set function
26490 alignment for -mcpu=niagara7 to 64 to match the I$ line.
26491 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
26493 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
26495 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
26497 2017-05-18 Marek Polacek <polacek@redhat.com>
26500 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
26501 (pass_ubsan::execute): Call gimple_assign_single_p instead of
26502 gimple_assign_load_p.
26504 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
26506 PR middle-end/80692
26507 * real.c (do_compare): Give decimal_do_compare preference over
26508 comparing just the signs.
26510 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
26512 * doc/md.texi (Canonicalization of Instructions): Describe the
26513 canonical form of instructions that inherently set a condition
26516 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
26518 PR middle-end/80775
26519 * tree-cfg.c: Move deletion of unreachable case statements to after
26520 the merging of consecutive case labels.
26522 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
26524 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
26525 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
26526 restoring of callee-saved registers.
26528 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
26530 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
26531 * config/visium/visium.c (single_set_and_flags): Likewise.
26532 * config/visium/visium.md (Substitutions): Likewise.
26534 2017-05-17 Martin Liska <mliska@suse.cz>
26536 * cfg.c: Introduce dump_flags_t type and
26537 use it instead of int type.
26539 * cfghooks.c: Likewise.
26540 * cfghooks.h (struct cfg_hooks): Likewise.
26541 * cfgrtl.c: Likewise.
26542 * cfgrtl.h: Likewise.
26543 * cgraph.c (cgraph_node::get_body): Likewise.
26544 * coretypes.h: Likewise.
26545 * domwalk.c: Likewise.
26546 * domwalk.h: Likewise.
26547 * dumpfile.c (struct dump_option_value_info): Likewise.
26548 (dump_enable_all): Likewise.
26549 (dump_switch_p_1): Likewise.
26550 (opt_info_switch_p): Likewise.
26551 * dumpfile.h (enum tree_dump_index): Likewise.
26552 (struct dump_file_info): Likewise.
26553 * genemit.c: Likewise.
26554 * generic-match-head.c: Likewise.
26555 * gengtype.c (open_base_files): Likewise.
26556 * gimple-pretty-print.c: Likewise.
26557 * gimple-pretty-print.h: Likewise.
26558 * graph.c (print_graph_cfg): Likewise.
26559 * graphite-scop-detection.c (dot_all_sese): Likewise.
26560 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
26561 * loop-unroll.c (report_unroll): Likewise.
26562 * passes.c (pass_manager::register_one_dump_file): Likewise.
26563 * print-tree.c: Likewise.
26564 * statistics.c: Likewise.
26565 * tree-cfg.c: Likewise.
26566 * tree-cfg.h: Likewise.
26567 * tree-dfa.c: Likewise.
26568 * tree-dfa.h: Likewise.
26569 * tree-dump.c (dump_function): Likewise.
26570 * tree-dump.h (struct dump_info): Likewise.
26571 * tree-pretty-print.c: Likewise.
26572 * tree-pretty-print.h: Likewise.
26573 * tree-ssa-live.c: Likewise.
26574 * tree-ssa-live.h: Likewise.
26575 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
26576 * tree-vect-loop.c: Likewise.
26577 * tree-vect-slp.c: Likewise.
26579 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
26580 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26582 PR tree-optimization/80457
26583 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
26584 of all arguments to a statement as scalar_to_vec operations.
26585 (vectorizable_call): Adjust call to vect_model_simple_cost for
26587 (vectorizable_conversion): Likewise.
26588 (vectorizable_assignment): Likewise.
26589 (vectorizable_shift): Likewise.
26590 (vectorizable_operation): Likewise.
26591 (vectorizable_comparison): Likewise.
26592 (vect_is_simple_cond): Record the def types for operands.
26593 (vectorizable_condition): Likewise, call vect_model_simple_cost.
26594 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
26595 for statement argument count.
26597 2017-05-16 Carl Love <cel@us.ibm.com>
26599 * config/rs6000/rs6000-c: Add support for built-in functions
26600 vector unsigned long long vec_bperm (vector unsigned long long,
26601 vector unsigned char)
26602 vector signed long long vec_mule (vector signed int,
26604 vector unsigned long long vec_mule (vector unsigned int,
26605 vector unsigned int)
26606 vector signed long long vec_mulo (vector signed int,
26608 vector unsigned long long vec_mulo (vector unsigned int,
26609 vector unsigned int)
26610 vector signed char vec_sldw (vector signed char,
26611 vector signed char,
26613 vector unsigned char vec_sldw (vector unsigned char,
26614 vector unsigned char,
26616 vector signed short vec_sldw (vector signed short,
26617 vector signed short,
26619 vector unsigned short vec_sldw (vector unsigned short,
26620 vector unsigned short,
26622 vector signed int vec_sldw (vector signed int,
26625 vector unsigned int vec_sldw (vector unsigned int,
26626 vector unsigned int,
26628 vector signed long long vec_sldw (vector signed long long,
26629 vector signed long long,
26631 vector unsigned long long vec_sldw (vector unsigned long long,
26632 vector unsigned long long,
26634 * config/rs6000/rs6000-c: Add support for built-in functions
26635 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
26636 * config/rs6000/altivec.h: Add defintion for vec_sldw.
26637 * doc/extend.texi: Update the built-in documentation for the
26638 new built-in functions.
26640 2017-05-16 Marek Polacek <polacek@redhat.com>
26644 * tree.c (save_expr): Don't fold the expression.
26646 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
26648 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
26649 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
26650 and (?*y,m). Update insn attributes.
26652 2017-05-16 Martin Liska <mliska@suse.cz>
26654 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
26655 flags argument of print_gimple_stmt, print_gimple_expr,
26656 print_generic_stmt and print_generic_expr.
26657 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
26658 * coretypes.h: Likewise.
26659 * except.c (dump_eh_tree): Likewise.
26660 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
26661 * gimple-pretty-print.h: Likewise.
26662 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
26663 (backprop::push_to_worklist): Likewise.
26664 (backprop::pop_from_worklist): Likewise.
26665 (backprop::process_use): Likewise.
26666 (backprop::intersect_uses): Likewise.
26667 (note_replacement): Likewise.
26668 * gimple-ssa-store-merging.c
26669 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
26670 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
26671 (pass_store_merging::execute): Likewise.
26672 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
26673 (ssa_base_cand_dump_callback): Likewise.
26674 (dump_incr_vec): Likewise.
26675 (replace_refs): Likewise.
26676 (replace_mult_candidate): Likewise.
26677 (create_add_on_incoming_edge): Likewise.
26678 (create_phi_basis): Likewise.
26679 (insert_initializers): Likewise.
26680 (all_phi_incrs_profitable): Likewise.
26681 (introduce_cast_before_cand): Likewise.
26682 (replace_one_candidate): Likewise.
26683 * gimplify.c (gimplify_expr): Likewise.
26684 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
26685 (set_rename): Likewise.
26686 (rename_uses): Likewise.
26687 (copy_loop_phi_nodes): Likewise.
26688 (add_close_phis_to_merge_points): Likewise.
26689 (copy_loop_close_phi_args): Likewise.
26690 (copy_cond_phi_args): Likewise.
26691 (graphite_copy_stmts_from_block): Likewise.
26692 (translate_pending_phi_nodes): Likewise.
26693 * graphite-poly.c (print_pdr): Likewise.
26694 (dump_gbb_cases): Likewise.
26695 (dump_gbb_conditions): Likewise.
26696 (print_scop_params): Likewise.
26697 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
26698 (build_cross_bb_scalars_use): Likewise.
26699 (gather_bbs::before_dom_children): Likewise.
26700 * hsa-dump.c (dump_hsa_immed): Likewise.
26701 * ipa-cp.c (print_ipcp_constant_value): Likewise.
26702 (get_replacement_map): Likewise.
26703 * ipa-inline-analysis.c (dump_condition): Likewise.
26704 (estimate_function_body_sizes): Likewise.
26705 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
26706 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
26707 * ipa-prop.c (ipa_dump_param): Likewise.
26708 (ipa_print_node_jump_functions_for_edge): Likewise.
26709 (ipa_modify_call_arguments): Likewise.
26710 (ipa_modify_expr): Likewise.
26711 (ipa_dump_param_adjustments): Likewise.
26712 (ipa_dump_agg_replacement_values): Likewise.
26713 (ipcp_modif_dom_walker::before_dom_children): Likewise.
26714 * ipa-pure-const.c (check_stmt): Likewise.
26715 (pass_nothrow::execute): Likewise.
26716 * ipa-split.c (execute_split_functions): Likewise.
26717 * omp-offload.c (dump_oacc_loop_part): Likewise.
26718 (dump_oacc_loop): Likewise.
26719 * trans-mem.c (tm_log_emit): Likewise.
26720 (tm_memopt_accumulate_memops): Likewise.
26721 (dump_tm_memopt_set): Likewise.
26722 (dump_tm_memopt_transform): Likewise.
26723 * tree-cfg.c (gimple_verify_flow_info): Likewise.
26724 (print_loop): Likewise.
26725 * tree-chkp-opt.c (chkp_print_addr): Likewise.
26726 (chkp_gather_checks_info): Likewise.
26727 (chkp_get_check_result): Likewise.
26728 (chkp_remove_check_if_pass): Likewise.
26729 (chkp_use_outer_bounds_if_possible): Likewise.
26730 (chkp_reduce_bounds_lifetime): Likewise.
26731 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
26732 (chkp_mark_completed_bounds): Likewise.
26733 (chkp_register_incomplete_bounds): Likewise.
26734 (chkp_mark_invalid_bounds): Likewise.
26735 (chkp_maybe_copy_and_register_bounds): Likewise.
26736 (chkp_build_returned_bound): Likewise.
26737 (chkp_get_bound_for_parm): Likewise.
26738 (chkp_build_bndldx): Likewise.
26739 (chkp_get_bounds_by_definition): Likewise.
26740 (chkp_generate_extern_var_bounds): Likewise.
26741 (chkp_get_bounds_for_decl_addr): Likewise.
26742 * tree-chrec.c (chrec_apply): Likewise.
26743 * tree-data-ref.c (dump_data_reference): Likewise.
26744 (dump_subscript): Likewise.
26745 (dump_data_dependence_relation): Likewise.
26746 (analyze_overlapping_iterations): Likewise.
26747 * tree-inline.c (expand_call_inline): Likewise.
26748 (tree_function_versioning): Likewise.
26749 * tree-into-ssa.c (dump_defs_stack): Likewise.
26750 (dump_currdefs): Likewise.
26751 (dump_names_replaced_by): Likewise.
26752 (dump_update_ssa): Likewise.
26753 (update_ssa): Likewise.
26754 * tree-object-size.c (pass_object_sizes::execute): Likewise.
26755 * tree-parloops.c (build_new_reduction): Likewise.
26756 (try_create_reduction_list): Likewise.
26757 (ref_conflicts_with_region): Likewise.
26758 (oacc_entry_exit_ok_1): Likewise.
26759 (oacc_entry_exit_single_gang): Likewise.
26760 * tree-pretty-print.h: Likewise.
26761 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
26762 (get_scalar_evolution): Likewise.
26763 (add_to_evolution): Likewise.
26764 (get_loop_exit_condition): Likewise.
26765 (analyze_evolution_in_loop): Likewise.
26766 (analyze_initial_condition): Likewise.
26767 (analyze_scalar_evolution): Likewise.
26768 (instantiate_scev): Likewise.
26769 (number_of_latch_executions): Likewise.
26770 (gather_chrec_stats): Likewise.
26771 (final_value_replacement_loop): Likewise.
26772 (scev_const_prop): Likewise.
26773 * tree-sra.c (dump_access): Likewise.
26774 (disqualify_candidate): Likewise.
26775 (create_access): Likewise.
26776 (reject): Likewise.
26777 (maybe_add_sra_candidate): Likewise.
26778 (create_access_replacement): Likewise.
26779 (analyze_access_subtree): Likewise.
26780 (analyze_all_variable_accesses): Likewise.
26781 (sra_modify_assign): Likewise.
26782 (initialize_constant_pool_replacements): Likewise.
26783 (find_param_candidates): Likewise.
26784 (decide_one_param_reduction): Likewise.
26785 (replace_removed_params_ssa_names): Likewise.
26786 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
26787 * tree-ssa-copy.c (dump_copy_of): Likewise.
26788 (copy_prop_visit_cond_stmt): Likewise.
26789 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
26790 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
26791 (record_equivalences_from_stmt): Likewise.
26792 * tree-ssa-dse.c (compute_trims): Likewise.
26793 (delete_dead_call): Likewise.
26794 (delete_dead_assignment): Likewise.
26795 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
26796 (forward_propagate_into_cond): Likewise.
26797 (pass_forwprop::execute): Likewise.
26798 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
26799 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
26801 (move_computations_worker): Likewise.
26802 (execute_sm): Likewise.
26803 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
26804 (remove_exits_and_undefined_stmts): Likewise.
26805 (remove_redundant_iv_tests): Likewise.
26806 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
26807 (adjust_iv_update_pos): Likewise.
26808 * tree-ssa-math-opts.c (bswap_replace): Likewise.
26809 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
26810 (value_replacement): Likewise.
26811 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
26812 * tree-ssa-pre.c (print_pre_expr): Likewise.
26813 (get_representative_for): Likewise.
26814 (create_expression_by_pieces): Likewise.
26815 (insert_into_preds_of_block): Likewise.
26816 (eliminate_insert): Likewise.
26817 (eliminate_dom_walker::before_dom_children): Likewise.
26818 (eliminate): Likewise.
26819 (remove_dead_inserted_code): Likewise.
26820 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
26821 * tree-ssa-reassoc.c (get_rank): Likewise.
26822 (eliminate_duplicate_pair): Likewise.
26823 (eliminate_plus_minus_pair): Likewise.
26824 (eliminate_not_pairs): Likewise.
26825 (undistribute_ops_list): Likewise.
26826 (eliminate_redundant_comparison): Likewise.
26827 (update_range_test): Likewise.
26828 (optimize_range_tests_var_bound): Likewise.
26829 (optimize_vec_cond_expr): Likewise.
26830 (rewrite_expr_tree): Likewise.
26831 (rewrite_expr_tree_parallel): Likewise.
26832 (linearize_expr): Likewise.
26833 (break_up_subtract): Likewise.
26834 (linearize_expr_tree): Likewise.
26835 (attempt_builtin_powi): Likewise.
26836 (attempt_builtin_copysign): Likewise.
26837 (transform_stmt_to_copy): Likewise.
26838 (transform_stmt_to_multiply): Likewise.
26839 (dump_ops_vector): Likewise.
26840 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
26841 (print_scc): Likewise.
26842 (set_ssa_val_to): Likewise.
26843 (visit_reference_op_store): Likewise.
26844 (visit_use): Likewise.
26845 (sccvn_dom_walker::before_dom_children): Likewise.
26846 (run_scc_vn): Likewise.
26847 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
26849 (expr_hash_elt::print): Likewise.
26850 (const_and_copies::pop_to_marker): Likewise.
26851 (const_and_copies::record_const_or_copy_raw): Likewise.
26852 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
26853 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
26854 (dump_predicates): Likewise.
26855 (find_uninit_use): Likewise.
26856 (warn_uninitialized_phi): Likewise.
26857 (pass_late_warn_uninitialized::execute): Likewise.
26858 * tree-ssa.c (verify_vssa): Likewise.
26859 (verify_ssa): Likewise.
26860 (maybe_optimize_var): Likewise.
26861 * tree-vrp.c (dump_value_range): Likewise.
26862 (dump_all_value_ranges): Likewise.
26863 (dump_asserts_for): Likewise.
26864 (register_edge_assert_for_2): Likewise.
26865 (vrp_visit_cond_stmt): Likewise.
26866 (vrp_visit_switch_stmt): Likewise.
26867 (vrp_visit_stmt): Likewise.
26868 (vrp_visit_phi_node): Likewise.
26869 (simplify_cond_using_ranges_1): Likewise.
26870 (fold_predicate_in): Likewise.
26871 (evrp_dom_walker::before_dom_children): Likewise.
26872 (evrp_dom_walker::push_value_range): Likewise.
26873 (evrp_dom_walker::pop_value_range): Likewise.
26874 (execute_early_vrp): Likewise.
26876 2017-05-16 Richard Biener <rguenther@suse.de>
26878 * dwarf2out.c (loc_list_from_tree_1): Do not create
26879 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
26881 2017-05-16 Richard Biener <rguenther@suse.de>
26883 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
26885 (note_variable_value_in_expr): If we resolved the decl ref
26886 do not push to the stack.
26888 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
26890 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
26891 operations in fast-math mode.
26892 (vaddq_f16): Likewise.
26893 (vmul_f16): Likewise.
26894 (vmulq_f16): Likewise.
26895 (vsub_f16): Likewise.
26896 (vsubq_f16): Likewise.
26897 * config/arm/neon.md (add<mode>3): New.
26899 (fma:<VH:mode>3): New. Also remove outdated comment.
26902 2017-05-16 Martin Liska <mliska@suse.cz>
26906 * ipa-devirt.c (warn_types_mismatch): Fix typo.
26907 (odr_types_equivalent_p): Likewise.
26909 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
26911 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
26913 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
26916 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
26917 non-interunit SSE move alternatives with '?'.
26918 (zero-extendsidi peephole2): New peephole to skip intermediate
26919 general register in SSE zero-extend sequence.
26921 2017-05-15 Jeff Law <law@redhat.com>
26923 * reorg.c (relax_delay_slots): Create a new variable to hold
26924 the temporary target rather than clobbering TARGET_LABEL.
26926 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
26927 missing argument to extract_bit_field call.
26928 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
26930 2017-05-15 Martin Liska <mliska@suse.cz>
26933 * gcc.c (process_command): Do not allow empty argument of -o option.
26935 2017-05-15 Renlin Li <renlin.li@arm.com>
26937 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
26938 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
26939 * config/aarch64/constraints.md (Usf): Add long call check.
26940 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
26941 (call_value): Likewise.
26942 (sibcall): Likewise.
26943 (sibcall_value): Likewise.
26945 (call_value_insn): New.
26946 (sibcall_insn): Update rtx pattern.
26947 (sibcall_value_insn): Likewise.
26948 (call_internal): Remove.
26949 (call_value_internal): Likewise.
26950 (sibcall_internal): Likewise.
26951 (sibcall_value_internal): Likewise.
26952 (call_reg): Likewise.
26953 (call_symbol): Likewise.
26954 (call_value_reg): Likewise.
26955 (call_value_symbol): Likewise.
26957 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
26960 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
26962 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
26964 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
26965 compatible with CCGOCmode and with CCZmode.
26967 2017-05-14 Martin Sebor <msebor@redhat.com>
26969 PR middle-end/77671
26970 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
26971 (gimple_fold_builtin_snprintf): Same.
26972 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
26973 (gimple_fold_builtin_snprintf): Same.
26974 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
26975 of character types.
26976 (is_call_safe): New function.
26977 (try_substitute_return_value): Call it.
26978 (try_simplify_call): New function.
26979 (pass_sprintf_length::handle_gimple_call): Call it.
26981 2017-05-14 Martin Sebor <msebor@redhat.com>
26983 PR middle-end/80669
26984 * builtins.c (expand_builtin_stpncpy): Simplify.
26986 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
26988 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
26989 * config/i386/i386.h
26990 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
26991 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
26992 (struct machine_function): Add new members call_ms2sysv,
26993 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
26994 (struct machine_frame_state): New fields sp_realigned and
26995 sp_realigned_offset.
26996 * config/i386/i386.c
26997 (enum xlogue_stub): New enum.
26998 (enum xlogue_stub_sets): New enum.
26999 (class xlogue_layout): New class.
27000 (struct ix86_frame): New fields stack_realign_allocate_offset,
27001 stack_realign_offset and outlined_save_offset. Modify comments to
27002 detail stack layout when using out-of-line stubs.
27003 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27004 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27005 -mcall-ms2sysv-xlogues.
27006 (stub_managed_regs): New static variable.
27007 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27008 registers managed by out-of-line stub.
27009 (disable_call_ms2sysv_xlogues): New function.
27010 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27011 m->call_ms2sysv when appropriate and compute frame layout for
27013 (sp_valid_at, fp_valid_at): New inline functions.
27014 (choose_basereg): New function.
27015 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27017 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27018 Use align parameter of choose_baseaddr to generated aligned SSE movs
27020 (pro_epilogue_adjust_stack): Modify to track
27021 machine_frame_state::sp_realigned.
27022 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
27023 (ix86_nsaved_sseregs): Likewise.
27024 (ix86_emit_save_regs): Likewise.
27025 (ix86_emit_save_regs_using_mov): Likewise.
27026 (ix86_emit_save_sse_regs_using_mov): Likewise.
27027 (get_scratch_register_on_entry): Likewise.
27028 (gen_frame_set): New function.
27029 (gen_frame_load): Likewise.
27030 (gen_frame_store): Likewise.
27031 (emit_outlined_ms2sysv_save): Likewise.
27032 (emit_outlined_ms2sysv_restore): Likewise.
27033 (ix86_expand_prologue): Modify stack re-alignment code and call
27034 emit_outlined_ms2sysv_save when appropriate.
27035 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
27036 parameter rtx_insn *insn, which allows the function to be used to only
27037 generate the notes.
27038 (ix86_expand_epilogue): Modify validity checks of frame and stack
27039 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
27040 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
27041 * config/i386/predicates.md
27042 (save_multiple): New predicate.
27043 (restore_multiple): Likewise.
27044 * config/i386/sse.md
27045 (save_multiple<mode>): New pattern.
27046 (save_multiple_realign<mode>): Likewise.
27047 (restore_multiple<mode>): Likewise.
27048 (restore_multiple_and_return<mode>): Likewise.
27049 (restore_multiple_leave_return<mode>): Likewise.
27050 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
27052 2017-05-14 Julia Koval <julia.koval@intel.com>
27054 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
27055 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
27056 (__builtin_ia32_xsetbv): New builtins.
27057 * config/i386/i386.c (ix86_expand_special_args_builtin):
27059 (ix86_expand_builtin): Special expand for new intrinsics.
27060 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
27061 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
27062 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
27064 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27066 * cfganal.c (inverted_post_order_compute): Change argument type
27068 * cfganal.h (inverted_post_order_compute): Adjust prototype.
27069 * df-core.c (rest_of_handle_df_initialize): Adjust.
27070 (rest_of_handle_df_finish): Likewise.
27071 (df_analyze_1): Likewise.
27072 (df_analyze): Likewise.
27073 (loop_inverted_post_order_compute): Change argument to be a vec *.
27074 (df_analyze_loop): Adjust.
27075 (df_get_n_blocks): Likewise.
27076 (df_get_postorder): Likewise.
27077 * df.h (struct df_d): Change field to be a vec.
27078 * lcm.c (compute_laterin): Adjust.
27079 (compute_available): Likewise.
27080 * lra-lives.c (lra_create_live_ranges_1): Likewise.
27081 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
27082 * tree-ssa-pre.c (compute_antic): Likewise.
27084 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27086 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
27087 (depth_first_search::depth_first_search): Change structure init
27088 function to this constructor.
27089 (depth_first_search::add_bb): Rename function to this member.
27090 (depth_first_search::execute): Likewise.
27091 (flow_dfs_compute_reverse_finish): Adjust.
27093 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27095 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
27096 (longest_simple_path): Likewise.
27097 * shrink-wrap.c (spread_components): Likewise.
27098 (disqualify_problematic_components): Likewise.
27099 (emit_common_heads_for_components): Likewise.
27100 (emit_common_tails_for_components): Likewise.
27101 (insert_prologue_epilogue_for_components): Likewise.
27103 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27105 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
27108 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27110 * df-core.c (df_set_blocks): Start using auto_bitmap.
27111 (df_compact_blocks): Likewise.
27112 * df-problems.c (df_rd_confluence_n): Likewise.
27113 * df-scan.c (df_insn_rescan_all): Likewise.
27114 (df_process_deferred_rescans): Likewise.
27115 (df_update_entry_block_defs): Likewise.
27116 (df_update_exit_block_uses): Likewise.
27117 (df_entry_block_bitmap_verify): Likewise.
27118 (df_exit_block_bitmap_verify): Likewise.
27119 (df_scan_verify): Likewise.
27120 * lra-constraints.c (lra_constraints): Likewise.
27121 (undo_optional_reloads): Likewise.
27122 (lra_undo_inheritance): Likewise.
27123 * lra-remat.c (calculate_gen_cands): Likewise.
27124 (do_remat): Likewise.
27125 * lra-spills.c (assign_spill_hard_regs): Likewise.
27126 (spill_pseudos): Likewise.
27127 * tree-ssa-pre.c (bitmap_set_and): Likewise.
27128 (bitmap_set_subtract_values): Likewise.
27130 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27132 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
27133 management with auto_bitmap.
27134 (fix_inter_tick): Likewise.
27135 (fix_recovery_deps): Likewise.
27136 * ira.c (add_store_equivs): Likewise.
27137 (find_moveable_pseudos): Likewise.
27138 (split_live_ranges_for_shrink_wrap): Likewise.
27139 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
27140 (rtx_reuse_manager::seen_def_p): Likewise.
27141 (rtx_reuse_manager::set_seen_def): Likewise.
27142 * print-rtl.h (class rtx_reuse_manager): Likewise.
27144 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27146 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
27148 (migrate_btr_def): Likewise.
27149 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
27150 * df-core.c (loop_post_order_compute): Likewise.
27151 (loop_inverted_post_order_compute): Likewise.
27152 * hsa-common.h: Likewise.
27153 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
27154 * init-regs.c (initialize_uninitialized_regs): Likewise.
27155 * ipa-inline.c (resolve_noninline_speculation): Likewise.
27156 (inline_small_functions): Likewise.
27157 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
27158 * ira.c (combine_and_move_insns): Likewise.
27159 (build_insn_chain): Likewise.
27160 * loop-invariant.c (find_invariants): Likewise.
27161 * lower-subreg.c (propagate_pseudo_copies): Likewise.
27162 * predict.c (tree_predict_by_opcode): Likewise.
27163 (predict_paths_leading_to): Likewise.
27164 (predict_paths_leading_to_edge): Likewise.
27165 (estimate_loops_at_level): Likewise.
27166 (estimate_loops): Likewise.
27167 * shrink-wrap.c (try_shrink_wrapping): Likewise.
27168 (spread_components): Likewise.
27169 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
27170 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
27171 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
27172 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
27173 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
27174 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
27175 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
27176 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
27177 (mark_threaded_blocks): Likewise.
27178 (thread_through_all_blocks): Likewise.
27179 * tree-ssa.c (verify_ssa): Likewise.
27180 (execute_update_addresses_taken): Likewise.
27181 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
27183 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27185 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
27187 (post_order_compute): Likewise.
27188 (inverted_post_order_compute): Likewise.
27189 (pre_and_rev_post_order_compute_fn): Likewise.
27191 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27193 * genrecog.c (int_set::int_set): Explicitly construct our
27194 auto_vec base class.
27195 * vec.h (auto_vec::auto_vec): New constructor.
27197 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27199 * bitmap.h (class auto_bitmap): New constructor taking
27200 bitmap_obstack * argument.
27202 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27204 * bitmap.h (class auto_bitmap): Change type of m_bits to
27205 bitmap_head, and adjust ctor / dtor and member operators.
27207 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
27209 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
27210 when returned register mode doesn't match original mode.
27212 2017-05-12 Jeff Law <law@redhat.com>
27213 Jakub Jelinek <jakub@redhat.com>
27215 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
27216 we look for cc setter after the compare-elim changes.
27217 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
27218 within the vector to match what compare-elim now expects.
27219 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
27220 (xorsi3_flags, one_cmplsi2_flags): Likewise.
27222 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
27223 after the compare-elim changes.
27224 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
27225 the vector to match what compare-elim now expects.
27226 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
27227 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
27228 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
27229 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
27230 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
27232 * config/visium/visium.c (single_set_and_flags): Fix where
27233 we look for cc setter after the compare-elim changes.
27234 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
27235 with the vector to match what compare-elim now expects.
27236 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
27237 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
27238 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
27239 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
27240 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
27241 (neg<mode>2_insn_set_overflow): Likewise.
27243 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
27245 PR middle-end/79794
27246 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
27247 maybe_expand_insn call, set ops[0].target. If still set after call,
27248 set alt_rtl. Add extra arg to recursive calls.
27249 (extract_bit_field): Add alt_rtl argument. Pass to
27251 * expmed.h (extract_bit_field): Fix prototype.
27252 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
27253 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
27254 to extract_bit_field_calls.
27255 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
27256 Pass alt_rtl to extract_bit_field calls.
27257 * calls.c (store_unaligned_arguments_into_psuedos)
27258 load_register_parameters): Pass extra NULL to extract_bit_field calls.
27259 * optabs.c (maybe_legitimize_operand): Clear op->target when call
27261 * optabs.h (struct expand_operand): Add target bitfield.
27263 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27265 * compare-elim.c (try_eliminate_compare): Canonicalize
27266 operation with embedded compare to
27267 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
27268 (set (reg) (operation)].
27270 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
27272 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27275 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
27276 cost of adding a carry flag for ADC instruction.
27277 [case MINUS]: Ignore the cost of subtracting a carry flag
27278 for SBB instruction.
27280 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
27282 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
27284 * config/rs6000/bmiintrin.h: New file.
27285 * config/rs6000/bmi2intrin.h: New file.
27286 * config/rs6000/x86intrin.h: New file.
27288 2017-05-12 Jeff Law <law@redhat.com>
27290 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
27293 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
27295 PR middle-end/80707
27296 * tree-cfg.c: Remove cfg edges of unreachable case statements.
27298 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27300 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27301 early expansion of vector divide builtins.
27302 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
27303 builtins identified as having unsigned arguments.
27305 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27307 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
27308 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
27309 expansion of vector logical operations (and, andc, or, xor,
27312 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27314 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
27315 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
27317 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27319 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27320 early GIMPLE expansion of vector multiplies.
27322 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27324 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
27325 TARGET_HAVE_MOVT conditional.
27326 (movt splitter): Likewise.
27328 2017-05-12 Richard Biener <rguenther@suse.de>
27330 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
27331 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27332 Fold all stmts not inplace.
27334 2017-05-12 Richard Biener <rguenther@suse.de>
27336 PR tree-optimization/80713
27337 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
27338 inserted_exprs bit for not removed stmts.
27340 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
27342 PR middle-end/69921
27343 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
27344 parallelized" attribute for parallelized OpenACC kernels.
27345 * omp-offload.c (execute_oacc_device_lower): Use it.
27347 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
27348 Set "oacc kernels" attribute.
27349 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
27350 parameter. Adjust all users.
27351 (oacc_fn_attrib_kernels_p): Remove function.
27352 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
27353 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
27354 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
27355 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
27356 assert "oacc kernels" attribute is set.
27358 2017-05-11 Carl Love <cel@us.ibm.com>
27360 * config/rs6000/rs6000-c: Add support for built-in functions
27361 vector unsigned char vec_popcnt (vector signed char)
27362 vector unsigned char vec_popcnt (vector unsigned char)
27363 vector unsigned short vec_popcnt (vector signed short)
27364 vector unsigned short vec_popcnt (vector unsigned short)
27365 vector unsigned int vec_popcnt (vector signed int)
27366 vector unsigned int vec_popcnt (vector unsigned int)
27367 vector unsigned long long vec_popcnt (vector signed long long)
27368 vector unsigned long long vec_popcnt (vector unsigned long long)
27369 vector signed long long vec_slo (vector signed long long,
27370 vector signed char)
27371 vector signed long long vec_slo (vector signed long long,
27372 vector unsigned char)
27373 vector unsigned long long vec_slo (vector unsigned long long,
27374 vector signed char)
27375 vector unsigned long long vec_slo (vector unsigned long long,
27376 vector unsigned char)
27377 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
27378 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
27379 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
27380 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
27381 * doc/extend.texi: Update the built-in documentation file for the
27382 new built-in functions.
27384 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27386 * attribs.h (sorted_attr_string): Move machine independent
27387 functions for target clone support from the i386 port to common
27388 code. Rename ix86_function_versions to common_function_versions.
27389 Rename make_name to make_unique_name.
27390 (common_function_versions): Likewise.
27391 (make_unique_name): Likewise.
27392 (make_dispatcher_decl): Likewise.
27393 (is_function_default_version): Likewise.
27394 * attribs.c (attr_strcmp): Likewise.
27395 (sorted_attr_string): Likewise.
27396 (common_function_versions): Likewise.
27397 (make_unique_name): Likewise.
27398 (make_dispatcher_decl): Likewise.
27399 (is_function_default_version): Likewise.
27400 * config/i386/i386.c (attr_strcmp): Likewise.
27401 (sorted_attr_string): Likewise.
27402 (ix86_function_versions): Likewise.
27403 (make_name): Likewise.
27404 (make_dispatcher_decl): Likewise.
27405 (is_function_default_version): Likewise.
27406 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
27408 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27411 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
27412 Account for direct move costs for vec_construct of integer
27415 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
27418 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
27419 (UNSPEC_STX_ATOMIC): Ditto.
27420 (loaddi_via_sse): New insn.
27421 (storedi_via_sse): Ditto.
27422 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
27423 Update corresponding peephole2 patterns.
27424 (atomic_storedi_fpu): Ditto.
27426 2017-05-11 Julia Koval <julia.koval@intel.com>
27428 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
27429 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
27431 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
27432 (__builtin_ia32_rsqrt14ss_mask): New builtins.
27433 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
27435 2017-05-11 Nathan Sidwell <nathan@acm.org>
27437 * graphite-poly.c: Include dumpfile.h.
27439 * dumpfle.h (dump_function): Declare here ...
27440 * tree-dump.h (dump_function): ... not here.
27441 * dumpfile.c: #include tree-cfg.h.
27442 (dump_function): Move here from ...
27443 * tree-dump.c (dump_function): ... here.
27444 * gimplify.c: #include splay-tree.h, not tree-dump.h.
27445 * graphite-poly.c: Don't include tree-dump.h.
27446 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
27447 * print-tree.c: Likewise.
27448 * stor-layout.c: Likewise.
27449 * tree-nested.c: Likewise.
27451 * dumpfile.c (dump_start): Use TDF_FLAGS.
27452 (dump_enable_all): Fix TDF_KIND check thinko.
27454 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
27456 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27457 array entries to represent two legal parameterizations of the
27458 overloaded __builtin_cmpb function, as represented by the
27459 P6_OV_BUILTIN_CMPB constant.
27460 (altivec_resolve_overloaded_builtin): Add special case handling
27461 for the __builtin_cmpb function, as represented by the
27462 P6_OV_BUILTIN_CMPB constant.
27463 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
27464 (BU_P6_64BIT_2): New macro.
27465 (BU_P6_OVERLOAD_2): New macro
27466 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
27467 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
27468 (CMPB): Add overload support to represent both 32-bit and 64-bit
27469 compare-bytes function.
27470 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
27471 support for TARGET_CMPB.
27472 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
27473 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27474 documentation of the __builtin_cmpb overloaded built-in function.
27476 2017-05-11 Richard Biener <rguenther@suse.de>
27478 PR tree-optimization/80705
27479 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
27480 bases are not vectorizable.
27482 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27484 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
27485 when counting register pressure.
27487 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27489 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
27490 (iv_ca_more_deps): Renamed to ...
27491 (iv_ca_compare_deps): ... this.
27492 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
27494 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27496 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
27498 (determine_group_iv_costs): ... here.
27499 (find_inv_vars_cb): Record inv var if it's not recorded before.
27501 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27503 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
27504 (get_shiftadd_cost): Ditto.
27506 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27508 * tree-ssa-address.c: Include header file.
27509 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
27511 (add_to_parts): Refactor.
27512 (addr_to_parts): New parameter. Update use of move_hint_to_base.
27513 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
27516 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27518 PR tree-optimization/53090
27519 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
27521 (extract_cond_operands): Detect condition with IV on both sides
27522 and return COMP_IV_EXPR_2.
27523 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
27524 (rewrite_use_compare): Simplify by removing call to function
27525 extract_cond_operands.
27527 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27529 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
27530 (extract_cond_operands): Detect condition comparing against non-
27531 invariant bound and return appropriate enum value.
27532 (find_interesting_uses_cond): Update use of extract_cond_operands.
27533 Handle its return value accordingly.
27534 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
27536 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27538 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
27539 nonlinear iv_use computation in loop invariant sensitive way.
27541 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27543 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
27544 (find_iv_candidates): Call relate_compare_use_with_all_cands.
27546 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27548 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
27549 (dump_cand): Support iv_cand.inv_exprs.
27550 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
27552 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
27555 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27557 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
27559 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
27560 as local function. Include necessary header files.
27561 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
27563 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27565 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
27567 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27569 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
27570 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
27571 RSHIFT_EXPR and BIT_NOT_EXPR.
27573 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27575 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
27576 (adjust_setup_cost): New parameter supporting round up adjustment.
27577 (struct address_cost_data): Delete.
27578 (force_expr_to_var_cost): Don't bound cost with spill_cost.
27579 (split_address_cost, ptr_difference_cost): Delete.
27580 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
27581 (struct ainc_cost_data): New struct.
27582 (get_address_cost_ainc): New function.
27583 (get_address_cost, get_computation_cost): Reimplement.
27584 (determine_group_iv_cost_address): Record inv_expr for all uses of
27586 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
27587 (iv_ca_has_deps): Reimplemented to ...
27588 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
27590 (iv_ca_extend): Call iv_ca_more_deps.
27592 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27594 * tree-ssa-address.c (struct mem_address): Move to header file.
27595 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
27596 * tree-ssa-address.h (struct mem_address): Move from C file.
27597 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
27599 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27601 * tree-affine.h (aff_combination_type): New interface.
27602 (aff_combination_zero_p): Remove static.
27603 (aff_combination_const_p): New interface.
27604 (aff_combination_singleton_var_p): New interfaces.
27606 2017-05-11 Richard Biener <rguenther@suse.de>
27608 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27609 Skip unreachable blocks and destinations.
27610 (eliminate): Move stmt removal and fixup ...
27611 (fini_eliminate): ... here. Skip inserted exprs.
27612 (pass_pre::execute): Move fini_pre after fini_eliminate.
27613 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
27614 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
27615 PRE to get rid of dead code that has invalid SSA form and
27616 split critical edges again.
27618 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27620 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27622 2017-05-11 Richard Biener <rguenther@suse.de>
27624 * passes.c (execute_function_todo): Verify loops if they are
27625 said to be up-to-date.
27626 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
27627 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
27629 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
27632 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
27633 handle calling assemble_external ourself.
27636 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
27637 modes with zero size. Enhance comment.
27639 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27641 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
27642 built-ins for vec_xl and vec_xst with short and char pointer
27645 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27647 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
27648 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
27649 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
27650 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
27651 (_mm_maskz_min_round_ss): New intrinsics.
27652 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
27653 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
27654 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
27655 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
27656 (__builtin_ia32_minss_mask_round): New builtins.
27657 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27658 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27659 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
27661 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
27662 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
27664 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
27667 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27669 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
27670 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
27671 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
27672 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
27673 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
27674 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
27675 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
27676 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27677 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27678 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
27679 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
27680 (__builtin_ia32_mulss_mask_round): New builtins.
27681 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27682 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27683 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
27685 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
27686 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27688 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27691 2017-05-10 Julia Koval <julia.koval@intel.com>
27693 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
27694 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
27695 (_mm256_setr_m128i): New intrinsics.
27697 2017-05-10 Julia Koval <julia.koval@intel.com>
27699 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
27700 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
27701 (_mm_maskz_rcp14_ss): New intrinsics.
27702 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
27703 (__builtin_ia32_rcp14ss_mask): New builtins.
27704 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
27706 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
27708 PR tree-optimization/51513
27709 * tree-cfg.c (gimple_seq_unreachable_p): New function.
27710 (assert_unreachable_fallthru_edge_p): Use it.
27711 (group_case_labels_stmt): Likewise.
27712 * tree-cfg.h: Prototype it.
27713 * stmt.c: Include cfghooks.h and tree-cfg.h.
27714 (emit_case_dispatch_table) <gap_label>: New local variable.
27715 Use it to fill dispatch table gaps.
27716 Test for default_label before updating probabilities.
27717 (expand_case) <default_label>: Remove unneeded initialization.
27718 Test for unreachable default case statement and remove its edge.
27719 Set default_label accordingly.
27720 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
27722 2017-05-10 Carl Love <cel@us.ibm.com>
27724 * config/rs6000/rs6000-c: Add support for built-in functions
27725 vector signed char vec_neg (vector signed char)
27726 vector signed short int vec_neg (vector short int)
27727 vector signed int vec_neg (vector signed int)
27728 vector signed long long vec_neg (vector signed long long)
27729 vector float vec_neg (vector float)
27730 vector double vec_neg (vector double)
27731 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
27733 * config/rs6000/altivec.h: Add define for vec_neg
27734 * doc/extend.texi: Update the built-in documentation for the
27735 new built-in functions.
27737 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27739 PR tree-optimization/77644
27740 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
27742 2017-05-10 Nathan Sidwell <nathan@acm.org>
27744 * dumpfile.h (TDI_lang_all): New.
27745 (TDF_KIND): New. Renumber others
27746 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
27748 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
27750 (get_dump_file_name): Adjust suffix generation.
27751 (dump_enable_all): Use TDF_KIND.
27752 * doc/invoke.texi (-fdump-lang-all): Document.
27754 * dumpfile.h: Tabify.
27756 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
27759 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
27760 Move member access before delete.
27762 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
27764 * tree-inline.c (expand_call_inline): Split block at stmt
27767 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
27770 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
27771 are now unused after splitting mov{sf,sd}_hardfloat.
27772 (f32_lr2): Likewise.
27773 (f32_lm): Likewise.
27774 (f32_lm2): Likewise.
27775 (f32_li): Likewise.
27776 (f32_li2): Likewise.
27777 (f32_lv): Likewise.
27778 (f32_sr): Likewise.
27779 (f32_sr2): Likewise.
27780 (f32_sm): Likewise.
27781 (f32_sm2): Likewise.
27782 (f32_si): Likewise.
27783 (f32_si2): Likewise.
27784 (f32_sv): Likewise.
27785 (f32_dm): Likewise.
27786 (f32_vsx): Likewise.
27787 (f32_av): Likewise.
27788 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
27789 For movsf, order stores so the VSX stores occur before the GPR
27790 store which encourages the register allocator to use a traditional
27791 FPR instead of a GPR. For movsd, order the stores so that the GPR
27792 store comes before the VSX stores to allow the power6 to work.
27793 This is due to the power6 not having a 32-bit integer store
27794 instruction from a FPR.
27795 (movsf_hardfloat): Likewise.
27796 (movsd_hardfloat): Likewise.
27798 2017-05-09 Martin Sebor <msebor@redhat.com>
27800 PR translation/80280
27801 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
27804 PR translation/80280
27805 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
27806 data member added in r247778.
27807 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
27809 2017-05-09 Nathan Sidwell <nathan@acm.org>
27811 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
27813 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
27816 2017-05-09 Marek Polacek <polacek@redhat.com>
27818 * doc/invoke.texi: Fix typo.
27820 2017-05-09 Richard Biener <rguenther@suse.de>
27822 * tree-vrp.c (vrp_val_is_max): Adjust comment.
27823 (vrp_val_is_min): Likewise.
27824 (set_value_range_to_value): Likewise.
27825 (set_value_range_to_nonnegative): Likewise.
27826 (gimple_assign_nonzero_p): Likewise.
27827 (gimple_stmt_nonzero_p): Likewise.
27828 (vrp_int_const_binop): Likewise. Remove unreachable case.
27829 (adjust_range_with_scev): Adjust comments.
27830 (compare_range_with_value): Likewise.
27831 (extract_range_from_phi_node): Likewise.
27832 (test_for_singularity): Likewise.
27834 2017-05-09 Richard Biener <rguenther@suse.de>
27836 * tree-vrp.c (get_single_symbol): Add assert that we don't
27837 get overflowed constants as invariant part.
27838 (compare_values_warnv): Add comment before the TREE_NO_WARNING
27839 checks. Use wi::cmp instead of recursing for integer constants.
27840 (compare_values): Just ignore whether we assumed undefined
27841 overflow instead of failing the compare.
27842 (extract_range_for_var_from_comparison_expr): Add comment before the
27843 TREE_NO_WARNING sets.
27844 (test_for_singularity): Likewise.
27845 (extract_range_from_comparison): Do not disable optimization
27846 when we assumed undefined overflow.
27847 (extract_range_basic): Remove init of unused var.
27849 2017-05-09 Richard Biener <rguenther@suse.de>
27851 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
27852 (extract_range_from_multiplicative_op_1): Adjust.
27853 (extract_range_from_binary_expr_1): Use int_const_binop.
27855 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
27858 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
27859 rs6000_store_data_bypass_p in seven define_bypass directives and
27860 in several comments.
27861 * config/rs6000/rs6000-protos.h: Add prototype for
27862 rs6000_store_data_bypass_p function.
27863 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
27864 function implements slightly different (rs6000-specific) semantics
27865 than store_data_bypass_p, returning false rather than aborting
27866 with assertion error when arguments do not satisfy the
27867 requirements of store data bypass.
27868 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
27869 rs6000_store_data_bypass_p.
27871 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
27873 * config/xtensa/xtensa-protos.h
27874 (xtensa_initial_elimination_offset): New declaration.
27875 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27876 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
27877 macro definition, add case for FRAME_POINTER_REGNUM when
27878 FRAME_GROWS_DOWNWARD.
27879 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
27880 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
27881 xtensa_initial_elimination_offset.
27883 2017-05-08 Nathan Sidwell <nathan@acm.org>
27885 * doc/invoke.texi: Alphabetize -fdump options.
27887 2017-05-08 Martin Sebor <msebor@redhat.com>
27889 PR translation/80280
27890 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
27892 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
27894 * target.def (compute_frame_layout): New optional target hook.
27895 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
27896 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
27897 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
27899 * reload1.c (verify_initial_elim_offsets): Likewise.
27900 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
27901 (use_simple_return_p): Call arm_compute_frame_layout if needed.
27902 (arm_get_frame_offsets): Split up into this ...
27903 (arm_compute_frame_layout): ... and this function.
27905 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
27907 * config/aarch64/constraints.md (Usa): New constraint.
27908 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
27910 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
27912 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
27913 with_multilib_list after it has been checked.
27915 2017-05-08 Richard Biener <rguenther@suse.de>
27917 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
27918 (bitmap_set_subtract_values): Likewise.
27920 2017-05-08 Richard Biener <rguenther@suse.de>
27922 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
27923 (gimple_assign_nonzero): ... this and remove strict_overflow_p
27925 (gimple_stmt_nonzero_warnv_p): Rename to ...
27926 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
27928 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
27929 (extract_range_basic): Adjust, do not disable propagation on
27930 strict overflow sensitive simplification.
27931 (vrp_visit_cond_stmt): Likewise.
27933 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
27935 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
27936 body size unconditionally.
27938 2017-05-07 Jeff Law <law@redhat.com>
27941 2017-05-06 Jeff Law <law@redhat.com>
27942 PR tree-optimization/78496
27943 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27946 PR tree-optimization/78496
27947 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27948 (simplify_stmt_using_ranges): Call it.
27949 (vrp_dom_walker::before_dom_children): Extract equivalences
27950 from an ASSERT_EXPR with an equality comparison against a
27953 2017-05-06 Jeff Law <law@redhat.com>
27955 PR tree-optimization/78496
27956 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27959 PR tree-optimization/78496
27960 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27961 (simplify_stmt_using_ranges): Call it.
27962 (vrp_dom_walker::before_dom_children): Extract equivalences
27963 from an ASSERT_EXPR with an equality comparison against a
27966 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27968 * lra-constraints.c (lra_copy_reg_equiv): New function.
27969 (split_reg): Use it to copy equivalence information from the
27970 original register to the spill register.
27972 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27974 PR rtl-optimization/75964
27975 * simplify-rtx.c (simplify_const_relational_operation): Remove
27976 invalid handling of comparisons of integer ABS.
27978 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
27980 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
27981 initialize to zero.
27982 (init_regs): Remove declaration.
27983 (function_arg_advance_32): Initialize error_p as boolean variable.
27985 2017-05-05 Nathan Sidwell <nathan@acm.org>
27987 * store-motion.c (remove_reachable_equiv_notes): Reformat long
27988 lines. Use for (;;).
27990 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27992 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
27993 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
27994 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
27995 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
27996 VF=2 that require versioning.
27998 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28000 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28003 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28005 * diagnostic.h (diagnostic_override_option_index): Convert from
28006 macro to inline function.
28008 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28010 * diagnostic.c (last_module_changed_p): New function.
28011 (set_last_module): New function.
28012 (diagnostic_report_current_module): Convert macro usage to
28013 the above functions.
28014 * diagnostic.h (diagnostic_context::last_module): Strengthen
28015 from const line_map * to const line_map_ordinary *.
28016 (diagnostic_last_module_changed): Delete macro.
28017 (diagnostic_set_last_module): Delete macro.
28019 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28021 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
28022 with diagnostic_report_diagnostic.
28023 (diagnostic_n_impl_richloc): Likewise.
28024 * diagnostic.h (report_diagnostic): Delete macro.
28025 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
28026 with diagnostic_report_diagnostic.
28027 * substring-locations.c (format_warning_va): Likewise.
28029 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28031 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
28032 save/restor of format_spec. Move option-printing code to...
28033 (print_option_information): ...this new function, and
28034 reimplement by simply printing to the pretty_printer,
28035 rather than appending to the format string.
28037 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28039 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
28040 handling logic into...
28041 (update_effective_level_from_pragmas): ...this new function.
28043 2017-05-04 Andrew Waterman <andrew@sifive.com>
28045 * config/riscv/riscv.opt (mstrict-align): New option.
28046 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
28047 (SLOW_UNALIGNED_ACCESS): Define.
28048 (riscv_slow_unaligned_access): Declare.
28049 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
28051 (riscv_slow_unaligned_access): New variable.
28052 (rocket_tune_info): Set slow_unaligned_access to true.
28053 (optimize_size_tune_info): Set slow_unaligned_access to false.
28054 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
28055 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
28056 (riscv_option_override): Set riscv_slow_unaligned_access.
28057 * doc/invoke.texi: Add -mstrict-align to RISC-V.
28059 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
28061 * config/riscv/riscv.md: Unify indentation.
28063 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28068 * config/rs6000/rs6000.md (u code attribute): Add FIX and
28070 (extendsi<mode>2): Add support for doing sign extension via
28071 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
28072 don't have ISA 3.0 instructions.
28073 (extendsi<mode>2 splitter): Likewise.
28074 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
28075 generate the normal insns since SImode can now go in vector
28076 registers. Disallow the special UNSPECs needed for previous
28077 machines to hide SImode being used. Add new insns
28078 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
28079 (fix_trunc<mode>si2_stfiwx): Likewise.
28080 (fix_trunc<mode>si2_internal): Likewise.
28081 (fixuns_trunc<mode>si2): Likewise.
28082 (fixuns_trunc<mode>si2_stfiwx): Likewise.
28083 (fctiw<u>z_<mode>_smallint): Likewise.
28084 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
28085 of floating point to 32-bit integer from doing a direct move to
28086 the GPR registers to do a store.
28087 (fctiwz_<mode>): Break long line.
28089 2017-05-05 Bin Cheng <bin.cheng@arm.com>
28091 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
28092 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
28093 (addr_list, addr_offset_valid_p): New.
28094 (split_address_groups): Check offset validity with above function.
28095 (gt-tree-ssa-loop-ivopts.h): Include header file.
28097 2017-05-05 Nathan Sidwell <nathan@acm.org>
28099 * config.gcc (arm*-*-*): Add missing 'fi'.
28101 2017-05-05 Steve Ellcey <sellcey@cavium.com>
28103 * doc/invoke.texi (-fopt-info): Explicitly say order of options
28104 included in -fopt-info does not matter.
28105 * doc/optinfo.texi (-fopt-info): Fix description of default
28106 behavour. Explicitly say order of options included in -fopt-info
28109 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28111 * config.gcc: Allow combinations of aprofile and rmprofile values for
28112 --with-multilib-list.
28113 * config/arm/t-multilib: New file.
28114 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
28115 variables. Remove setting of ISA and floating-point ABI in
28116 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
28117 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
28118 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
28119 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
28121 * config/arm/t-rmprofile: Likewise except for the matches changes.
28122 * doc/install.texi (--with-multilib-list): Document the combination of
28123 aprofile and rmprofile values and warn about pitfalls in doing that.
28125 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28127 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
28128 (movdi_aarch64): Likewise.
28130 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28132 PR tree-optimization/80632
28133 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
28135 (build_arrays): Initialize it for virtual phis.
28136 (fix_phi_nodes): Use it for virtual phis.
28138 PR tree-optimization/80558
28139 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
28140 [x, y] op z into [x op, y op z] for op & or | if conditions
28143 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
28144 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28147 * config/arm/arm.md (use_literal_pool): Remove.
28148 (64-bit immediate split): No longer takes cost into consideration
28149 if arm_disable_literal_pool is enabled.
28150 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
28151 used when arm_disable_literal_pool is enabled.
28152 (arm_max_const_double_inline_cost): Remove use of
28153 arm_disable_literal_pool.
28154 (push_minipool_fix): Add assert.
28155 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
28156 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
28157 (no_literal_pool_sf_immediate): New.
28159 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28161 PR tree-optimization/80613
28162 * tree-ssa-dce.c (propagate_necessity): Remove cases for
28163 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
28165 2017-05-05 Richard Biener <rguenther@suse.de>
28167 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
28169 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
28171 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
28172 of this flag from insn conditions due to removal from r247495.
28174 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28176 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
28178 (arm_early_store_addr_dep_ptr): Likewise.
28179 * config/arm/aarch-common-protos.h
28180 (arm_early_load_addr_dep_ptr): Add prototype.
28181 (arm_early_store_addr_dep_ptr): Likewise.
28182 * config/arm/cortex-a53.md: Add new bypasses.
28184 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28186 * tree.c (next_type_uid): Change type to unsigned.
28187 (type_hash_canon): Decrement back next_type_uid if
28188 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
28189 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
28192 2017-05-04 Martin Sebor <msebor@redhat.com>
28194 * builtins.c: Fix a trivial typo in a comment.
28196 PR middle-end/79234
28197 * builtins.c (check_sizes): Adjust to handle reading past the end.
28198 Avoid printing excessive upper bound of ranges. Use %E to print
28199 tree nodes instead of converting them to %wu.
28200 (expand_builtin_memchr): New function.
28201 (compute_dest_size): Rename...
28202 (compute_objsize): ...to this.
28203 (expand_builtin_memcpy): Adjust.
28204 (expand_builtin_mempcpy): Adjust.
28205 (expand_builtin_strcat): Adjust.
28206 (expand_builtin_strcpy): Adjust.
28207 (check_strncat_sizes): Adjust.
28208 (expand_builtin_strncat): Adjust.
28209 (expand_builtin_strncpy): Adjust and simplify.
28210 (expand_builtin_memset): Adjust.
28211 (expand_builtin_bzero): Adjust.
28212 (expand_builtin_memcmp): Adjust.
28213 (expand_builtin): Handle memcmp.
28214 (maybe_emit_chk_warning): Check strncat just once.
28216 2017-05-04 Martin Sebor <msebor@redhat.com>
28218 PR preprocessor/79214
28219 PR middle-end/79222
28220 PR middle-end/79223
28221 * builtins.c (check_sizes): Add inlining context and issue
28222 warnings even when -Wno-system-headers is set.
28223 (check_strncat_sizes): Same.
28224 (expand_builtin_strncat): Same.
28225 (expand_builtin_memmove): New function.
28226 (expand_builtin_stpncpy): Same.
28227 (expand_builtin): Handle memmove and stpncpy.
28229 2017-05-04 Bin Cheng <bin.cheng@arm.com>
28231 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
28232 which is not used any more.
28234 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28236 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
28238 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28240 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
28241 (cortexa53_tunings): Likewise.
28242 (cortexa57_tunings): Likewise.
28243 (cortexa72_tunings): Likewise.
28244 (cortexa73_tunings): Likewise.
28246 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28248 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
28249 Set loop alignment to 8.
28251 2017-05-04 Martin Sebor <msebor@redhat.com>
28253 PR translation/80280
28254 * builtins.c (expand_builtin_object_size): Add missing quoting to
28255 %D and like directives.
28256 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
28257 (hsa_type_for_tree_type): Same.
28258 (verify_function_arguments): Same.
28259 * symtab.c (symbol_table::change_decl_assembler_name): Same.
28260 * varasm.c (get_section): Same.
28263 2017-05-04 Martin Sebor <msebor@redhat.com>
28265 PR translation/80280
28266 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
28268 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28270 * config/aarch64/aarch64.c (generic_addrcost_table):
28271 Change HI/TI mode setting.
28273 2017-05-04 Martin Jambor <mjambor@suse.cz>
28275 PR tree-optimization/80622
28276 * tree-sra.c (comes_initialized_p): New function.
28277 (build_accesses_from_assign): Only set write lazily when
28278 comes_initialized_p is false.
28279 (analyze_access_subtree): Use comes_initialized_p.
28280 (propagate_subaccesses_across_link): Assert !comes_initialized_p
28281 instead of testing for PARM_DECL.
28283 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28285 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
28286 constraint on operand 0 to allow more general addressing modes.
28287 Adjust output template.
28288 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
28290 * config/aarch64/aarch64-protos.h
28291 (aarch64_address_valid_for_prefetch_p): Declare prototype.
28292 * config/aarch64/constraints.md (Dp): New address constraint.
28293 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
28296 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
28298 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
28299 update use of estimate_ipcp_clone_size_and_time.
28300 (estimate_local_effects): Update use of
28301 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
28302 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
28303 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
28304 Return nonspecialized time.
28306 2017-05-04 Richard Biener <rguenther@suse.de>
28308 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
28309 for the last VUSE which def dominates the PHI. Directly call
28311 (get_continuation_for_phi_1): Remove.
28313 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
28315 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
28316 to explain the use of truncating division. Cap the number of
28317 iterations to the maximum given by nb_iterations_upper_bound,
28320 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28322 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
28323 * configure: Regenerate.
28324 * config.in: Regenerate.
28325 * config/i386/driver-mingw32.c: new file.
28326 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
28327 * config.host: Link driver-mingw32.o on MinGW host.
28328 * doc/install.texi: Document new --enable-mingw-wildcard configure
28331 2017-05-04 Marek Polacek <polacek@redhat.com>
28333 PR tree-optimization/80612
28334 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
28336 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28337 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
28339 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
28340 (movt splitter): Likewise.
28341 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
28342 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
28343 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
28344 block for Thumb-1 with MOVT.
28345 (thumb2_legitimate_address_p): Move code block ...
28346 (can_avoid_literal_pool_for_label_p): ... into this new function.
28347 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
28349 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
28350 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
28351 "M-profile targets with the MOVT instruction".
28353 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28355 * config/arm/arm-builtins.c (arm_init_builtins): Rename
28356 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
28357 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
28359 2017-05-04 Martin Liska <mliska@suse.cz>
28361 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
28362 variable cond_code.
28364 2017-05-04 Richard Biener <rguenther@suse.de>
28366 * tree.c (array_at_struct_end_p): Handle arrays at struct
28367 end with flexarrays more conservatively. Refactor and treat
28368 arrays of arrays or aggregates more strict. Fix
28369 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
28370 * tree.c (array_at_struct_end_p): Adjust prototype.
28371 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
28372 * gimple-fold.c (get_range_strlen): Likewise.
28373 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
28375 2017-05-04 Richard Biener <rguenther@suse.de>
28377 PR tree-optimization/31130
28378 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
28380 (supports_overflow_infinity): Likewise.
28381 (is_negative_overflow_infinity): Likewise.
28382 (is_positive_overflow_infinity): Likewise.
28383 (is_overflow_infinity): Likewise.
28384 (stmt_overflow_infinity): Likewise.
28385 (overflow_infinity_range_p): Likewise.
28386 (usable_range_p): Remove as always returning true.
28387 (make_overflow_infinity): Remove.
28388 (negative_overflow_infinity): Likewise.
28389 (positive_overflow_infinity): Likewise.
28390 (avoid_overflow_infinity): Likewise.
28391 (set_value_range): Adjust accordingly.
28392 (set_value_range_to_nonnegative): Likewise, remove now unused
28393 overflow_infinity arg.
28394 (vrp_operand_equal_p): Adjust.
28395 (update_value_range): Likewise.
28396 (range_int_cst_singleton_p): Likewise.
28397 (operand_less_p): Likewise.
28398 (compare_values_warnv): Likewise.
28399 (extract_range_for_var_from_comparison_expr): Likewise.
28400 (vrp_int_const_binop): Likewise.
28401 (zero_nonzero_bits_from_vr): Likewise.
28402 (extract_range_from_multiplicative_op_1): Likewise.
28403 (extract_range_from_binary_expr_1): Likewise.
28404 (extract_range_from_unary_expr): Likewise.
28405 (extract_range_from_comparison): Likewise.
28406 (extract_range_basic): Likewise.
28407 (adjust_range_with_scev): Likewise.
28408 (compare_ranges): Likewise.
28409 (compare_range_with_value): Likewise.
28410 (dump_value_range): Likewise.
28411 (test_for_singularity): Likewise, remove strict_overflow_p parameter
28413 (simplify_cond_using_ranges): Adjust.
28415 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28417 * brig-builtins.def: Added a builtin for class_f64.
28418 * builtin-types.def: Added a builtin type needed by class_f64.
28420 2017-05-03 Jason Merrill <jason@redhat.com>
28422 * timevar.def: Add TV_CONSTEXPR.
28424 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28426 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
28428 2017-05-03 Martin Jambor <mjambor@suse.cz>
28430 * ipa-prop.c (ipa_update_after_lto_read): Removed.
28431 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
28432 * ipa-cp.c (ipcp_propagate_stage): Do not call
28433 ipa_update_after_lto_read.
28434 * ipa-inline.c (ipa_inline): Likewise.
28436 2017-05-03 Martin Jambor <mjambor@suse.cz>
28438 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
28439 tag. Added a default constructor and a destructor.
28440 (ipa_edge_args_sum_t): New class;
28441 (ipa_edge_args_sum): Declare.
28442 (ipa_edge_args_vector): Remove declaration.
28443 (IPA_EDGE_REF): Use ipa_edge_args_sum.
28444 (ipa_free_edge_args_substructures): Remove declaration.
28445 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
28446 (ipa_edge_args_info_available_for_edge_p): Likewise.
28447 * ipa-prop.c (ipa_edge_args_vector): Removed.
28448 (edge_removal_hook_holder): Likewise.
28449 (edge_duplication_hook_holder): Likewise.
28450 (ipa_edge_args_sum): New variable.
28451 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
28452 ipa_edge_args_vector.
28453 (ipa_free_edge_args_substructures): Likewise.
28454 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
28455 ipa_edge_args_vector.
28456 (ipa_edge_removal_hook): Turned into method
28457 ipa_edge_args_sum_t::remove.
28458 (ipa_edge_duplication_hook): Turned into method
28459 ipa_edge_args_sum_t::duplicate.
28460 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
28461 registering edge hooks.
28462 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
28463 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
28464 ipa_edge_args_sum instead of ipa_edge_args_vector.
28465 * ipa-profile.c (ipa_profile): Likewise.
28467 2017-05-03 Martin Jambor <mjambor@suse.cz>
28469 * symbol-summary.h (function_summary): New method exists.
28470 (function_summary::symtab_removal): Deallocate through release.
28471 (call_summary): New class.
28472 (gt_ggc_mx): New overload.
28473 (gt_pch_nx): Likewise.
28474 (gt_pch_nx): Likewise.
28476 2017-05-03 Jeff Law <law@redhat.com>
28478 PR tree-optimization/78496
28479 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
28480 from simplify_cond_using_ranges. Split off code to walk
28481 backwards through casts into ...
28482 (simplify_cond_using_ranges_2): New function.
28483 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
28484 (execute_vrp): After identifying jump threads, call
28485 simplify_cond_using_ranges_2.
28487 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
28490 * ipa-inline.h (inline_summary): Add ctor.
28491 (create_ggc): Do not use ggc_cleared_alloc.
28493 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
28494 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28496 * gcc.c (handle_braces): Support escaping in switch matching
28498 * doc/invoke.texi (Spec Files): Document it.
28499 Remove superfluous @code markup in items.
28501 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28503 * diagnostic-show-locus.c (struct column_range): New struct.
28504 (get_affected_columns): New function.
28505 (get_printed_columns): New function.
28506 (struct correction): New struct.
28507 (correction::ensure_capacity): New function.
28508 (correction::ensure_terminated): New function.
28509 (struct line_corrections): New struct.
28510 (line_corrections::~line_corrections): New dtor.
28511 (line_corrections::add_hint): New function.
28512 (layout::print_trailing_fixits): Reimplement in terms of the new
28514 (selftest::test_overlapped_fixit_printing): New function.
28515 (selftest::diagnostic_show_locus_c_tests): Call it.
28517 2017-05-03 Nathan Sidwell <nathan@acm.org>
28519 Canonicalize canonical type hashing
28520 * tree.h (type_hash_canon_hash): Declare.
28521 * tree.c (type_hash_list, attribute_hash_list): Move into
28522 type_hash_canon_hash.
28523 (build_type_attribute_qual_variant): Break out hash code calc into
28524 type_hash_canon_hash.
28525 (type_hash_canon_hash): New. Generic type hash computation.
28526 (build_range_type_1, build_array_type_1, build_function_type,
28527 build_method_type_directly, build_offset_type, build_complex_type,
28528 make_vector_type): Call it.
28530 2017-05-03 Richard Biener <rguenther@suse.de>
28532 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
28533 When all DRs have unknown misaligned do not always peel
28534 when there is a store but apply the same costing model as if
28535 there were only loads.
28537 2017-05-03 Richard Biener <rguenther@suse.de>
28540 PR tree-optimization/80492
28541 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28542 compare_base_decls returning dont-know properly.
28544 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
28546 * config/arm/iterators.md (CCSI): New mode iterator.
28547 (arch): New mode attribute.
28548 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
28549 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
28550 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
28551 code iterator for success result mode.
28552 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
28553 the corresponding new insn generators.
28555 2017-05-03 Bin Cheng <bin.cheng@arm.com>
28558 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28559 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28561 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
28563 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
28564 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
28565 (DDR_A): Wrap DDR argument in brackets.
28566 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
28567 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
28568 (DDR_REVERSED_P): Likewise.
28570 2017-05-03 Jakub Jelinek <jakub@redhat.com>
28572 PR tree-optimization/79472
28573 * tree-switch-conversion.c (struct switch_conv_info): Add
28574 contiguous_range and default_case_nonstandard fields.
28575 (collect_switch_conv_info): Compute contiguous_range and
28576 default_case_nonstandard fields, don't clear final_bb if
28577 contiguous_range and only the default case doesn't have the required
28579 (check_all_empty_except_final): Set default_case_nonstandard instead
28580 of failing if contiguous_range and the default case doesn't have empty
28582 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
28583 and only the default case doesn't have the required constants. Skip
28585 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
28586 if default_case_nonstandard.
28587 (build_constructors): Build constant 1 just once. Assert that default
28588 values aren't inserted in between cases if contiguous_range. Skip
28590 (build_arrays): Skip virtual phis.
28591 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
28592 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
28593 Handle virtual phis.
28594 (gen_inbound_check): Handle default_case_nonstandard case.
28595 (process_switch): Adjust check_final_bb caller. Call
28596 gather_default_values with the first non-default case instead of
28597 default case if default_case_nonstandard.
28599 2017-05-02 Nathan Sidwell <nathan@acm.org>
28601 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
28602 check. Fix formatting.
28604 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
28606 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
28607 errors when comparing specialized and unspecialized times.
28609 2017-05-02 David Malcolm <dmalcolm@redhat.com>
28611 * diagnostic-show-locus.c
28612 (layout::should_print_annotation_line_p): Make private.
28613 (layout::print_annotation_line): Make private.
28614 (layout::annotation_line_showed_range_p): Make private.
28615 (layout::show_ruler): Make private.
28616 (layout::print_source_line): Make private. Pass in line and
28617 line_width, rather than calling location_get_source_line. Drop
28619 (layout::print_leading_fixits): New method.
28620 (layout::print_any_fixits): Rename to...
28621 (layout::print_trailing_fixits): ...this, and make private.
28622 Don't print newline fixits.
28623 (diagnostic_show_locus): Move logic for printing one row into...
28624 (layout::print_line): ...this new function. Move the
28625 location_get_source_line call and error-handling from
28626 print_source_line to here. Call print_leading_fixits, and rename
28627 print_any_fixits to print_trailing_fixits.
28628 (selftest::test_fixit_insert_containing_newline): Update now that
28629 newlines are partially supported.
28630 (selftest::test_fixit_insert_containing_newline_2): New test.
28631 (selftest::test_fixit_replace_containing_newline): Update comments.
28632 (selftest::diagnostic_show_locus_c_tests): Call the new test.
28633 * edit-context.c (class added_line): New class.
28634 (class edited_line): Describe newline handling in comment.
28635 (edited_line::actually_edited_p): New method.
28636 (edited_line::print_content): Delete redundant decl.
28637 (edited_line::m_predecessors): New field.
28638 (edited_file::print_content): Call edited_line::print_content.
28639 (edited_file::print_diff): Update to support newlines.
28640 (edited_file::print_diff_hunk): Likewise.
28641 (edited_file::print_run_of_changed_lines): New function.
28642 (edited_file::print_diff_line): Convert to...
28643 (print_diff_line): ...this.
28644 (edited_file::get_effective_line_count): New function.
28645 (edited_line::edited_line): Initialize new field m_predecessors.
28646 (edited_line::~edited_line): Clean up m_predecessors.
28647 (edited_line::apply_fixit): Handle newlines.
28648 (edited_line::get_effective_line_count): New function.
28649 (edited_line::print_content): New function.
28650 (edited_line::print_diff_lines): New function.
28651 (selftest::test_applying_fixits_insert_containing_newline): New
28653 (selftest::test_applying_fixits_replace_containing_newline): New
28655 (selftest::insert_line): New function.
28656 (selftest::test_applying_fixits_multiple_lines): Add example of
28658 (selftest::edit_context_c_tests): Call the new tests.
28660 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28662 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
28663 parameter cand. Update dump information.
28664 (get_computation_cost): Update uses.
28666 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28668 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
28669 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
28670 (get_computation_at, rewrite_use_address): Update use of
28671 get_computation_aff.
28673 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28675 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
28676 (get_computation): Delete.
28677 (get_computation_cost): Implement like get_computation_cost_at.
28678 Use get_computation_at.
28679 (get_computation_cost_at): Delete.
28680 (rewrite_use_nonlinear_expr): Use get_computation_at.
28681 (rewrite_use_compare, remove_unused_ivs): Ditto.
28683 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28685 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
28687 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28689 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
28690 (ivopts_global_cost_for_size): Rename parameter and update uses.
28691 (iv_ca_recount_cost): Update uses.
28692 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
28693 candidates seperately in n_invs and n_cands.
28694 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
28696 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28698 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
28699 (find_inv_vars_cb): New.
28700 (find_depends): Renamed to ...
28701 (find_inv_vars): ... this.
28702 (add_candidate_1, force_var_cost): Call find_inv_vars.
28703 (split_address_cost, determine_group_iv_cost_cond): Ditto.
28705 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28707 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
28708 inv_vars. Add inv_exprs.
28709 (struct iv_cand): Rename depends_on to inv_vars.
28710 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
28711 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
28712 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
28713 (dump_cand): Dump inv_vars.
28714 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
28715 (record_invariant, find_depends, add_candidate_1): Ditto.
28716 (set_group_iv_cost, force_var_cost): Ditto.
28717 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
28718 (get_computation_cost_at, get_computation_cost): Ditto.
28719 (determine_group_iv_cost_generic): Ditto.
28720 (determine_group_iv_cost_address): Ditto.
28721 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
28722 (determine_group_iv_costs): Ditto.
28723 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
28724 (iv_ca_set_remove_invariants): Renamed to ...
28725 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
28726 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
28727 (iv_ca_set_add_invariants): Renamed to ...
28728 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
28729 (iv_ca_set_cp): Use iv_ca_set_add_invs.
28730 (iv_ca_has_deps): Support inv_vars and inv_exprs.
28731 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
28732 (create_new_ivs): Remove useless dump.
28734 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28736 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
28738 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
28739 (iv_ca_set_no_cp, create_new_iv): Ditto.
28741 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28743 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28745 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28747 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
28748 function tree_check2.
28750 2017-05-02 Martin Liska <mliska@suse.cz>
28752 * doc/gcov.texi: Add missing preposition.
28753 * gcov.c (function_info::function_info): Properly fill up
28754 all member variables.
28756 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28758 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
28760 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28762 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
28764 2017-05-02 Martin Liska <mliska@suse.cz>
28767 * lto-streamer-in.c (lto_read_tree_1): Remove
28768 LTO_STREAMER_DEBUG.
28769 * lto-streamer.c (struct tree_hash_entry): Likewise.
28770 (struct tree_entry_hasher): Likewise.
28771 (tree_entry_hasher::hash): Likewise.
28772 (tree_entry_hasher::equal): Likewise.
28773 (lto_streamer_init): Likewise.
28774 (lto_orig_address_map): Likewise.
28775 (lto_orig_address_get): Likewise.
28776 (lto_orig_address_remove): Likewise.
28777 * lto-streamer.h: Likewise.
28778 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
28779 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
28781 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
28783 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
28784 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
28785 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
28786 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
28787 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
28788 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
28789 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
28790 (mm_maskz_sub_ss): New intrinsics.
28791 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28792 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28793 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
28794 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
28795 (__builtin_ia32_subss_mask_round): New builtins.
28796 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28797 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28798 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
28800 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
28801 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28803 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28806 2017-05-02 Martin Jambor <mjambor@suse.cz>
28808 PR tree-optimization/78687
28809 * tree-sra.c (access): New field parent.
28810 (process_subtree_disqualification): New function.
28811 (disqualify_candidate): Call it.
28812 (build_accesses_from_assign): Reset write flag if creating an
28814 (build_access_subtree): Fill in parent field and also prpagate
28815 down grp_write flag.
28816 (create_artificial_child_access): New parameter set_grp_write, set
28817 grp_write to its value.
28818 (propagate_subaccesses_across_link): Also propagate grp_write flag
28820 (propagate_all_subaccesses): Push the closest parent back to work
28821 queue if add_access_to_work_queue returned true.
28823 2017-05-02 Richard Biener <rguenther@suse.de>
28825 * common.opt (fstrict-overflow): Alias negative to fwrapv.
28826 * doc/invoke.texi (fstrict-overflow): Remove all traces of
28827 -fstrict-overflow documentation.
28828 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
28829 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
28830 flag_strict_overflow.
28831 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
28832 * lto-opts.c (lto_write_options): Do not stream it.
28833 * lto-wrapper.c (merge_and_complain): Do not handle it.
28834 * opts.c (default_options_table): Do not set -fstrict-overflow.
28835 (finish_options): Likewise do not clear it when sanitizing.
28836 * simplify-rtx.c (simplify_const_relational_operation): Do not
28837 test flag_strict_overflow.
28839 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28841 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
28842 using enabled attribute.
28843 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
28844 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
28845 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
28846 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
28847 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
28848 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
28849 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
28850 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
28851 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
28852 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
28854 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28856 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
28858 2017-05-02 Richard Biener <rguenther@suse.de>
28860 PR tree-optimization/80591
28862 2017-04-10 Richard Biener <rguenther@suse.de>
28864 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28867 2017-05-02 Richard Biener <rguenther@suse.de>
28869 PR tree-optimization/80549
28870 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
28871 (cleanup_tree_cfg_noloop): Create forwarders to known loop
28872 headers if they do not have a preheader.
28874 2017-05-02 Martin Liska <mliska@suse.cz>
28877 * common.opt: Fix typo.
28878 * doc/invoke.texi: Likewise.
28880 2017-05-01 Jan Beulich <jbeulich@suse.com>
28882 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
28883 swapping, add (x,x,m,x,n) alternative.
28885 2017-05-01 Nathan Sidwell <nathan@acm.org>
28887 * calls.c (combine_pending_stack_adjustment_and_call): Remove
28888 unnecessary unadjusted_alignment check.
28890 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28893 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
28895 * gimplify.c (gimplify_cilk_detach): New function.
28896 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
28897 * tree-core.h: Document EXPR_CILK_SPAWN.
28898 * tree.h (EXPR_CILK_SPAWN): Define.
28900 2017-05-01 David Malcolm <dmalcolm@redhat.com>
28902 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
28903 to use new fixit_hint representation, using the "replace" logic.
28904 (get_line_span_for_fixit_hint): Likewise.
28905 (layout::print_any_fixits): Likewise.
28906 (selftest::test_one_liner_many_fixits): Rename to...
28907 (selftest::test_one_liner_many_fixits_1): ...this, and update
28908 comment and expected output to reflect that the multiple fix-it
28909 hints are now consolidated into one insertion.
28910 (selftest::test_one_liner_many_fixits_2): New test.
28911 (selftest::test_diagnostic_show_locus_one_liner): Update for
28913 (selftest::test_fixit_consolidation): Update for fix-it API
28915 * diagnostic.c (print_parseable_fixits): Likewise.
28916 * edit-context.c (edited_line::m_line_events): Convert from
28917 auto_vec <line_event *> to auto_vec <line_event>.
28918 (class line_event): Convert from abstract base class to a concrete
28919 class, taking over the role of replace_event.
28920 (class insert_event): Delete.
28921 (class replace_event): Rename to class line_event. Convert to
28923 (edit_context::add_fixits): Reimplement.
28924 (edit_context::apply_insert): Delete.
28925 (edit_context::apply_replace): Rename to...
28926 (edit_context::apply_fixit): ...this. Convert to half-open range.
28927 (edited_file::apply_insert): Delete.
28928 (edited_file::apply_replace): Rename to...
28929 (edited_file::apply_fixit): ...this.
28930 (edited_line::~edited_line): Drop deletion of events.
28931 (edited_line::apply_insert): Delete.
28932 (edited_line::apply_replace): Rename to...
28933 (edited_line::apply_fixit): ...this. Convert to half-open range.
28934 Update for change to type of m_line_events.
28935 * edit-context.h (edit_context::apply_insert): Delete.
28936 (edit_context::apply_replace): Rename to...
28937 (edit_context::apply_fixit): ...this.
28939 2017-05-01 Martin Sebor <msebor@redhat.com>
28941 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
28944 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
28947 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
28948 __get_cpuid_max returns 0.
28949 (__get_cpuid_count): Ditto.
28951 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
28953 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
28954 replacement expression is another instance of one of its arguments.
28956 2017-05-01 Jakub Jelinek <jakub@redhat.com>
28959 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
28960 check for stack push/pop autoinc.
28961 * config/i386/i386.c (ix86_agi_dependent): Return false
28962 if the only reason why modified_in_p returned true is that
28963 addr is SP based and set_insn is a push or pop.
28965 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28967 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
28970 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28973 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
28974 (account_size_time): Use two predicates - exec_pred and
28976 (evaluate_conditions_for_known_args): Compute both clause and
28978 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
28979 (inline_summary_t::duplicate): Update.
28980 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
28982 (compute_inline_parameters): Likewise.
28983 (estimate_edge_size_and_time): Update caluclation of time.
28984 (estimate_node_size_and_time): Compute both time and nonspecialized
28986 (estimate_ipcp_clone_size_and_time): Update.
28987 (inline_merge_summary): Update.
28988 (do_estimate_edge_time): Update.
28989 (do_estimate_edge_size): Update.
28990 (do_estimate_edge_hints): Update.
28991 (inline_read_section, inline_write_summary): Stream both new predicates.
28992 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
28994 (compute_inlined_call_time): Cleanup.
28995 (big_speedup_p): Update.
28996 (edge_badness): Update.
28997 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
28998 (size_time_entry): Replace predicate by exec_predicate and
28999 nonconst_predicate.
29000 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29001 (estimate_edge_time): Return also nonspec_time.
29002 (reset_edge_growth_cache): Update.
29004 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29006 PR rtl-optimization/80491
29007 * ifcvt.c (noce_process_if_block): When looking for x setter
29008 with missing else_bb, don't check only the insn right before
29009 cond_earliest, but look for the last insn that x is modified in
29010 within the same bb.
29012 PR rtl-optimization/80491
29013 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29015 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29017 PR tree-optimization/80487
29018 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
29020 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29022 PR tree-optimization/79697
29023 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
29024 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
29025 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
29027 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
29028 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
29030 2017-04-28 Martin Sebor <msebor@redhat.com>
29032 PR tree-optimization/80523
29033 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
29034 (init_target_to_host_charmap, target_to_host, target_strtol10): New
29036 (maybe_warn, format_directive, parse_directive): Use new functions.
29037 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
29039 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
29041 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
29043 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
29045 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
29046 target_header_dir): Set correctly.
29047 * configure: Regenerated.
29048 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29049 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29050 instead of SYSTEM_HEADER_DIR.
29052 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
29054 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
29055 (estimate_local_effects): Likewise.
29056 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
29057 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
29058 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
29059 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
29060 do_estimate_edge_time, estimate_edge_time): Likewise.
29061 * ipa-inline-analysis.c (estimate_node_size_and_time,
29062 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
29063 (estimate_time_after_inlining): Remove.
29065 2017-04-28 Martin Liska <mliska@suse.cz>
29067 * doc/gcov.texi: Enhance documentation of gcov.
29069 2017-04-28 Martin Liska <mliska@suse.cz>
29071 * doc/gcov.texi: Sort options in alphabetic order.
29072 * doc/gcov-dump.texi: Likewise.
29073 * doc/gcov-tool.texi: Likewise.
29074 * gcov.c (print_usage): Likewise.
29075 * gcov-dump.c (print_usage): Likewise.
29076 * gcov-tool.c (print_merge_usage_message): Likewise.
29077 (print_rewrite_usage_message): Likewise.
29078 (print_overlap_usage_message): Likewise.
29080 2017-04-28 Martin Liska <mliska@suse.cz>
29082 PR gcov-profile/53915
29083 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
29085 2017-04-28 Martin Liska <mliska@suse.cz>
29087 PR gcov-profile/79891
29088 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
29089 is marked by compiler as living on a line.
29090 (get_cycles_count): Remove usage of the union.
29091 (output_intermediate_file): Likewise.
29092 (find_source): Fix GNU coding style.
29093 (accumulate_line_counts): Remove old non-all block mode.
29094 (output_lines): Remove usage of the union.
29095 * profile.c (output_location): Include all BBs, even if
29096 belonging to a same line (and file) as a previous BB.
29098 2017-04-28 Martin Liska <mliska@suse.cz>
29100 * gcov.c (process_args): Handle new argument 'w'.
29101 (read_graph_file): Assign ID to BBs.
29102 (output_branch_count): Display BB # if verbose flag is set.
29103 (output_lines): Likewise for arcs.
29104 (print_usage): Add '--verbose' option help.
29105 * doc/gcov.texi: Document --verbose (-w) option.
29107 2017-04-28 Martin Liska <mliska@suse.cz>
29109 * gcov.c (struct block_location_info): New struct.
29110 (process_file): Fill up the new structure.
29111 (read_graph_file): Replace usage of encoding by the newly added
29113 (add_line_counts): Likewise.
29114 (accumulate_line_counts): Remove usage of the union.
29115 (function_info::function_info): New function.
29116 (function_info::~function_info): Likewise.
29117 (process_file): Call delete instead of release_function.
29118 (release_function): Release the function.
29119 (release_structures): Call delete instead of release_function.
29120 (solve_flow_graph): Replace usage of num_blocks.
29121 (find_exception_blocks): Likewise.
29122 (output_lines): Fix GNU coding style.
29124 2017-04-28 Martin Liska <mliska@suse.cz>
29127 * coverage.c (coverage_remove_note_file): New function.
29128 * coverage.h: Declare the function.
29129 * toplev.c (finalize): Clean if an error has been seen.
29131 2017-04-28 Martin Liska <mliska@suse.cz>
29133 PR gcov-profile/80031
29134 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
29135 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
29136 * gcov.c (read_graph_file): Read just number of blocks.
29137 * profile.c (branch_prob): Do not stream 0 flags per a basic
29140 2017-04-28 Martin Liska <mliska@suse.cz>
29142 * gcov-dump.c (tag_*): Add new argument to declarations.
29143 (dump_gcov_file): Likewise.
29144 (tag_blocks): Add and use new argument depth.
29145 (tag_arcs): Likewise.
29146 (tag_lines): Likewise.
29147 (tag_counters): Likewise.
29148 (tag_summary): Likewise.
29149 (dump_working_sets): Use depth to do a proper indentation.
29151 2017-04-28 Jakub Jelinek <jakub@redhat.com>
29154 * cgraph.h (symtab_node::debug_symtab): No longer inline.
29155 * symtab.c (symtab_node::debug_symtab): Move definition here.
29157 2017-04-28 Richard Biener <rguenther@suse.de>
29159 * lto-streamer.h (LTO_major_version): Bump to 7.
29161 2017-04-28 Richard Biener <rguenther@suse.de>
29163 * tree-vrp.c (assert_info): New struct.
29164 (add_assert_info): New helper.
29165 (register_edge_assert_for_2): Refactor to add asserts to a vector
29167 (register_edge_assert_for_1): Likewise.
29168 (register_edge_assert_for): Likewise.
29169 (finish_register_edge_assert_for): New helper actually registering
29170 asserts where live on edge.
29171 (find_conditional_asserts): Adjust.
29172 (find_switch_asserts): Likewise.
29173 (evrp_dom_walker::try_find_new_range): Generalize.
29174 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
29176 2017-04-27 Marek Polacek <polacek@redhat.com>
29179 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
29180 arg10 and arg11 to itype.
29182 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
29184 * doc/extend.texi (Object Size Checking): Improve grammar.
29186 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
29189 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
29190 that the logic for permitting reciprocal estimates matches that
29193 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29196 * tree.c (type_cache_hasher::equal): Only compare
29197 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
29198 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
29199 non-aggregate element types.
29200 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
29201 about the flag on ARRAY_TYPEs in the comment, formatting fix.
29203 2017-04-27 Richard Biener <rguenther@suse.de>
29205 PR middle-end/80533
29206 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
29207 stripping ARRAY_REFs from MEM_EXPR make sure we're not
29208 keeping a reference to a trailing array.
29210 2017-04-27 Richard Biener <rguenther@suse.de>
29212 PR middle-end/80539
29213 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
29214 being in loop-closed SSA form conservatively.
29215 (chrec_fold_multiply_poly_poly): Likewise.
29217 2017-04-27 Tamar Christina <tamar.christina@arm.com>
29219 PR middle-end/79665
29220 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
29221 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
29223 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29226 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
29227 (aarch64_function_arg_alignment): Return unsigned int again, but still
29228 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
29229 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29230 Don't emit -Wpsabi note.
29231 (aarch64_function_arg_boundary): Likewise.
29232 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29235 2017-04-26 Nathan Sidwell <nathan@acm.org>
29237 * tree.h (crc32_unsigned_n): Declare.
29238 (crc32_unsigned, crc32_unsigned): Make inline.
29239 * tree.c (crc32_unsigned_bits): Replace with ...
29240 (crc32_unsigned_n): ... this.
29241 (crc32_unsigned, crc32_byte): Remove.
29242 (crc32_string): Remove unnecessary braces.
29244 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29246 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
29247 * ipa-inline-analysis.c (MAX_TIME): Remove.
29248 (account_size_time): Use sreal for time.
29249 (dump_inline_summary): Update.
29250 (estimate_function_body_sizes): Update.
29251 (estimate_edge_size_and_time): Update.
29252 (estimate_calls_size_and_time): Update.
29253 (estimate_node_size_and_time): Update.
29254 (inline_merge_summary): Update.
29255 (inline_update_overall_summary): Update.
29256 (estimate_time_after_inlining): Update.
29257 (inline_read_section): Update.
29258 (inline_write_summary): Update.
29259 * ipa-inline.c (compute_uninlined_call_time): Update.
29260 (compute_inlined_call_time): Update.
29261 (recursive_inlining): Update.
29262 (inline_small_functions): Update.
29263 (dump_overall_stats): Update.
29264 * ipa-inline.h: Include sreal.h.
29265 (size_time_entry): Turn time to sreal.
29266 (inline_summary): Turn self_time nad time to sreal.
29268 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29270 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
29272 (sreal::stream_out, sreal::stream_in): New.
29273 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
29275 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29277 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
29280 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
29283 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
29285 (dimode_scalar_chain::compute_convert_gain): Ditto.
29286 (dimode_scalar_chain::make_vector_copies): Ditto.
29287 (dimode_scalar_chain::convert_reg): Ditto.
29288 (dimode_scalar_chain::convert_insn): Ditto.
29289 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
29290 (VI248_AVX512BW_1): New mode iterator.
29291 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
29292 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
29295 2017-04-25 Martin Sebor <msebor@redhat.com>
29297 PR tree-optimization/80497
29298 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
29299 constants are representable in HOST_WIDE_INT.
29300 (parse_directive): Ditto.
29302 2017-04-25 Martin Sebor <msebor@redhat.com>
29305 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
29306 (new_zero_array): Adjust signature.
29307 (dom_info::dom_init): Used unsigned rather that size_t.
29308 (dom_info::dom_info): Same.
29310 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29311 Jakub Jelinek <jakub@redhat.com>
29314 * config/arm/arm.c: Include gimple.h.
29315 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29316 returns negative, increment ncrn only if it returned positive.
29317 (arm_needs_doubleword_align): Return int instead of bool,
29318 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
29319 members, but if there is any such non-FIELD_DECL
29320 > PARM_BOUNDARY aligned decl, return -1 instead of false.
29321 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29322 returns negative, increment nregs only if it returned positive.
29323 (arm_setup_incoming_varargs): Likewise.
29324 (arm_function_arg_boundary): Emit -Wpsabi note if
29325 arm_needs_doubleword_align returns negative, return
29326 DOUBLEWORD_ALIGNMENT only if it returned positive.
29328 2017-04-25 Marek Polacek <polacek@redhat.com>
29331 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29332 first argument to type.
29334 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
29337 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
29338 type checks to test for compatibility instead of equality.
29340 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29341 Jakub Jelinek <jakub@redhat.com>
29344 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
29346 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
29347 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
29348 the alignment computation, but return their maximum in warn_alignment.
29349 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29350 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
29352 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
29353 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29356 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29358 * config/arc/simdext.md (dmpyh): Fix typo.
29360 2017-04-25 Richard Biener <rguenther@suse.de>
29362 PR tree-optimization/80492
29363 * alias.c (compare_base_decls): Handle registers with asm
29364 specification conservatively.
29365 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29366 compare_base_decls returning dont-know properly.
29368 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29370 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
29371 (legitimate_offset_address_p): New function.
29372 (arc_legitimate_address_p): Use above function.
29374 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29376 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
29378 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29380 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
29381 ACCH registers whenever they are available.
29383 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29385 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
29386 double regs fix when not used.
29388 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29390 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
29392 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
29393 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
29395 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29397 * config/arc/arc.c (arc_output_addsi): Check for h-register class
29398 when emitting short ADD instructions.
29400 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29402 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
29404 (cmpsi_cc_c_insn): Likewise.
29405 (cbranchsi4_scratch): Compute proper instruction length using
29406 compact_hreg_operand.
29407 * config/arc/predicates.md (compact_hreg_operand): New predicate.
29409 2017-04-25 Richard Biener <rguenther@suse.de>
29411 PR middle-end/80509
29412 * passes.c (pass_manager::pass_manager): Initialize
29413 m_name_to_pass_map.
29415 2017-04-25 Richard Biener <rguenther@suse.de>
29417 PR tree-optimization/79201
29418 * tree-ssa-sink.c (statement_sink_location): Handle calls.
29420 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29423 * config/s390/vector.md: Split MEM->GPR vector moves for
29424 non-s_operand addresses.
29426 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29429 * config/s390/predicates.md (reload_const_wide_int_operand): New
29431 * config/s390/s390.md ("movti"): Remove d/P alternative.
29432 ("movti_bigconst"): New pattern definition.
29434 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29437 * s390-protos.h (s390_expand_cs_hqi): Removed.
29438 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
29439 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
29440 modes as well as CCZ1mode and CCZmode.
29441 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
29442 signature of s390_emit_compare_and_swap.
29443 (s390_expand_cs_hqi): Likewise, make static.
29444 (s390_expand_cs_tdsi): Generate an explicit compare before trying
29445 compare-and-swap, in some cases.
29446 (s390_expand_cs): Wrapper function.
29447 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
29449 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
29450 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
29451 patterns for small and large integers. Forbid symref memory operands.
29452 Move expander to s390.c. Require cc register.
29453 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
29454 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
29455 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
29456 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
29457 symref memory operands. Remove CC mode and call s390_match_ccmode
29459 ("atomic_exchange<mode>"): Allow and implement all integer modes.
29461 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29463 * config/s390/s390.md (define_peephole2): New peephole to help
29464 combining the load-and-test pattern with volatile memory.
29466 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29468 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
29469 with CCZmode for TARGET_Z196.
29471 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29473 PR rtl-optimization/80501
29474 * combine.c (make_compound_operation_int): Set subreg_code to SET
29475 even for AND with mask of the sign bit of mode.
29477 PR rtl-optimization/80500
29478 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
29479 sum's initial value.
29481 2017-04-25 Julian Brown <julian@codesourcery.com>
29482 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29484 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
29486 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
29488 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
29490 2017-04-25 Julian Brown <julian@codesourcery.com>
29491 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29493 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
29494 (thunderx2t99_sha): New Reservation.
29496 2017-04-25 Julian Brown <julian@codesourcery.com>
29497 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29499 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
29500 type for 1-element load.
29502 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
29504 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
29506 2017-04-24 Martin Jambor <mjambor@suse.cz>
29508 PR tree-optimization/80293
29509 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
29510 char arrays not totally scalarizable if it is false.
29511 (analyze_all_variable_accesses): Pass correct value in the new
29512 parameter. Add a statistics counter.
29514 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
29516 PR middle-end/79931
29517 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
29519 2017-04-24 Richard Biener <rguenther@suse.de>
29521 PR tree-optimization/80494
29522 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
29523 out for complex types.
29525 2017-04-24 Richard Biener <rguenther@suse.de>
29527 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
29528 * tree-ssa-sccvn.c (print_scc): Print SCC size.
29529 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
29530 (DFS): Adjust and never fail.
29531 (sccvn_dom_walker::fail): Remove.
29532 (sccvn_dom_walker::before_dom_children): Adjust.
29533 (run_scc_vn): Likewise and never fail.
29534 * tree-ssa-pre.c (pass_pre::execute): Adjust.
29535 (pass_fre::execute): Likewise.
29537 2017-04-24 Richard Biener <rguenther@suse.de>
29539 PR tree-optimization/79725
29540 * tree-ssa-sink.c (statement_sink_location): Return whether
29541 failure reason was zero uses. Move that check later.
29542 (sink_code_in_bb): Deal with zero uses by removing the stmt
29545 2017-04-24 Richard Biener <rguenther@suse.de>
29548 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
29549 pointer-based references.
29551 2017-04-24 Richard Biener <rguenther@suse.de>
29554 * pass_manager.h (pass_manager::operator new): Remove.
29555 (pass_manager::operator delete): Likewise.
29556 * passes.c (pass_manager::operator new): Remove.
29557 (pass_manager::operator delete): Likewise.
29558 (pass_manager::pass_manager): Zero individual pass members.
29560 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
29563 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
29564 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
29565 Check "XEXP (src, 1)" operand here.
29566 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
29567 Check "XEXP (src, 1)" operand here.
29568 (dimode_scalar_chain::make_vector_copies): Detect count register
29569 of a shift instruction. Zero extend count register from QImode
29570 to DImode to satisfy vector shift pattern count operand predicate.
29571 Substitute vector shift count operand with a DImode copy.
29572 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
29575 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
29577 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
29578 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29579 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
29580 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29581 (UNSPEC_NOREX_MEM): Remove definition.
29583 2017-04-21 Richard Biener <rguenther@suse.de>
29585 PR tree-optimization/79547
29586 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29587 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
29588 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
29589 without any constraints.
29591 2017-04-21 Richard Biener <rguenther@suse.de>
29593 PR tree-optimization/78847
29594 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
29596 2017-04-21 Richard Biener <rguenther@suse.de>
29598 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
29599 (build_distinct_type_copy): Likewise.
29600 (build_variant_type_copy): Likewise.
29601 * tree.c (build_qualified_type): Pass down mem-stat info.
29602 (build_distinct_type_copy): Likewise.
29603 (build_variant_type_copy): Likewise.
29605 2017-04-21 Richard Biener <rguenther@suse.de>
29607 PR tree-optimization/80237
29608 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
29610 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
29611 for a simplified result.
29613 2016-04-21 Richard Biener <rguenther@suse.de>
29615 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
29616 sth as strict as a simple_iv but a chrec without symbols and an
29617 operand defined in the loop we are peeling (and not some subloop).
29618 (propagate_constants_for_unrolling): Propagate all constants.
29620 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29623 * config/i386/i386.c (print_reg): Remove assert for disalowed
29624 regno values, call output_operand_lossage instead.
29626 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29629 * config/i386/constraints.md (Yc): New register constraint.
29630 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
29631 Use Yc constraint for alternative 2 of operand 0. Remove
29632 preferred_for_speed attribute.
29634 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29636 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
29637 lastprivate clauses in SIMT case.
29639 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
29641 * doc/invoke.texi (-Wextra-semi): Document new warning option.
29643 2017-04-20 Richard Biener <rguenther@suse.de>
29645 PR tree-optimization/57796
29646 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
29647 as N scalar stores.
29648 (vect_model_load_cost): Cost gathers as N scalar loads.
29650 2017-04-20 Richard Biener <rguenther@suse.de>
29652 * ggc-page.c (ggc_allocated_p): Rename to ...
29653 (safe_lookup_page_table_entry): ... this and return the lookup
29655 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
29657 2017-04-20 Richard Biener <rguenther@suse.de>
29659 PR tree-optimization/80453
29660 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
29661 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
29662 from the conditions.
29663 (vn_phi_eq): Pass them down.
29664 (vn_phi_lookup): Record them.
29665 (vn_phi_insert): Likewise.
29667 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
29669 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
29670 uninitialized variable warning to avoid buffer overrun.
29672 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29675 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
29676 is suppressed for '{ 0 }' in C.
29678 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29680 * BASE-VER: Set to 8.0.0.
29682 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
29684 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
29685 priority .init_array and .fini_array section with SECTION_NOTYPE
29688 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29690 PR middle-end/80423
29691 * tree.h (build_array_type): Add typeless_storage default argument.
29692 * tree.c (type_cache_hasher::equal): Also compare
29693 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
29694 (build_array_type): Add typeless_storage argument, set
29695 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
29697 (build_nonshared_array_type): Adjust build_array_type_1 caller.
29698 (build_array_type): Likewise. Add typeless_storage argument.
29700 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29701 Jakub Jelinek <jakub@redhat.com>
29703 PR tree-optimization/80426
29704 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
29705 operation on symbolic operands, also compute the overflow for the
29706 invariant part when the operation degenerates into a negation.
29708 2017-04-19 Jakub Jelinek <jakub@redhat.com>
29711 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
29712 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
29715 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
29717 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
29720 * config/avr/avr.c (tree.h): Include it.
29721 (cgraph.h): Include it.
29722 (avr_encode_section_info): Don't warn for uninitialized progmem
29723 variable if it's just an alias.
29725 2017-04-19 Richard Biener <rguenther@suse.de>
29728 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
29729 when needed by AutoPGO.
29731 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
29734 * doc/lto.texi: Remove an extra 'that'.
29736 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
29738 PR rtl-optimization/80429
29739 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
29740 are only used in debug insns.
29742 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29743 Vladimir Makarov <vmakarov@redhat.com>
29745 * config/sparc/predicates.md (input_operand): Add comment. Return
29746 true for any memory operand when LRA is in progress.
29747 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
29749 2017-04-18 Jeff Law <law@redhat.com>
29752 * mips.md ({return,simple_return}_internal): Do not overwrite
29755 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29757 PR tree-optimization/80443
29758 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
29759 instead of adding 1, subtract -1 and similarly instead of subtracting
29762 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
29764 PR rtl-optimization/80357
29765 * haifa-sched.c (tmp_bitmap): New variable.
29766 (model_recompute): Handle duplicate use records.
29767 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
29768 (free_global_sched_pressure_data): Free it.
29770 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
29773 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29774 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29775 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29776 instead of SYSTEM_HEADER_DIR.
29778 2017-04-18 Jeff Law <law@redhat.com>
29780 PR middle-end/80422
29781 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
29782 predecessors after walking up the insn chain.
29784 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29787 * dwarf2out.c (modified_type_die): Try harder not to emit internal
29788 sizetype type into debug info.
29790 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29793 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
29794 unneeded test for TARGET_UPPER_REGS_SF.
29795 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
29797 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29800 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
29801 instead of gsi_after_labels.
29803 2017-04-18 Jeff Law <law@redhat.com>
29805 * regcprop.c (maybe_mode_change): Avoid creating copies of the
29809 2017-04-13 Jeff Law <law@redhat.com>
29810 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29811 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29813 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
29816 * config/avr/avr.c (intl.h): Include it.
29817 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
29819 2017-04-18 Martin Liska <mliska@suse.cz>
29821 PR gcov-profile/78783
29822 * gcov-tool.c (gcov_output_files): Validate that destination
29823 file is either removed by the tool or by a user.
29825 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29826 Guy Benyei <guybe@mellanox.com>
29828 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
29829 block, and do not negate it, the stored id is already negative.
29831 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29833 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
29835 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29838 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
29839 masks of options that should be turned off if the VSX vector
29840 options are turned off.
29841 (OTHER_P8_VECTOR_MASKS): Likewise.
29842 (OTHER_VSX_VECTOR_MASKS): Likewise.
29843 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
29844 rs6000_disable_incompatible_switches to validate no type switches
29846 (rs6000_incompatible_switch): New function to disallow turning on
29847 other vector options if -mno-vsx, -mno-power8-vector, or
29848 -mno-power9-vector are specified.
29850 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29852 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
29854 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29856 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
29857 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
29858 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
29859 (ARG_POINTER_CFA_OFFSET): Likewise.
29861 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29863 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
29864 conditions to take advantage of various optimizations.
29866 2017-04-13 Jeff Law <law@redhat.com>
29868 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29869 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29870 (zero_extendsidi2_dext): Likewise.
29872 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29875 * fold-const.c (fold_ternary_loc): Revert
29876 use op0 instead of fold_convert_loc (loc, type, arg0) part of
29879 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
29881 PR rtl-optimization/80343
29882 * lra-remat.c (update_scratch_ops): Assign original hard reg to
29883 new scratch pseudo.
29885 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
29888 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
29889 to ubsan_encode_value.
29891 2017-04-13 Jeff Law <law@redhat.com>
29893 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
29894 appearing in DEBUG_INSNs.
29896 2017-04-13 Martin Liska <mliska@suse.cz>
29898 PR gcov-profile/80413
29899 * gcov-io.c (gcov_write_string): Copy to buffer just when
29900 allocated size is greater than zero.
29902 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29905 * dwarf2out.c (decls_for_scope): Ignore declarations of
29906 current_function_decl in BLOCK_NONLOCALIZED_VARS.
29908 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
29911 * ipa-visibility.c (non_local_p): Fix typos.
29912 (localize_node): When localizing symbol in same comdat group,
29913 dissolve the group only when we know external symbols are going
29915 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
29917 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29919 PR tree-optimization/79390
29920 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
29921 order does not result in usable sequence, retry with reversed operand
29927 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
29928 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
29929 op0 instead of fold_convert_loc (loc, type, arg0).
29931 2017-04-12 Jeff Law <law@redhat.com>
29933 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
29934 has a delay slot in the generated code.
29936 * config/cris/cris.md (cris_preferred_reload_class): Return
29937 GENNONACR_REGS rather than GENERAL_REGS.
29939 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29942 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
29943 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
29944 signedness of the result type.
29946 2017-04-12 Richard Biener <rguenther@suse.de>
29947 Jeff Law <law@redhat.com>
29949 PR tree-optimization/80359
29950 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
29951 trim stores to TARGET_MEM_REFs.
29953 2017-04-12 Richard Biener <rguenther@suse.de>
29955 PR tree-optimization/79390
29956 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
29957 threading case even more.
29959 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
29962 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
29963 for quad_address_p for TImode, instead of just not indexed_address.
29965 2017-04-12 Richard Biener <rguenther@suse.de>
29966 Bernd Edlinger <bernd.edlinger@hotmail.de>
29968 PR middle-end/79671
29969 * alias.c (component_uses_parent_alias_set_from): Handle
29970 TYPE_TYPELESS_STORAGE.
29971 (get_alias_set): Likewise.
29972 * tree-core.h (tree_type_common): Add typeless_storage flag.
29973 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
29974 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
29975 for types containing members with TYPE_TYPELESS_STORAGE.
29976 (place_field): Likewise.
29977 (layout_type): Likewise for ARRAY_TYPE.
29978 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
29979 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
29980 TYPE_TYPELESS_STORAGE.
29981 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
29983 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29986 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29987 first argument to type.
29989 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29993 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
29994 CONST0_RTX (mode) rather than const0_rtx where appropriate.
29995 (rs6000_expand_binop_builtin): Likewise.
29996 (rs6000_expand_ternop_builtin): Likewise; also add missing
29997 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
29998 vshasigma built-ins.
29999 * doc/extend.texi: Document that vec_xxpermdi's third argument
30000 must be a constant.
30002 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30004 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30005 Use shift_const cost parameter when calculating gain of STV shifts.
30007 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30009 PR rtl-optimization/70478
30010 * lra-constraints.c (process_alt_operands): Check memory for
30011 disfavoring memory insn operand.
30013 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30015 PR middle-end/80100
30016 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30017 left shift in unsigned HOST_WIDE_INT type.
30019 PR rtl-optimization/80385
30020 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
30021 (not (neg X)) into (plus X -1) for complex or non-integral modes.
30024 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
30025 if they have any depend clauses.
30027 2017-04-11 Martin Liska <mliska@suse.cz>
30030 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
30031 * ipa-split.c (split_function): Create a local comdat symbol
30032 if caller is in a comdat group.
30034 2017-04-11 Martin Liska <mliska@suse.cz>
30037 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
30040 2017-04-11 Martin Sebor <msebor@redhat.com>
30042 PR middle-end/80364
30043 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
30044 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
30046 (directive::set_width, directive::set_precision, format_character):
30048 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
30051 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
30054 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
30055 conflict, set target->arch_name instead of target->cpu_name.
30057 2017-04-11 Richard Biener <rguenther@suse.de>
30059 PR tree-optimization/80374
30060 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
30061 build_zero_cst, remove fold_convertible_p check again.
30063 2017-04-11 Martin Liska <mliska@suse.cz>
30066 * ubsan.c (instrument_object_size): Do not instrument register
30069 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30072 * config/i386/i386-builtin-types.def
30073 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
30074 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
30075 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
30076 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
30077 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
30078 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
30079 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
30080 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
30081 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
30082 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
30083 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
30084 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
30085 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
30086 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
30087 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
30088 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
30089 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
30090 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
30091 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
30092 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
30093 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
30094 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
30095 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
30096 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
30097 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
30098 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
30099 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
30100 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
30101 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
30102 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
30103 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
30104 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
30105 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
30106 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
30107 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
30108 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
30109 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
30110 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
30111 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
30112 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
30113 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
30114 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
30115 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
30116 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
30117 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
30118 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
30120 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
30121 flag to second_arg_count, handle 4 argument function type _COUNT
30122 aliases, handle second_arg_count on second argument rather than last.
30124 2017-04-10 Jeff Law <law@redhat.com>
30126 PR tree-optimization/80374
30127 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
30128 record anything if we can not convert integer_zero_node to the
30131 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30134 * config/rs6000/rs6000.c (rs6000_option_override_internal):
30135 Enhance special handling given to the TARGET_P9_MINMAX option in
30136 relation to certain other options.
30138 2017-04-10 Bin Cheng <bin.cheng@arm.com>
30140 PR tree-optimization/80153
30141 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
30142 remove POINTER_PLUS_EXPR's base part directly, rather than through
30145 2017-04-10 Richard Biener <rguenther@suse.de>
30146 Bin Cheng <bin.cheng@arm.com>
30148 PR tree-optimization/80153
30149 * tree-affine.c (aff_combination_to_tree): Get base pointer from
30150 the first element of pointer type aff_tree. Build result expr in
30152 (add_elt_to_tree): Convert to type unconditionally. Remove other
30153 fold_convert calls.
30154 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
30155 (rewrite_use_nonlinear_expr): Check invariant using iv information.
30157 2017-04-10 Richard Biener <rguenther@suse.de>
30159 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30162 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
30164 PR rtl-optimization/70478
30165 * lra-constraints.c (curr_small_class_check): New.
30166 (update_and_check_small_class_inputs): New.
30167 (process_alt_operands): Update curr_small_class_check. Disfavor
30168 alternative insn memory operands. Check available regs for small
30171 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30174 * config/mips/mips.opt (-mvirt): Update description.
30175 * doc/invoke.texi (-mvirt): Likewise.
30177 2017-04-10 Richard Biener <rguenther@suse.de>
30179 PR middle-end/80362
30180 * fold-const.c (fold_binary_loc): Look at unstripped ops when
30181 looking for NEGATE_EXPR in -A / -B to A / B folding.
30183 2017-04-10 Martin Liska <mliska@suse.cz>
30185 PR gcov-profile/80224
30186 * gcov.c (print_usage): Fix usage string.
30187 (get_gcov_intermediate_filename): Remove.
30188 (output_gcov_file): Use both for normal and intermediate format.
30189 (generate_results): Do not initialize special file for
30190 intermediate format.
30192 2017-04-10 Richard Biener <rguenther@suse.de>
30194 PR tree-optimization/80304
30195 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
30198 2017-04-10 Nathan Sidwell <nathan@acm.org>
30201 * config/rs6000/rs6000.c (rs6000_vector_type): New.
30202 (rs6000_init_builtins): Use it.
30204 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30206 * config/arm/arm.md (<mrc>): Add mode to SET source.
30207 (<mrrc>): Likewise.
30209 2017-04-10 Richard Biener <rguenther@suse.de>
30211 PR middle-end/80344
30212 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
30214 2017-04-10 Jakub Jelinek <jakub@redhat.com>
30217 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
30218 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
30219 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
30220 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
30221 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
30222 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
30223 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
30224 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
30225 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
30226 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
30227 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
30228 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
30229 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
30230 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
30231 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
30232 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
30233 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
30234 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
30235 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
30236 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
30237 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
30238 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
30239 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
30241 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
30243 PR rtl-optimization/70478
30244 * lra-constraints.c: Reverse the last patch.
30246 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
30248 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
30249 Add comment for WCHAR_T.
30251 2017-04-08 Martin Liska <mliska@suse.cz>
30254 2017-04-07 Martin Liska <mliska@suse.cz>
30257 * ipa-split.c (split_function): Add function part to a same comdat
30260 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30263 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
30265 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
30267 * rs6000/rs6000.c (vec_load_pendulum): Rename...
30268 (vec_pairing): ...to this.
30269 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
30270 (rs6000_sched_init): Adjust for name change.
30271 (struct rs6000_sched_context): Likewise.
30272 (rs6000_init_sched_context): Likewise.
30273 (rs6000_set_sched_context): Likewise.
30275 2017-04-07 Jakub Jelinek <jakub@redhat.com>
30281 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
30283 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
30284 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
30285 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
30287 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
30289 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
30291 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30293 PR rtl-optimization/70703
30294 * ira-color.c (update_conflict_hard_regno_costs): Use
30295 int64_t instead of HOST_WIDE_INT.
30297 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30299 PR rtl-optimization/70478
30300 * lra-constraints.c (process_alt_operands): Disfavor alternative
30301 insn memory operands.
30303 2017-04-07 Jeff Law <law@redhat.com>
30305 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
30306 CALL and NOTE_INSN_CALL_ARG_LOCATION.
30308 2017-04-07 Martin Liska <mliska@suse.cz>
30311 * config/aarch64/aarch64.c (aarch64_process_target_attr):
30312 Show error message instead of an ICE.
30314 2017-04-07 Martin Liska <mliska@suse.cz>
30317 * ipa-split.c (split_function): Add function part to a same comdat
30320 2017-04-07 Richard Biener <rguenther@suse.de>
30322 PR middle-end/80341
30323 * tree.c (get_unwidened): Also handle ! for_type case for
30325 * convert.c (do_narrow): Split out from ...
30326 (convert_to_integer_1): ... here. Do not pass final truncation
30327 type to get_unwidened for TRUNC_DIV_EXPR.
30329 2017-04-07 Richard Biener <rguenther@suse.de>
30331 * tree-affine.c (wide_int_ext_for_comb): Take type rather
30333 (aff_combination_const): Adjust.
30334 (aff_combination_scale): Likewise.
30335 (aff_combination_add_elt): Likewise.
30336 (aff_combination_add_cst): Likewise.
30337 (aff_combination_convert): Likewise.
30338 (add_elt_to_tree): Likewise. Remove unused argument.
30339 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
30341 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
30343 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
30345 * config/arm/arm.c (arm_default_short_enums): Use
30346 ARM_DEFAULT_SHORT_ENUMS.
30347 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
30349 2017-04-06 Jakub Jelinek <jakub@redhat.com>
30352 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
30353 members with redundant out-of-class redeclaration.
30355 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30358 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
30359 * config/i386/i386.md (*zero_extendsidi2):
30360 Add (?*x,*x) and (?*v,*v) alternatives.
30362 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30365 * config/i386/i386.c (ix86_expand_builtin)
30366 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
30367 mode from insn data. Convert operands to insn operand mode.
30368 Copy operands that don't satisfy insn predicate to a register.
30370 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
30372 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
30375 2017-04-06 Richard Biener <rguenther@suse.de>
30377 PR tree-optimization/80334
30378 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
30379 preserve alignment of accesses.
30381 2017-04-06 Richard Biener <rguenther@suse.de>
30383 PR tree-optimization/80262
30384 * tree-sra.c (build_ref_for_offset): Preserve address-space
30386 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
30387 Drop useless address-space information on MEM_REF offsets.
30389 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
30391 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
30393 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30395 PR rtl-optimization/70703
30396 * ira-color.c (update_conflict_hard_regno_costs): Use
30397 HOST_WIDE_INT instead of long.
30399 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
30402 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
30403 not defined for x86_64 target. Add -mmmx target option when __SSE2__
30405 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
30406 for x86_64 target. Handle -m3dnowa option.
30408 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30410 PR rtl-optimization/70703
30411 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
30412 (update_conflict_hard_regno_costs): Use long instead of unsigned
30413 arithmetic for cost calculation.
30415 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30416 Bernd Edlinger <bernd.edlinger@hotmail.de>
30419 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
30422 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
30425 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
30426 ptr_mode with Pmode throughout.
30427 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
30428 into probe_stack_range and use DImode.
30430 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
30433 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
30434 call_eh_return is true.
30436 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30438 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
30439 Initialize last_match_fntype_index.
30441 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30444 * tree-nvr.c: Include internal-fn.h.
30445 (pass_return_slot::execute): Ignore internal calls without
30448 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30449 Richard Biener <rguenther@suse.de>
30452 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
30453 captures used multiple times, except for the last use.
30454 * generic-match-head.c: Include gimplify.h.
30456 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30458 PR tree-optimization/79390
30459 * target.h (struct noce_if_info): Declare.
30460 * targhooks.h (default_noce_conversion_profitable_p): Declare.
30461 * target.def (noce_conversion_profitable_p): New target hook.
30462 * ifcvt.h (struct noce_if_info): New type, moved from ...
30463 * ifcvt.c (struct noce_if_info): ... here.
30464 (noce_conversion_profitable_p): Renamed to ...
30465 (default_noce_conversion_profitable_p): ... this. No longer
30467 (noce_try_store_flag_constants, noce_try_addcc,
30468 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
30469 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
30470 instead of noce_conversion_profitable_p.
30471 * config/i386/i386.c: Include ifcvt.h.
30472 (ix86_option_override_internal): Don't override
30473 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
30474 (ix86_noce_conversion_profitable_p): New function.
30475 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
30476 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
30477 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
30478 * doc/tm.texi: Regenerated.
30480 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30482 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
30485 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30488 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
30489 instructions for small multiply cores.
30491 2017-04-04 Jeff Law <law@redhat.com>
30493 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
30495 (mips_expand_vec_perm_const): Initialize elements in orig_perm
30496 that are not set by the loop over the elements.
30498 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30501 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
30502 int mode, convert_modes it to mode as unsigned, otherwise use
30503 lowpart_subreg to mode rather than SImode.
30504 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
30505 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
30506 Use DImode instead of SImode for the shift count operand.
30507 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
30510 2017-04-04 Richard Biener <rguenther@suse.de>
30512 PR middle-end/80281
30513 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
30514 arithmetic done for the negate or the plus. Simplify.
30515 (A - (-B) -> A + B): Likewise.
30516 * fold-const.c (split_tree): Make sure to not negate pointers.
30518 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
30520 PR rtl-optimization/60818
30521 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
30522 a compare of comparisons with the thing compared if this results
30523 in a different machine mode.
30525 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
30527 * alias.c (base_alias_check): Fix typo in comment.
30528 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
30529 * cgraphunit.c (symbol_table::compile): Likewise.
30530 * collect2.c (maybe_run_lto_and_relink): Likewise.
30531 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
30532 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
30533 * config/avr/avr.c (avr_map_op_t): Likewise.
30534 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
30535 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
30536 * config/epiphany/epiphany.md (movcc): Likewise.
30537 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
30538 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
30540 * config/mips/mips.c (mips_save_restore_reg): Likewise.
30541 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
30542 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
30543 * config/sh/sh.c (sh_rtx_costs): Likewise.
30544 * fold-const.c (fold_truth_andor): Likewise.
30545 * genautomata.c (collapse_flag): Likewise.
30546 * gengtype.h (struct type::u::s): Likewise.
30547 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
30548 * input.c (FORMAT_AMOUNT): Likewise.
30549 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
30550 (known_aggs_to_agg_replacement_list): Likewise.
30551 * ipa-inline-analysis.c: Likewise.
30552 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
30553 * ipa-polymorphic-call.c
30554 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
30555 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
30556 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
30558 * modulo-sched.c (apply_reg_moves): Likewise.
30559 * omp-expand.c (build_omp_regions_1): Likewise.
30560 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
30561 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
30562 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
30563 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
30564 * value-prof.c: Likewise.
30565 * var-tracking.c (val_reset): Likewise.
30567 2017-04-03 Richard Biener <rguenther@suse.de>
30569 PR tree-optimization/80275
30570 * fold-const.c (split_address_to_core_and_offset): Handle
30573 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
30575 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
30576 descriptors is at least equal to that of functions.
30578 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30580 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
30582 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30585 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
30586 (mov<IMOD4:mode>): New expander.
30587 (*mov<IMOD4:mode>_internal): New insn and split pattern.
30589 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
30591 PR rtl-optimization/79405
30592 * fwprop.c (propagations_left): New variable.
30593 (forward_propagate_into): Decrement it.
30594 (fwprop_init): Initialize it.
30595 (fw_prop): If the variable has reached zero, stop propagating.
30596 (fwprop_addr): Ditto.
30598 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30601 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
30602 a FUNCTION_DECL, pass it as decl instead of origin to
30605 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
30607 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
30610 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
30613 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
30614 TARGET_VSX_SMALL_INTEGER.
30616 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30618 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30619 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
30621 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30623 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
30624 extraction from odd-numbered MSA register.
30626 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30628 PR middle-end/80173
30629 * expmed.c (store_bit_field_1): Don't attempt to create
30630 a word subreg out of hard registers wider than word if they
30631 have HARD_REGNO_NREGS of 1 for their mode.
30633 PR middle-end/80163
30634 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
30635 conversions to integer types wider than word and pointer.
30638 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
30639 (rtx_equal_for_cselib_p): Pass 0 to it.
30640 * cselib.c (cselib_hasher::equal): Likewise.
30641 (rtx_equal_for_cselib_1): Add depth argument. If depth
30642 is 128, don't look up VALUE locs and punt. Increment
30643 depth in recursive calls when walking VALUE locs.
30645 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
30647 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
30648 (make_gcov_file_name): Use the canonical path name for generating
30650 (read_line): Fix handling of files with ascii null bytes.
30652 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
30654 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
30655 to initialise a vector register instead
30656 of using a const_int.
30658 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30660 PR translation/80189
30661 * gimplify.c (omp_default_clause): Use %qs instead of %s in
30662 diagnostic messages.
30664 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
30667 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
30668 (dfp_diex_<mode>): Update mode of operand 1.
30669 * doc/extend.texi (dxex, dxexq): Document change to return type.
30670 (diex, diexq): Document change to argument type.
30672 2017-03-30 Martin Jambor <mjambor@suse.cz>
30675 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
30676 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
30677 it reflects the signature changes performed at the callee side.
30678 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
30679 to cgraph_build_function_type_skip_args.
30680 (build_function_decl_skip_args): Adjust call to the above function.
30682 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30685 * config/i386/sse.md
30686 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
30687 register as dest whenever it is a MEM not rtx_equal_p to the
30688 corresponding dup operand, and when forcing into reg move the
30689 reg into the memory afterwards.
30690 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
30691 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
30692 for the force_reg mode.
30693 (avx512vl_vextractf128<mode>): Use register as dest either
30694 always when a MEM, or when it is a MEM not rtx_equal_p to the
30695 corresponding dup operand, or even not when it is a CONST_VECTOR
30696 depending on the mode and lo vs. hi.
30697 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
30699 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
30700 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
30701 Likewise. Require that operands[2] is even.
30702 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
30703 Remove extraneous parens. Require that operands[2] is a multiple
30705 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
30706 operands[0] is a MEM if <mask_applied>, the predicates/constraints
30707 disallow memory then.
30709 2017-03-30 Richard Biener <rguenther@suse.de>
30711 PR tree-optimization/77498
30712 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
30713 to non-constants over backedges.
30715 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
30717 PR rtl-optimization/80233
30718 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
30719 as last_combined_insn. Do not test for BARRIER_P separately.
30721 2017-03-29 Andreas Schwab <schwab@suse.de>
30724 * calls.c (prepare_call_address): Convert funexp to Pmode before
30725 copying to temp reg.
30727 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30729 PR tree-optimization/80158
30730 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
30731 Handle possible future case of more than one alternate
30733 (replace_rhs_if_not_dup): Likewise.
30734 (replace_one_candidate): Likewise.
30736 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
30738 PR rtl-optimization/80193
30739 * ira.c (ira): Do not check allocation for LRA.
30741 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
30743 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
30744 (nvptx_output_simt_exit): Declare.
30745 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
30746 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
30747 (init_softstack_frame): Move initialization of crtl->is_leaf to...
30748 (nvptx_declare_function_name): ...here. Emit declaration of local
30749 memory space buffer for omp_simt_enter insn.
30750 (nvptx_output_unisimt_switch): New.
30751 (nvptx_output_softstack_switch): New.
30752 (nvptx_output_simt_enter): New.
30753 (nvptx_output_simt_exit): New.
30754 * config/nvptx/nvptx.h (struct machine_function): New fields
30755 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
30756 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
30757 (UNSPECV_SIMT_EXIT): Ditto.
30758 (omp_simt_enter_insn): New insn.
30759 (omp_simt_enter): New expansion.
30760 (omp_simt_exit): New insn.
30761 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
30763 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
30764 (expand_GOMP_SIMT_ENTER_ALLOC): New.
30765 (expand_GOMP_SIMT_EXIT): New.
30766 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
30767 (GOMP_SIMT_ENTER_ALLOC): Ditto.
30768 (GOMP_SIMT_EXIT): Ditto.
30769 * target-insns.def (omp_simt_enter): New insn.
30770 (omp_simt_exit): Ditto.
30771 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
30773 (lower_rec_simd_input_clauses): Implement SIMT privatization.
30774 (lower_rec_input_clauses): Likewise.
30775 (lower_lastprivate_clauses): Handle SIMT privatization.
30777 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
30778 (ompdevlow_adjust_simt_enter): New.
30779 (find_simtpriv_var_op): New.
30780 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
30781 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
30783 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
30784 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
30785 (copy_decl_for_dup_finish): Ditto.
30787 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
30789 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
30792 * config/i386/i386.c (ix86_option_override_internal): Always
30793 allow -mpreferred-stack-boundary=3 for 64-bit targets.
30795 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30797 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
30799 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30801 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
30802 mark new edge's irreducible flag accordign to it.
30803 (vect_do_peeling): Check loop preheader edge's irreducible flag
30804 and pass it to function slpeel_add_loop_guard.
30806 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
30808 PR tree-optimization/80218
30809 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30810 Update block frequencies and counts.
30812 2017-03-28 Richard Biener <rguenther@suse.de>
30814 PR tree-optimization/78644
30815 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
30816 of a simplification result we may not use it at all.
30818 2017-03-28 Richard Biener <rguenther@suse.de>
30821 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
30822 without arguments, generate default definition of a SSA name.
30824 2017-03-28 Richard Biener <rguenther@suse.de>
30826 PR middle-end/80222
30827 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
30828 TYPE_REF_CAN_ALIAS_ALL references.
30829 * fold-const.c (fold_indirect_ref_1): Likewise.
30831 2017-03-28 Martin Liska <mliska@suse.cz>
30834 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
30835 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
30837 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30838 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
30840 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
30841 (EXTRA_SPECS): Define.
30842 (SUBTARGET_EXTRA_SPECS): Likewise.
30843 (SUBTARGET_CPP_SPEC): Likewise.
30844 * config/arc/elf.h (EXTRA_SPECS): Renamed to
30845 SUBTARGET_EXTRA_SPECS.
30846 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
30848 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30850 * config/arc/simdext.md (vst64_insn): Update pattern.
30851 (vld32wh_insn): Likewise.
30852 (vld32wl_insn): Likewise.
30853 (vld64_insn): Likewise.
30854 (vld32_insn): Likewise.
30856 2017-03-28 Marek Polacek <polacek@redhat.com>
30859 * fold-const.c (fold_comparison): Use protected_set_expr_location
30860 instead of SET_EXPR_LOCATION.
30862 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
30864 * tree.c (add_expr): Avoid name lookup warning.
30866 2017-03-27 Jeff Law <law@redhat.com>
30868 PR tree-optimization/80216
30869 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
30870 function name. Limit recursion depth.
30871 (record_temporary_equivalences): Corresponding changes.
30873 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
30875 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
30878 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30881 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
30883 * cfgcleanup.c (reg_note_cfa_p): New array.
30884 (insns_have_identical_cfa_notes): New function.
30885 (old_insns_match_p): Don't cross-jump in between /f
30886 and non-/f instructions. If both i1 and i2 are frame related,
30887 verify all CFA notes, their order and content.
30889 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
30892 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
30893 HImode and SImode with zero extend to DImode to one insn.
30894 (bswap<mode>2_extenddi): Likewise.
30895 (bswapsi2_extenddi): Likewise.
30896 (bswaphi2_extendsi): Likewise.
30897 (bswaphi2): Combine bswap HImode and SImode into one insn.
30898 Separate memory insns from swapping register.
30899 (bswapsi2): Likewise.
30900 (bswap<mode>2): Likewise.
30901 (bswaphi2_internal): Delete, no longer used.
30902 (bswapsi2_internal): Likewise.
30903 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
30904 store, and gpr<-gpr swap insns.
30905 (bswap<mode>2_store): Likewise.
30906 (bswaphi2_reg): Register only splitter, combine with the splitter.
30907 (bswaphi2 splitter): Likewise.
30908 (bswapsi2_reg): Likewise.
30909 (bswapsi2 splitter): Likewise.
30910 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
30911 the insns into load, store, and register/register insns.
30912 (bswapdi2_ldbrx): Likewise.
30913 (bswapdi2_load): Likewise.
30914 (bswapdi2_store): Likewise.
30915 (bswapdi2_reg): Likewise.
30917 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
30919 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
30920 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
30922 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
30925 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
30927 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30928 special handling for target option conflicts between dform
30929 options (-mpower9-dform, -mpower9-dform-vector,
30930 -mpower9-dform-scalar) and -mno-direct-move.
30932 2017-03-27 Richard Biener <rguenther@suse.de>
30934 PR tree-optimization/80181
30935 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
30937 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30939 * config/arc/predicates.md (move_double_src_operand): Replace the
30940 call to move_double_src_operand with a call to address_operand.
30942 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30944 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
30945 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
30946 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
30948 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30950 * config/arc/predicates.md (long_immediate_loadstore_operand):
30951 Consider scaled addresses cases.
30953 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30955 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
30956 restored when in interrupt.
30957 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
30958 doesn't have delay slot.
30960 2017-03-27 Richard Biener <rguenther@suse.de>
30963 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
30964 inlined thunk clones.
30966 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30969 * asan.c (instrument_derefs): Copy over last operand from
30970 original COMPONENT_REF to the new COMPONENT_REF with
30971 DECL_BIT_FIELD_REPRESENTATIVE.
30972 * ubsan.c (instrument_object_size): Likewise.
30974 2017-03-27 Richard Biener <rguenther@suse.de>
30976 PR tree-optimization/80170
30977 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
30978 sure DR/SCEV didnt fold in constants we do not see when looking
30979 at the reference base alignment.
30981 2017-03-27 Richard Biener <rguenther@suse.de>
30983 PR middle-end/80171
30984 * gimple-fold.c (fold_ctor_reference): Properly guard against
30985 NULL return value from canonicalize_constructor_val.
30987 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
30990 * config/i386/i386.c (ix86_expand_builtin)
30991 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
30992 flags reg setting and flags reg using instructions.
30993 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
30994 clobbering instructions to zero extend op2.
30996 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
30998 * doc/install.texi (Configuration) <--with-aix-soname>:
30999 Update link to AIX ld.
31001 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31003 PR rtl-optimization/80160
31004 PR rtl-optimization/80159
31005 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31006 reg_alternate_class into account.
31008 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31011 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31012 to consider in curr_insn_transform.
31014 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31016 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31017 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31018 and emit_mode_inner.
31020 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31022 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
31023 argument to the overloaded builtin variants. Use the new flag to
31024 deprecate certain builtin variants.
31025 * config/s390/s390-builtin-types.def: Add new builtin types.
31026 * config/s390/s390-builtins.h: Support new flags field for
31027 overloaded builtins.
31028 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
31029 (s390_macro_to_expand): Enable vector float data type.
31030 (s390_cpu_cpp_builtins_internal): Indicate support of the new
31031 builtins by incrementing the __VEC__ version number.
31032 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
31034 (s390_resolve_overloaded_builtin): Emit error messages depending
31035 on the builtin flags.
31036 * config/s390/s390.c (s390_expand_builtin): Support additional
31037 flags argument. Change error message to match the messages
31038 emitted in s390-c.c.
31039 * config/s390/s390.md: New UNSPEC_* constants.
31040 (op_type): Add new instruction types.
31041 * config/s390/vecintrin.h: Add new builtins and test data class
31043 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
31044 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
31045 (VEC_INEXACT, VEC_NOINEXACT): New constants.
31046 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
31047 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
31048 ("vec_mergel<mode>"): V_HW -> VEC_HW.
31050 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
31051 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
31052 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
31053 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
31055 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
31056 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
31057 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
31058 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
31060 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
31061 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
31062 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
31063 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
31064 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
31065 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
31066 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
31068 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
31069 ("vec_scatter_element<V_HW_4:mode>_DI")
31070 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
31071 ("vec_fpint<mode>", "vflls")
31072 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
31073 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
31074 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
31075 ("*vec_cmphe<mode>_cc"): ... these.
31077 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
31078 mode constant instead of magic value.
31080 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31082 * config/s390/s390.c (s390_expand_vec_compare): Support other
31083 vector floating point modes than just V2DF.
31084 (s390_expand_vcond): Likewise.
31085 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
31086 (s390_cannot_change_mode_class): Prevent mode changes between TF
31087 and V1TF in vector registers.
31088 * config/s390/s390.md (DF, SF): New mode attributes.
31089 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
31090 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
31091 SFmode support for VRs.
31092 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
31094 (VFT, VF_HW): New mode iterators.
31095 (vw, sdx): New mode attributes.
31096 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
31097 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
31098 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
31099 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
31100 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
31101 also the new vector floating point modes. Renaming to ...
31103 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
31104 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
31105 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
31106 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
31107 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
31108 ("vec_unordered<mode>"): ... these.
31110 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
31111 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
31112 ("*vec_extendv2df"): New insn definitions.
31114 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31116 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
31117 ("mulditi3_2", "*muldi3_sign"): New patterns.
31118 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
31119 rename the pattern definition.
31121 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31123 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
31125 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
31127 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31129 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
31130 instruction if possible.
31131 * config/s390/vector.md (vec_halfnumelts): New mode
31133 ("*vec_vllezlf<mode>"): New pattern.
31135 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31137 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
31138 ("popcountv4si2", "popcountv2di2"): Rename to ...
31139 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
31140 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
31142 ("popcount<mode>2_vxe"): New pattern.
31144 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31146 * common/config/s390/s390-common.c (processor_flags_table): Add
31148 * config.gcc: Add arch12.
31149 * config/s390/driver-native.c (s390_host_detect_local_cpu):
31150 Default to arch12 for unknown CPU model numbers.
31151 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
31152 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
31153 PROCESSOR_max sanity check.
31154 * config/s390/s390-opts.h (enum processor_type): Add
31156 * config/s390/s390.c (processor_table): Add arch12.
31157 (s390_expand_builtin): Add check for B_VXE flag.
31158 (s390_issue_rate): Add PROCESSOR_ARCH12.
31159 (s390_get_sched_attrmask): Likewise.
31160 (s390_get_unit_mask): Likewise.
31161 (s390_sched_score): Enable z13 scheduling for arch12.
31162 (s390_sched_reorder): Likewise.
31163 (s390_sched_variable_issue): Likewise.
31164 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
31166 (s390_tune_attr): Use z13 scheduling also for arch12.
31167 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
31168 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
31169 (TARGET_VXE_P): New macros.
31170 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
31171 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
31172 * config/s390/s390.opt: Add arch12 as processor_type.
31174 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31176 * config/s390/s390.md
31177 ("fixuns_truncdddi2", "fixuns_trunctddi2")
31178 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
31179 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
31181 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
31182 Rename expanders to ...
31184 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
31185 ("fixuns_truncdddi2_emu"): ... these.
31187 ("fixuns_trunc<mode>si2_emu"): New expander.
31189 ("*fixuns_truncdfdi2_z13"): Rename to ...
31190 ("*fixuns_truncdfdi2_vx"): ... this.
31192 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31194 * config/s390/2964.md: Remove the single element vector compare
31195 instructions which are no longer used.
31196 * config/s390/s390.c (s390_select_ccmode): Remove handling of
31198 (s390_canonicalize_comparison): Remove handling of DFmode
31200 (s390_expand_vec_compare_scalar): Remove function.
31201 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
31202 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
31204 ("*cmp<mode>_ccs"): Add wfcdb instruction.
31206 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31208 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
31210 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
31211 will anyway by matched by mov<mode>_64dfp.
31213 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31215 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
31216 vlef/vstef. Add missing operand to vleif.
31218 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31220 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
31221 pair for all vector types with 64 bit elements.
31222 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
31223 * config/s390/vector.md (V_HW_64): ... here.
31224 (V_128_NOSINGLE): New mode iterator.
31225 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
31226 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
31227 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
31228 ("*vec_load_pairv2di"): Change to ...
31229 ("*vec_load_pair<mode>"): ... this one.
31231 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31233 * config/s390/constraints.md: Add comments.
31234 (jKK): Reject element sizes > 8 bytes.
31235 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
31237 * config/s390/s390.md: Add the s_operand checks formerly in
31238 s390_split_ok_p to various splitters where they are still
31240 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
31241 for 128 bit vectors. Plus two splitters.
31243 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31245 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
31248 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31251 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
31252 error if the boundary argument is not constant.
31254 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31256 PR rtl-optimization/80112
31257 * loop-doloop.c (doloop_condition_get): Don't check condition
31258 if cmp isn't SET with IF_THEN_ELSE src.
31260 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31262 PR tree-optimization/80158
31263 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
31264 replacing a candidate statement, also replace it for the
31265 candidate's alternate interpretation.
31266 (replace_rhs_if_not_dup): Likewise.
31267 (replace_one_candidate): Likewise.
31269 2017-03-24 Richard Biener <rguenther@suse.de>
31271 PR tree-optimization/80167
31272 * graphite-isl-ast-to-gimple.c
31273 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
31275 (translate_isl_ast_to_gimple::get_rename): Likewise.
31277 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31279 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
31280 handling of certain combinations of target options, including the
31281 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
31282 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
31284 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31287 * config/arm/arm.md (*load_multiple): Add reload_completed to
31288 matching condition.
31290 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31291 Richard Biener <rguenth@suse.de>
31293 PR tree-optimization/79908
31294 PR tree-optimization/80136
31295 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31296 been cast away, gimplify_and_add suffices.
31298 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
31300 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
31302 2017-03-23 Richard Biener <rguenther@suse.de>
31304 PR tree-optimization/80032
31305 * gimplify.c (gimple_push_cleanup): Forced unconditional
31306 cleanups still have to go to the conditional_cleanups
31309 2017-03-22 Jakub Jelinek <jakub@redhat.com>
31311 PR tree-optimization/80072
31312 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
31314 (next_operand_entry_id): Change type to unsigned int.
31315 (sort_by_operand_rank): Make sure to return the right return value
31316 even if unsigned fields are bigger than INT_MAX.
31317 (struct oecount): Change cnt and id type to unsigned int.
31318 (oecount_hasher::equal): Formatting fix.
31319 (oecount_cmp): Make sure to return the right return value
31320 even if unsigned fields are bigger than INT_MAX.
31321 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
31324 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
31325 TREE_READONLY on result if writing it more than once.
31328 * doc/invoke.texi (-fsanitize=thread): Document that with
31329 -fnon-call-exceptions atomics are not able to throw
31333 * tsan.c: Include tree-eh.h.
31334 (instrument_builtin_call): Call maybe_clean_eh_stmt or
31335 maybe_clean_or_replace_eh_stmt where needed.
31336 (instrument_memory_accesses): Add cfg_changed argument.
31337 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
31338 if it returned true.
31339 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
31341 PR rtl-optimization/63191
31342 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
31343 wrapper function, moved the whole old content into ...
31344 (ix86_delegitimize_address_1): ... this. New inline function.
31345 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
31346 true as last argument instead of ix86_delegitimize_address.
31348 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31350 * config/aarch64/aarch64.c (generic_branch_cost): Copy
31351 cortexa57_branch_cost.
31353 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31355 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
31357 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31360 * doc/md.texi (Constraints): Document wA constraint.
31361 * config/rs6000/constraints.md (wA): New.
31362 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
31363 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
31364 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
31365 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
31367 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
31370 * gimplify.c (is_oacc_declared): New function.
31371 (oacc_default_clause): Use it to set default flags for acc declared
31372 variables inside parallel regions.
31373 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
31374 declared variables.
31375 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
31376 declare attribute to any decl as necessary.
31378 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
31381 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
31382 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
31383 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
31384 (arm_arch_lpae): This.
31385 * config/arm/arm.c (arm_arch7ve): Rename into ...
31386 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
31387 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
31390 2017-03-22 Martin Liska <mliska@suse.cz>
31393 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
31394 error message instead of an ICE.
31396 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31398 * doc/extend.texi (6.11 Additional Floating Types): Revise.
31400 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
31402 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31404 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31407 2017-03-21 Martin Sebor <msebor@redhat.com>
31409 * doc/extend.texi: Use "cannot" instead of "can't."
31410 * doc/hostconfig.texi: Same.
31411 * doc/install.texi: Same.
31412 * doc/invoke.texi: Same.
31413 * doc/loop.texi: Same.
31414 * doc/md.texi: Same.
31415 * doc/objc.texi: Same.
31416 * doc/rtl.texi: Same.
31417 * doc/tm.texi: Same.
31418 * doc/tm.texi.in: Same.
31419 * doc/trouble.texi: Same.
31421 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
31424 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
31425 (collect_checksum_attributes): Set it.
31426 (die_checksum_ordered): Use it.
31428 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31430 PR tree-optimization/79908
31431 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
31432 change: For a VA_ARG whose LHS has been cast away, use
31433 force_gimple_operand to construct the side effects.
31435 2017-03-21 David Malcolm <dmalcolm@redhat.com>
31437 PR translation/80001
31438 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
31439 more amenable to translation.
31440 (oacc_loop_auto_partitions): Likewise.
31442 2017-03-21 Marek Polacek <polacek@redhat.com>
31443 Martin Sebor <msebor@redhat.com>
31445 PR tree-optimization/80109
31446 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
31447 on INTEGRAL_TYPE_P.
31449 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31450 Segher Boessenkool <segher@kernel.crashing.org>
31453 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
31454 check reg_used_between_p between insn and one of succ or succ2
31455 depending on if succ is artificial insn not inserted into insn
31458 2017-03-21 Martin Liska <mliska@suse.cz>
31460 PR gcov-profile/80081
31461 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
31462 * doc/gcc.texi: Include gcov-dump stuff.
31463 * doc/gcov-dump.texi: New file.
31465 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
31467 PR rtl-optimization/79150
31468 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
31469 conditional jump, if the jump is the last insn of the loop.
31471 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31472 Richard Biener <rguenth@suse.de>
31474 PR tree-optimization/79908
31475 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31476 been cast away, use force_gimple_operand to construct the side
31479 2017-03-21 Martin Liska <mliska@suse.cz>
31481 PR libfortran/79956
31482 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
31485 2017-03-21 Brad Spengler <spender@grsecurity.net>
31488 * plugin.c (htab_hash_plugin): New function.
31489 (add_new_plugin): Use it and adjust.
31490 (parse_plugin_arg_opt): Adjust.
31491 (init_one_plugin): Likewise.
31493 2017-03-21 Richard Biener <rguenther@suse.de>
31495 PR tree-optimization/80032
31496 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
31497 if set force the cleanup to happen unconditionally.
31498 (gimplify_target_expr): Push inserted clobbers with force_uncond
31499 to avoid them being removed by control-dependent DCE.
31501 2017-03-21 Richard Biener <rguenther@suse.de>
31503 PR tree-optimization/80122
31504 * tree-inline.c (copy_bb): Do not expans va-arg packs or
31505 va_arg_pack_len when the inlined call stmt requires pack
31507 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
31509 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31512 * tsan.c (instrument_builtin_call): If the memory model argument
31513 is not a constant, assume it is valid.
31516 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
31519 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31521 PR rtl-optimization/79910
31522 * combine.c (can_combine_p): Do not allow combining an I0 or I1
31523 if its dest is used by an insn before I2 (other than the combined
31524 insns themselves, which are properly handled already).
31526 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31529 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31531 * combine.c (record_used_regs): New static function.
31532 (try_combine): Handle situations where there is an additional
31533 instruction between I2 and I3 which needs to have a LOG_LINK
31537 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31539 * combine.c (try_combine): Delete redundant i1 test. Call
31540 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31542 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31545 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
31546 alternatives 13/14.
31548 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31550 PR tree-optimization/80054
31551 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
31552 the optimization if a PHI or any of its arguments is not dominated
31553 by the candidate's basis. Use gphi* rather than gimple* as
31555 (replace_profitable_candidates): Clean up a gimple* variable that
31556 should be a gphi* variable.
31558 2017-03-20 Martin Sebor <msebor@redhat.com>
31561 * doc/extend.texi (attribute constructor): Document present limitation.
31563 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
31566 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
31567 __POWER9_VECTOR__ #ifdef control, change template definition to
31568 use Power9-specific built-in function.
31569 (vec_any_eq): Likewise.
31570 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
31571 to control outcomes from this test.
31572 (vector_ae_<mode>p): For VEC_F modes, likewise.
31574 2017-03-20 Ian Lance Taylor <iant@google.com>
31576 * config/i386/i386.c (ix86_function_regparm): Save an extra
31577 register for -fsplit-stack with DECL_STATIC_CHAIN.
31579 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31582 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
31583 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
31585 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31587 * config/riscv/riscv.c (riscv_print_operand): Use "fence
31589 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
31592 2017-03-20 Marek Polacek <polacek@redhat.com>
31595 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
31597 2017-03-20 Richard Biener <rguenther@suse.de>
31599 PR tree-optimization/80113
31600 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
31601 allocate extra SSA name for PHI def.
31602 (add_close_phis_to_outer_loops): Likewise.
31603 (add_close_phis_to_merge_points): Likewise.
31604 (copy_loop_close_phi_args): Likewise.
31605 (copy_cond_phi_nodes): Likewise.
31607 2017-03-20 Martin Liska <mliska@suse.cz>
31609 PR middle-end/79753
31610 * tree-chkp.c (chkp_build_returned_bound): Do not build
31611 returned bounds for a LHS that's not a BOUNDED_P type.
31613 2017-03-20 Martin Liska <mliska@suse.cz>
31617 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
31618 COMPLEX_CST and VECTOR_CST.
31620 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31623 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
31624 target operand. A new splitter adds the clobber statement in case
31625 the target operand is dead anyway.
31627 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
31629 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
31630 to age-old versions of binutils and glibc.
31632 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
31634 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
31636 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31638 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
31640 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31642 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
31643 requirement for binutils 2.13.
31645 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31647 * combine.c (try_combine): Delete redundant i1 test. Call
31648 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31650 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
31652 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
31653 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
31655 <riscv64-*-elf>: Re-arrange section
31656 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
31657 <riscv32-*-linux>: Likewise.
31658 <riscv64-*-elf>: Likewise
31659 <riscv64-*-linux>: Likewise.
31661 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
31664 * aarch64.opt(verbose-cost-dump): Fix typo.
31666 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
31669 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
31670 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
31672 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31674 * reload.c (find_reloads): When reloading a nonoffsettable address,
31675 use RELOAD_OTHER for it and its address reloads.
31677 PR rtl-optimization/79910
31678 * combine.c (record_used_regs): New static function.
31679 (try_combine): Handle situations where there is an additional
31680 instruction between I2 and I3 which needs to have a LOG_LINK
31683 2017-03-17 Jeff Law <law@redhat.com>
31685 PR tree-optimization/71437
31686 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
31687 conditional in the hash table first.
31688 (vrp_dom_walker::before_dom_children): Extract condition from
31689 ASSERT_EXPR. Record condition, its inverion and any implied
31690 conditions as well.
31692 2017-03-17 Marek Polacek <polacek@redhat.com>
31693 Markus Trippelsdorf <markus@trippelsdorf.de>
31695 PR tree-optimization/80079
31696 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
31699 2017-03-17 Richard Biener <rguenther@suse.de>
31701 PR middle-end/80075
31702 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
31703 Properly verify the LHS before the RHS possibly claims to be
31705 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
31708 2017-03-17 Martin Jambor <mjambor@suse.cz>
31710 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
31711 (List of -O2 options): Likewise.
31712 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
31715 2017-03-17 Tom de Vries <tom@codesourcery.com>
31717 * gcov-dump.c (print_usage): Print bug_report_url.
31719 2017-03-17 Richard Biener <rguenther@suse.de>
31721 PR middle-end/80050
31722 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
31723 (parser::peek): Likewise.
31725 2017-03-17 Richard Biener <rguenther@suse.de>
31727 PR tree-optimization/80048
31728 * sese.c (free_sese_info): Properly release rename_map and
31729 copied_bb_map elements.
31731 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
31733 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
31734 Add linked-list forward and backlinks. Insert on
31735 construction, remove on destruction.
31736 (class pass_store_merging): Add m_stores_head field.
31737 (pass_store_merging::terminate_and_process_all_chains):
31738 Iterate over m_stores_head list.
31739 (pass_store_merging::terminate_all_aliasing_chains):
31741 (pass_store_merging::execute): Check for debug stmts first.
31742 Push new chains onto the m_stores_head stack.
31744 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
31747 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
31748 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
31749 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
31751 2017-03-16 Jeff Law <law@redhat.com>
31753 PR tree-optimization/71437
31754 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
31755 member function. Implementation moved into after_dom_children
31756 member function and into the threader's thread_outgoing_edges
31758 (dom_opt_dom_walker::after_dom_children): Simplify by moving
31759 some code into new thread_outgoing_edges.
31760 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
31761 definition. Simplify marker handling (do it here). Assume we always
31762 have the available expression and the const/copies tables.
31763 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
31765 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
31766 * tree-vrp.c (equiv_stack): No longer file scoped.
31767 (vrp_dom_walker): New class.
31768 (vrp_dom_walker::before_dom_children): New member function.
31769 (vrp_dom_walker::after_dom_children): Likewise.
31770 (identify_jump_threads): Setup domwalker. Use it rather than
31771 walking edges in a random order by hand. Simplify setup/finalization.
31772 (finalize_jump_threads): Remove.
31773 (vrp_finalize): Do not call identify_jump_threads here.
31774 (execute_vrp): Do it here instead and call thread_through_all_blocks
31777 PR tree-optimization/71437
31778 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
31780 (simplify_stmt_for_jump_threading): Add basic_block argument. All
31782 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
31783 (dom_opt_dom_walker::thread_across_edge): Remove
31784 handle_dominating_asserts argument. All callers changed.
31785 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
31786 changes. Remove calls to lhs_of_dominating_assert. Other
31787 uses of handle_dominating_asserts turn into unconditional code
31788 (simplify_control_stmt_condition_1): Likewise.
31789 (simplify_control_stmt_condition): Likewise.
31790 (thread_through_normal_block, thread_across_edge): Likewise.
31791 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
31792 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
31793 object if it is not an SSA_NAME.
31794 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
31795 before calling into the VRP specific simplifiers.
31796 (identify_jump_threads): Remove handle_dominating_asserts
31799 2017-03-16 Jakub Jelinek <jakub@redhat.com>
31802 * tree-diagnostic.c (default_tree_printer): No longer static.
31803 * tree-diagnostic.h (default_tree_printer): New prototype.
31805 2017-03-16 Tamar Christina <tamar.christina@arm.com>
31807 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
31808 Change ins into fmov.
31810 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31812 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
31813 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
31814 Use h_con constraint for operand 1.
31815 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
31816 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
31818 2017-03-15 Jeff Law <law@redhat.com>
31820 PR tree-optimization/71437
31821 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
31822 (record_temporary_equivalences): Use it.
31824 PR tree-optimization/71437
31825 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
31826 tree-ssa-scopedtables.
31827 (lookup_avail_expr, build_and_record_new_cond): Likewise.
31828 (record_conditions, record_cond, vuse_eq): Likewise.
31829 (record_edge_info): Adjust to API tweak of record_conditions.
31830 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
31831 (record_temporary_equivalences, optimize_stmt): Likewise.
31832 (eliminate_redundant_computations): Likewise.
31833 (record_equivalences_from_stmt): Likewise.
31834 * tree-ssa-scopedtables.c: Include options.h and params.h.
31835 (vuse_eq): New function, moved from tree-ssa-dom.c
31836 (build_and_record_new_cond): Likewise.
31837 (record_conditions): Likewise. Accept vector of conditions rather
31838 than edge_equivalence structure for first argument.
31839 for the first argument.
31840 (avail_exprs_stack::lookup_avail_expr): New member function, moved
31841 from tree-ssa-dom.c.
31842 (avail_exprs_stack::record_cond): Likewise.
31843 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
31844 from tree-ssa-dom.c.
31845 (avail_exprs_stack): Add new member functions lookup_avail_expr
31847 (record_conditions): Declare.
31849 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
31852 * lra-constraints.c (process_alt_operands): Increase reject for
31853 reloading an input/output operand.
31855 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
31858 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
31859 insns to convert from signed/unsigned char/short to IEEE 128-bit
31861 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
31863 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
31866 * config/i386/i386.c (ix86_vector_duplicate_value): Create
31867 subreg of inner mode for values already in registers.
31869 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
31871 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
31872 iteration reg is used after the loop.
31874 2017-03-14 Martin Sebor <msebor@redhat.com>
31876 PR tree-optimization/79800
31877 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
31878 precision in negative-positive range.
31879 (format_floating): Call non-const overload with adjusted precision.
31881 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31884 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
31887 2017-03-14 Martin Sebor <msebor@redhat.com>
31889 PR middle-end/80020
31890 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
31891 * builtins.def (aligned_alloc): Use it.
31894 * Makefile.in (GTFILES): Add calls.c.
31895 * calls.c: Include "gt-calls.h".
31897 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
31899 PR rtl-optimization/79728
31900 * regs.h (struct target_regs): New field
31901 x_contains_allocatable_regs_of_mode.
31902 (contains_allocatable_regs_of_mode): New macro.
31903 * reginfo.c (init_reg_sets_1): Initialize it, and change
31904 contains_reg_of_mode so it includes global regs as well.
31905 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
31906 rather than contains_regs_of_mode.
31908 2017-03-14 Martin Liska <mliska@suse.cz>
31910 * doc/invoke.texi: Document options that can't be combined with
31911 -fcheck-pointer-bounds.
31913 2017-03-14 Martin Liska <mliska@suse.cz>
31915 PR middle-end/79831
31916 * doc/invoke.texi (-Wchkp): Document the option.
31918 2017-03-14 Martin Liska <mliska@suse.cz>
31920 * Makefile.in: Install gcov-dump.
31922 2017-03-14 Martin Liska <mliska@suse.cz>
31924 * multiple_target.c (expand_target_clones): Bail out for
31925 an invalid attribute.
31927 2017-03-14 Richard Biener <rguenther@suse.de>
31929 * alias.c (struct alias_set_entry): Pack properly.
31930 * cfgloop.h (struct loop): Likewise.
31931 * cse.c (struct set): Likewise.
31932 * ipa-utils.c (struct searchc_env): Likewise.
31933 * loop-invariant.c (struct invariant): Likewise.
31934 * lra-remat.c (struct cand): Likewise.
31935 * recog.c (struct change_t): Likewise.
31936 * rtl.h (struct address_info): Likewise.
31937 * symbol-summary.h (function_summary): Likewise.
31938 * tree-loop-distribution.c (struct partition): Likewise.
31939 * tree-object-size.c (struct object_size_info): Likewise.
31940 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
31941 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
31942 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
31943 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
31944 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
31945 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
31946 (struct _stmt_vec_info): Likewise.
31948 2017-03-14 Martin Liska <mliska@suse.cz>
31951 * multiple_target.c (create_dispatcher_calls): Check that
31952 a target can create a function dispatcher.
31954 2017-03-14 Martin Liska <mliska@suse.cz>
31957 * multiple_target.c (expand_target_clones): Drop local.local
31958 flag for default implementation.
31960 2017-03-14 Richard Biener <rguenther@suse.de>
31962 PR tree-optimization/80030
31963 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
31965 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
31967 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
31968 gcc_fallthrough() instead of __attribute__((fallthrough));
31970 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31972 * doc/gcc.texi: Remove "up" link to (DIR).
31973 * doc/gccint.texi: Ditto.
31975 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31977 * doc/install.texi (Specific) <avr>: Remove reference to
31980 2017-03-13 Jeff Law <law@redhat.com>
31982 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
31983 attribute rather than comments.
31985 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
31986 match_scratch operand is highest.
31988 2017-03-13 Martin Liska <mliska@suse.cz>
31990 PR middle-end/78339
31991 * ipa-pure-const.c (warn_function_noreturn): If the declarations
31992 is a CHKP clone, use original declaration.
31994 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31996 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
31997 (arc_conditional_register_usage): Use a different allocation order
31998 when optimizing for size.
31999 * common/config/arc/arc-common.c (arc_option_optimization_table):
32000 Section anchors default on when optimizing for size.
32002 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32004 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32006 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32008 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32009 * config/arc/arc.md (cpu_facility): Add cd variant.
32010 (*movqi_insn): Add code density variant.
32011 (*movhi_insn): Likewise.
32012 (*movqi_insn): Likewise.
32013 (*addsi3_mixed): Likewise.
32014 (subsi3_insn): Likewise.
32016 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32018 * config/arc/arc.md (movsi_cond_exec): Update constraint.
32020 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32022 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
32023 expressions with MINUS and UNARY ops.
32025 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32028 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
32030 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
32031 between vec_select and vector argument.
32032 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
32033 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
32034 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
32035 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
32036 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
32037 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
32039 2017-03-13 Richard Biener <rguenther@suse.de>
32042 * params.def (vect-max-peeling-for-alignment): Fix typo.
32044 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32046 * doc/install.texi (Specific) <mips-*-*>: Remove description of
32047 issue that only occurred with binutils below 2.18.
32049 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32051 * doc/install.texi (Specific) <cris-axis-elf>: No longer
32052 refer to binutils 2.11/2.12 minimum.
32054 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32056 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
32057 ftp.kernel.org and simplify binutils requirement.
32059 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
32061 * doc/invoke.texi (Warning Options): Fix spelling of link-time
32063 (Optimize Options): Ditto. Also remove redundancy.
32065 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32067 PR translation/79848
32068 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
32070 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
32071 to G_ to avoid double translation.
32073 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32075 PR translation/79923
32076 * auto-profile.c (get_combined_location): Convert leading
32077 character of diagnostics to lower case and remove trailing period.
32078 (read_profile): Likewise for various diagnostics.
32079 * config/arm/arm.c (arm_option_override): Remove trailing period
32080 from various diagnostics.
32081 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
32082 (msp430_expand_delay_cycles): Likewise.
32084 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32087 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
32088 full command-line argument, rather than just "str".
32089 (aarch64_validate_march): Likewise.
32090 (aarch64_validate_mtune): Likewise.
32092 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
32094 PR rtl-optimization/78911
32095 * lra-assigns.c (must_not_spill_p): New function.
32096 (spill_for): Use it.
32098 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32100 PR tree-optimization/79981
32101 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
32102 ATOMIC_COMPARE_EXCHANGE ifn result.
32103 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
32104 IFN_ATOMIC_COMPARE_EXCHANGE.
32106 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32109 * opts.c (parse_sanitizer_options): Add missing question mark to
32110 "did you mean" message.
32112 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32114 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
32116 (VMULEUH_UNS): Likewise.
32117 (VMULOUB_UNS): Likewise.
32118 (VMULOUH_UNS): Likewise.
32119 * config/rs6000/rs6000.c (builtin_function_type): Remove
32120 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
32122 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32125 * read-rtl-function.c (function_reader::read_rtx_operand): Update
32126 x with result of extra_parsing_for_operand_code_0.
32127 (function_reader::extra_parsing_for_operand_code_0): Convert
32128 return type from void to rtx, returning x. When reading
32129 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
32130 larger size containing struct block_symbol.
32132 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
32134 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
32135 -mfloat128-hardware without -m64.
32137 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
32140 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
32141 entries to the case statement that marks unsigned arguments to
32142 overloaded functions.
32144 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32146 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
32147 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
32149 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
32152 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
32153 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
32155 2017-03-10 Martin Liska <mliska@suse.cz>
32159 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
32160 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
32163 2017-03-10 Olivier Hainque <hainque@adacore.com>
32165 * tree-switch-conversion (array_value_type): Start by resetting
32166 candidate type to it's main variant.
32168 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32170 PR rtl-optimization/79909
32171 * combine.c (try_combine): Use simplify_replace_rtx on individual
32172 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
32173 of the whole CALL_INSN_FUNCTION_USAGE.
32175 PR tree-optimization/79972
32176 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
32177 get_range_info on SSA_NAMEs. Formatting fixes.
32179 2017-03-10 Richard Biener <rguenther@suse.de>
32180 Jakub Jelinek <jakub@redhat.com>
32182 PR tree-optimization/77975
32183 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
32184 edge to be constant.
32185 (get_val_for): For constant x return it. Formatting fix.
32186 (loop_niter_by_eval): Avoid pointless looping if the next iteration
32187 would use the same bases as the current one.
32189 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32191 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
32192 instead of vec_select for V1TImode.
32193 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
32195 (VSX_LE_128): Add V1TI to this mode iterator.
32196 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
32197 (*vsx_le_perm_store_<mode>): Likewise.
32198 (pre-reload splitter for VSX stores): Likewise.
32199 (post-reload splitter for VSX stores): Likewise.
32200 (*vsx_xxpermdi2_le_<mode>): Likewise.
32201 (*vsx_lxvd2x2_le_<mode>): Likewise.
32202 (*vsx_stxvd2x2_le_<mode>): Likewise.
32204 2017-03-09 Michael Eager <eager@eagercon.com>
32206 Correct failures with --enable-checking=yes,rtl.
32208 * config/microblaze/microblaze.c (microblaze_expand_shift):
32209 Replace GET_CODE test with CONST_INT_P and INTVAL test with
32210 test for const0_rtx.
32211 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
32212 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
32214 2017-03-09 Richard Biener <rguenther@suse.de>
32216 PR tree-optimization/79977
32217 * graphite-scop-detection.c (scop_detection::merge_sese):
32218 Handle the case of extra exits to blocks dominating the entry.
32220 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
32222 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
32225 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
32227 PR rtl-optimization/79949
32228 * lra-constraints.c (process_alt_operands): Check memory when
32229 trying to predict a cycle. Print about the overall increase.
32231 2017-03-09 Richard Biener <rguenther@suse.de>
32233 PR middle-end/79971
32234 * gimple-expr.c (useless_type_conversion_p): Preserve
32235 TYPE_SATURATING for fixed-point types.
32237 2017-03-09 Richard Biener <rguenther@suse.de>
32240 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
32241 alignment of BLKmode params.
32243 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32246 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
32247 (VALL_NO_V2Q): Likewise.
32249 * config/aarch64/aarch64-simd.md
32250 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
32252 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
32253 VALL_NO_V2Q mode iterator.
32254 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
32256 2017-03-09 Martin Liska <mliska@suse.cz>
32258 PR tree-optimization/79631
32259 * tree-chkp-opt.c (chkp_is_constant_addr): Call
32260 tree_int_cst_sign_bit just for INTEGER constants.
32262 2017-03-09 Martin Liska <mliska@suse.cz>
32266 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
32269 2017-03-09 Marek Polacek <polacek@redhat.com>
32272 * tree.c (inchash::add_expr): Handle TREE_VEC.
32274 2017-03-09 Martin Liska <mliska@suse.cz>
32277 (chkp_narrow_size_and_offset): New function.
32278 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
32279 (void chkp_parse_bit_field_ref): New function.
32280 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
32281 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
32283 2017-03-09 Martin Liska <mliska@suse.cz>
32286 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
32287 (chkp_find_bounds_1): Remove gcc_unreachable.
32289 2017-03-09 Jakub Jelinek <jakub@redhat.com>
32292 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
32293 BUILT_IN_SYNC*, determine the access type from the size suffix and
32294 always build a MEM_REF with that type. Handle forgotten
32295 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
32298 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
32299 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
32300 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
32301 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
32302 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
32303 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
32304 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
32305 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
32306 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
32307 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
32308 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
32309 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
32310 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
32311 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
32312 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
32313 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
32314 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
32315 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
32316 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
32317 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
32318 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
32319 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
32320 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
32321 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
32322 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
32323 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
32324 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
32325 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
32326 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
32327 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
32328 definitions outside of __OPTIMIZE__ guarded section.
32331 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
32332 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
32333 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
32334 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
32337 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32339 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
32340 ("vfenez<mode>"): Add missing constraints.
32342 2017-03-08 Martin Sebor <msebor@redhat.com>
32345 * config/nds32/nds32.c (nds32_option_override):
32346 Fix misspelled diagnostic.
32348 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32351 * gimplify.c (gimplify_omp_for): Replace index var in outer
32352 taskloop statement with an artificial variable and add
32353 OMP_CLAUSE_PRIVATE clause for it.
32355 2017-03-08 Richard Biener <rguenther@suse.de>
32357 PR tree-optimization/79955
32358 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
32359 for accesses that are completely outside of the variable.
32361 2017-03-08 Andrew Haley <aph@redhat.com>
32363 PR tree-optimization/79943
32364 * tree-ssa-loop-split.c (compute_new_first_bound): When
32365 calculating the new upper bound, (END-BEG) should be added, not
32368 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32370 * config/avr/avr.md (setmemhi): Make sure match_dup
32371 operand number comes before match_scratch.
32373 2017-03-08 Richard Biener <rguenther@suse.de>
32375 PR tree-optimization/79920
32376 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
32377 with ncopies == 1 to ...
32378 (vect_transform_slp_perm_load): ... here. Properly compute
32379 all element loads by iterating VF times over the group. Do
32380 not handle ncopies (computed in a broken way) in
32381 vect_create_mask_and_perm.
32383 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32386 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
32387 is a uniform vector, use uniform_vector_p return value instead of
32388 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
32390 2017-03-07 Marek Polacek <polacek@redhat.com>
32392 PR middle-end/79809
32393 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
32394 (alloca_call_type): Likewise.
32396 2017-03-07 Martin Liska <mliska@suse.cz>
32398 * gcov.c (process_args): Put comment to correct location.
32400 2017-03-07 Martin Liska <mliska@suse.cz>
32402 PR middle-end/68270
32403 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
32404 Use array_at_struct_end_p instead of DECL_CHAIN (field).
32405 (chkp_narrow_bounds_for_field): Likewise.
32406 (chkp_parse_array_and_component_ref): Pass one more argument to
32409 2017-03-07 Richard Biener <rguenther@suse.de>
32411 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
32414 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
32416 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
32417 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
32419 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32422 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
32423 to end of description.
32424 (PARAM_MAX_STORES_TO_MERGE): Likewise.
32426 2017-03-07 Jakub Jelinek <jakub@redhat.com>
32428 PR rtl-optimization/79901
32429 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
32431 (*avx512f_<code><mode>3<mask_name>): ... this.
32432 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
32433 iterator instead of VI8_AVX2_AVX512BW.
32435 PR rtl-optimization/79901
32436 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
32437 min/max expander, expand it using expand_vec_cond_expr.
32440 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
32443 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32446 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
32447 to void * for PCH reasons.
32448 * dwarf2out.c (output_loc_operands, output_die): Cast
32449 v.val_vec.array to unsigned char *.
32451 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
32454 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
32457 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
32459 PR rtl-optimization/79571
32460 * lra-constraints.c (process_alt_operands): Calculate static
32461 reject and subtract it from overall when only addresses will be
32464 2017-03-06 Julia Koval <julia.koval@intel.com>
32467 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
32468 incoming stack boundary to 128 for 64-bit targets.
32470 2017-03-06 Richard Biener <rguenther@suse.de>
32472 PR tree-optimization/79894
32473 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
32474 to NULL after folding it.
32476 2017-03-06 Richard Biener <rguenther@suse.de>
32478 PR tree-optimization/79824
32479 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
32480 check disabling peeling for gaps.
32482 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
32484 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
32485 attributes): Document gettimeofday.
32487 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32489 * config/s390/s390.c (s390_option_override_internal): Set
32490 PARAM_MIN_VECT_LOOP_BOUND
32492 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32494 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
32495 * config/s390/s390.md: Likewise.
32497 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32500 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
32501 (<avx2_avx512>_perm<mode>): Rename to ...
32502 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
32504 (<avx512>_perm<mode>_mask): Rename to ...
32505 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
32507 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
32508 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
32509 instead of VI8F_256_512.
32510 (avx512f_perm<mode>): New define_expand.
32511 (avx512f_perm<mode>_mask): Likewise.
32512 (avx512f_perm<mode>_1<mask_name>): New define_insn.
32513 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
32515 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32517 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
32518 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
32520 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
32522 2017-03-06 Martin Liska <mliska@suse.cz>
32525 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
32526 when having a SSA NAME w/o VAR_DECL assigned to it.
32528 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32530 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
32531 msa_dpsub_<su>_d): Fix MODE for vec_select.
32533 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32535 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
32537 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
32539 2017-03-06 Richard Biener <rguenther@suse.de>
32541 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
32542 * plugin.c (register_plugin_info): Likewise.
32543 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
32545 2017-03-05 Jakub Jelinek <jakub@redhat.com>
32547 * config/i386/sse.md (sse_storehps, sse_storelps,
32548 avx_<castmode><avxsizesuffix>_<castmode>,
32549 avx512f_<castmode><avxsizesuffix>_<castmode>,
32550 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
32551 in condition that at least one operand is not a MEM.
32553 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32555 PR middle-end/79805
32556 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
32557 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
32559 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
32560 gimple_call_nothrow_p flag based on whether original builtin can throw.
32561 If it can, emit following stmts on the fallthrough edge.
32562 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
32563 don't create new bb if inserting just debug stmts on the edge, try to
32564 insert them on the fallthru bb or just reset debug stmts.
32566 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
32569 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
32570 restore recog_data (including the operand rtxes inside it) around
32571 the call to get_insn_template.
32573 2017-03-03 Martin Sebor <msebor@redhat.com>
32575 PR tree-optimization/79699
32576 * context.c (context::~context): Free MPFR caches to avoid
32577 a memory leak on program exit.
32579 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32581 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
32582 Use wide_int::ulow () instead of .elt (0).
32584 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32586 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
32587 (*pushxf): Limit oF constraint to 32bit targets and add oC
32588 constraint for 64bit targets.
32589 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
32590 (*pushdf): Change rmF constraint to rmC.
32592 2017-03-03 Martin Liska <mliska@suse.cz>
32594 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
32595 Remove unused variable.
32597 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32600 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
32601 is a memory operand, increase num_memory.
32602 (ix86_expand_args_builtin): Likewise.
32604 2017-03-03 Jan Hubicka <jh@suse.cz>
32607 * ipa-devirt.c (maybe_record_node): Properly handle
32608 __cxa_pure_virtual visibility.
32610 2017-03-03 Martin Liska <mliska@suse.cz>
32612 PR tree-optimization/79803
32613 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
32615 (pass_loop_prefetch::execute): Disabled optimization if an
32616 assumption about L1 cache size is not met.
32618 2017-03-03 Martin Liska <mliska@suse.cz>
32620 PR rtl-optimization/79574
32621 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
32622 (hash_scan_set): Likewise.
32623 (dump_hash_table): Likewise.
32624 (hoist_code): Likewise.
32626 2017-03-03 Richard Biener <rguenther@suse.de>
32628 * fixed-value.c (fixed_from_string): Restore use of elt (1)
32629 in place of uhigh ().
32630 (fixed_convert_from_real): Likewise.
32632 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32635 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
32637 2017-03-03 Richard Biener <rguenther@suse.de>
32639 PR middle-end/79818
32640 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
32641 TYPE_OVERFLOW_UNDEFINED check.
32643 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32645 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
32647 (vector_ae_<mode>_p): Likewise.
32648 (vector_nez_<mode>_p): Likewise.
32649 (vector_ne_v2di_p): Likewise.
32650 (vector_ae_v2di_p): Likewise.
32651 (vector_ne_<mode>_p): Likewise.
32652 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
32654 (vsx_tsqrt<mode>2_fe): Likewise.
32656 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
32659 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
32661 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32663 PR rtl-optimization/79780
32664 * cprop.c (one_cprop_pass): When second and further conditional trap
32665 in a single basic block is turned into an unconditional trap, turn it
32666 into a deleted note to avoid RTL verification failures.
32668 2017-03-02 Richard Biener <rguenther@suse.de>
32670 * fold-const.c (const_binop): Use ulow () instead of elt (0).
32672 2017-03-02 Richard Biener <rguenther@suse.de>
32674 PR tree-optimization/79345
32676 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
32677 param and abort the walk, returning -1 if it is hit.
32678 (walk_aliased_vdefs): Take a limit param and pass it on.
32679 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
32680 defaulting to 0 and return a signed int.
32681 * tree-ssa-uninit.c (struct check_defs_data): New struct.
32682 (check_defs): New helper.
32683 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
32684 about uninitialized memory.
32685 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
32686 bogus uninitialized warning.
32687 (fixed_convert_from_real): Likewise.
32689 2017-03-02 Bin Cheng <bin.cheng@arm.com>
32691 PR tree-optimization/66768
32692 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
32693 iv_use if base object can't be determined.
32695 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32697 PR tree-optimization/79345
32698 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
32699 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
32700 (get_pattern_stats): Initialize it.
32701 * genemit.c (gen_expand): Verify match_scratch numbers come after
32702 match_operand/match_dup numbers.
32703 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
32704 match_scratch numbers.
32705 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
32707 * config/s390/s390.md (trunctdsd2): Likewise.
32709 2017-03-02 Richard Biener <rguenther@suse.de>
32711 * wide-int.h (wide_int_storage::operator=): Implement in terms
32714 2017-03-02 Richard Biener <rguenther@suse.de>
32716 PR tree-optimization/79777
32717 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
32718 the to insert expression to sth existing.
32720 2017-03-01 Martin Sebor <msebor@redhat.com>
32722 PR middle-end/79692
32723 * gimple-ssa-sprintf.c
32724 (directive::known_width_and_precision): New function.
32725 (format_integer): Use it.
32726 (get_mpfr_format_length): Consider the full range of precision
32727 when computing %g output with the # flag. Set the likely byte
32728 count to 3 rather than 1 when precision is indeterminate.
32729 (format_floating): Correct the lower bound of precision.
32731 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32733 * doc/invoke.texi: Document default code model for 64-bit Linux.
32735 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32738 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
32739 udiv rather than div since input pattern is unsigned.
32741 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
32743 * config/i386/i386.c (print_reg): Warn for values of
32744 unsupported size in integer register.
32746 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
32749 * config/rs6000/predicates.md (current_file_function_operand): Do
32750 not allow self calls to be local if the function is replaceable.
32752 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
32755 * config/rs6000/altivec.h (vec_ctz and others): Change the
32756 preprocessor macro that controls conditional compilation from
32757 _ARCH_PWR9 to __POWER9_VECTOR__.
32758 (vec_all_ne): Change parameterization of __altivec_scalar_pred
32759 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
32760 control (instead of _ARCH_PWR9 control) so that template
32761 definition uses power9-specific function.
32762 (vec_any_eq): Likewise.
32763 (vec_all_ne): Change macro definition to use a power9-specific
32764 expansion under #ifdef __POWER9_VECTOR__ control (instead of
32765 _ARCH_PWR9 control).
32766 (vec_any_eq) Likewise.
32767 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
32768 expansion for CMPNEF to remove support for xvcmpnesp instruction.
32769 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
32770 support for xvcmpnedp instruction.
32771 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
32772 macro expansion so that Power9 implementation of vec_all_ne does
32773 not use the AltiVec predicate framework.
32774 (VCMPNEH_P): Likewise.
32775 (VCMPNEW_P): Likewise.
32776 (VCMPNED_P): Likewise.
32777 (VCMPNEFP_P): Likewise.
32778 (VCMPNEDP_P): Likewise.
32779 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
32780 implementation of vec_any_eq to not use AltiVec predicate
32782 (VCMPAEH_P): Likewise.
32783 (VCMPAEW_P): Likewise.
32784 (VCMPAED_P): Likewise.
32785 (VCMPAEFP_P): Likewise.
32786 (VCMPAEDP_P): Likewise.
32787 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
32788 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
32789 not use the AltiVec predicate framework.
32790 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
32791 of vec_any_eq to not use AltiVec predicate framework.
32792 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32793 support for predefined __POWER9_VECTOR__ macro to indicate that
32794 Power9 instruction selection is enabled.
32795 (altivec_overloaded_builtins): Remove extraneous
32796 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
32797 function argument types RS6000_BTI_bool_V16QI and
32798 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
32799 entry for overloaded function argument types RS6000_BTI_bool_V4SI
32800 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
32801 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
32802 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
32803 Power9 for implementations of vec_cmpne. Change the signature for
32804 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
32805 (representing vec_all_ne) to remove the previously described first
32806 argument of type RS6000_BTI_INTSI, as this was an artifact of
32807 reliance on the AltiVec predicate framework, which is no longer
32808 used in the implementation of these functions. Add
32809 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
32810 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
32811 since, unlike the AltiVec predicate framework implementation, we
32812 do not share function descriptors between vec_alle and vec_anyeq.
32813 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
32814 set of modes that receive special treatment even when
32815 TARGET_P9_VECTOR is true. The special treatment emits code that
32816 does not depend on Power9 instructions.
32817 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
32818 define_expand to not rely on AltiVec predicate framework.
32819 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32821 (vector_ne_v2di_p): Change this define_expand to not rely on
32822 AltiVec predicate framework.
32823 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
32825 (vector_ne_<mode>_p): Change this define_expand to not rely on
32826 AltiVec predicate framework.
32827 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32829 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
32830 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
32831 define_insn pattern.
32832 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
32833 define_insn pattern because the xvcmpne<VSs>. instruction is not
32835 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
32836 instruction is not supported.
32838 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32840 * config/nvptx/nvptx.c: Include intl.h.
32842 2017-03-01 Martin Jambor <mjambor@suse.cz>
32845 * ipa-prop.h (ipa_bits): Removed field known.
32846 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
32847 to pointers. Adjusted their comments to warn about their sharing.
32848 (ipcp_transformation_summary): Change bits to a vector of pointers.
32849 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
32850 (ipa_get_ipa_bits_for_value): Declare.
32851 * tree-vrp.h (value_range): Mark as GTY((for_user)).
32852 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
32853 (ipa_bits_hash_table): Likewise.
32854 (ipa_vr_ggc_hash_traits): Likewise.
32855 (ipa_vr_hash_table): Likewise.
32856 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
32857 being pointers and vr_known being removed.
32858 (ipa_set_jf_unknown): Likewise.
32859 (ipa_get_ipa_bits_for_value): New function.
32860 (ipa_set_jfunc_bits): Likewise.
32861 (ipa_get_value_range): New overloaded functions.
32862 (ipa_set_jfunc_vr): Likewise.
32863 (ipa_compute_jump_functions_for_edge): Use the above functions to
32864 construct bits and vr parts of jump functions.
32865 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
32866 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32868 (ipcp_grow_transformations_if_necessary): Also allocate
32869 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32871 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
32872 them. Fix too long lines.
32873 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
32874 vr_known being removed.
32875 (ipa_read_jump_function): Use new setter functions to construct bits
32876 and vr parts of jump functions or set them to NULL.
32877 (write_ipcp_transformation_info): Adjust for bits being pointers.
32878 (read_ipcp_transformation_info): Likewise.
32879 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
32881 Include gt-ipa-prop.h.
32882 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
32884 (ipcp_store_bits_results): Likewise.
32885 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
32886 Do not write to existing jump functions but use a temporary instead.
32888 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32891 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
32892 attempt to use its first operand as BIT_FIELD_REF base.
32894 2017-03-01 Richard Biener <rguenther@suse.de>
32896 PR middle-end/79721
32897 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
32898 interpolating formula in wrapping arithmetic.
32899 (chrec_apply): Convert chrec_evaluate return value to wanted type.
32901 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32903 PR tree-optimization/79734
32904 * tree-vect-generic.c (expand_vector_condition): Optimize
32905 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
32906 Handle VEC_COND_EXPR where comparison has different inner width from
32907 type's inner width.
32909 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
32911 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
32912 markup, and similar issues. Remove @opindex entries for things
32913 that aren't options. Add missing -mmpy-option entries.
32915 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32917 PR tree-optimization/79737
32918 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
32919 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
32920 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
32921 instead of byte_size. Formatting fix.
32922 (shift_bytes_in_array_right): Formatting fix.
32924 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
32927 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
32928 condition on optimize for the leaf function test.
32930 2017-02-28 Martin Liska <mliska@suse.cz>
32933 * read-rtl-function.c (function_reader::handle_unknown_directive):
32934 Bail out when one uses -flto.
32936 2017-02-28 Martin Liska <mliska@suse.cz>
32938 * common.opt: Replace space with tabular for options of <number>
32940 * config/i386/i386.opt: Show <number> value for
32941 -mlarge-data-threshold.
32942 * opts.c (print_filtered_help): Do not display number in hexadecimal
32945 2017-02-28 Martin Liska <mliska@suse.cz>
32947 * common.opt: Fix --help=option -Q for options which are of
32950 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
32952 * config/i386/i386.c (print_reg): Error out for values
32953 of 8-bit size in invalid integer register.
32955 2017-02-28 Martin Sebor <msebor@redhat.com>
32957 PR tree-optimization/79691
32958 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
32960 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32963 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
32964 gcc_unreachable with output_operand_lossage.
32966 2017-02-28 Richard Biener <rguenther@suse.de>
32968 PR tree-optimization/79740
32969 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
32971 (visit_nary_op): Insert the nary into the hashtable if we
32972 pattern-matched sth.
32973 * tree-ssa-pre.c (eliminate_insert): Robustify.
32975 2017-02-28 Richard Biener <rguenther@suse.de>
32977 PR middle-end/79731
32978 * fold-const.c (decode_field_reference): Reject out-of-bound
32981 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32983 * config/i386/i386.c: Include intl.h.
32984 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
32985 instead of just cond ? "..." : "...".
32986 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
32987 * coverage.c (read_counts_file): Likewise.
32988 * omp-offload.c: Include intl.h.
32989 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
32990 of just cond ? "..." : "...".
32991 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
32992 of just cond ? "..." : "...".
32994 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
32997 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
32999 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33000 'tune for' CPU name.
33001 * config/arm/arm-cpu-data.h: Regenerated.
33003 2017-02-28 Richard Biener <rguenther@suse.de>
33005 PR tree-optimization/79732
33006 * tree-inline.c (expand_call_inline): Do not shadow var.
33008 2017-02-28 Richard Biener <rguenther@suse.de>
33010 PR tree-optimization/79723
33011 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33012 address-space properly.
33014 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33016 * doc/optinfo.texi (Optimization groups): Fix option used for
33018 * doc/invoke.texi (-fopt-info): Document "omp".
33019 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
33020 (OPTGROUP_ALL): Add OPTGROUP_OMP.
33021 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
33022 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
33023 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
33025 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
33027 * dumpfile.c (optgroup_options): Instead of "openmp", associate
33028 OPTGROUP_OMP with "omp".
33030 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
33033 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
33034 for arithmetic shift of unsigned V2DI.
33036 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
33038 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
33039 arc/linux.h headers.
33040 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
33041 (LINK_SPEC): Likewise.
33042 (ARC_TLS_EXTRA_START_SPEC): Likewise.
33043 (EXTRA_SPECS): Likewise.
33044 (STARTFILE_SPEC): Likewise.
33045 (ENDFILE_SPEC): Likewise.
33046 (LIB_SPEC): Likewise.
33047 (TARGET_SDATA_DEFAULT): Likewise.
33048 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
33049 (MULTILIB_DEFAULTS): Likewise.
33050 (DWARF2_UNWIND_INFO): Likewise.
33051 * config/arc/big.h: New file.
33052 * config/arc/elf.h: Likewise.
33053 * config/arc/linux.h: Likewise.
33054 * config/arc/t-uClibc: Remove.
33056 2017-02-27 Bin Cheng <bin.cheng@arm.com>
33058 PR tree-optimization/77536
33059 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
33060 (tree_transform_and_unroll_loop): Use above function to compute the
33061 estimated niter of unrolled loop and use it when scaling profile.
33062 Also use count info rather than frequency if it's non-zero.
33063 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
33064 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
33065 (vect_transform_loop): Call above function.
33067 2017-02-27 Richard Biener <rguenther@suse.de>
33069 PR tree-optimization/45397
33070 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
33071 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
33072 (visit_nary_op): Add pattern matching for CSEing sign-changed
33073 or truncated operations with wider ones.
33075 2017-02-27 Richard Biener <rguenther@suse.de>
33077 PR tree-optimization/79690
33078 * tree-vect-stmts.c (vectorizable_store): Use vector type
33079 built from the DR with address-space.
33081 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
33083 * doc/invoke.texi (Optimize Options): Refine the description
33084 of asan-use-after-return.
33086 2017-02-25 Alan Modra <amodra@gmail.com>
33088 PR rtl-optimization/79584
33089 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
33090 base, not ad->base_term, the reg within base. Remove assertion
33091 that ad->base == ad->base_term. Replace gen_int_mode using
33092 bogus mode with const0_rtx.
33094 2017-02-25 Jakub Jelinek <jakub@redhat.com>
33096 PR middle-end/79396
33097 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
33098 FMA_EXPR like tcc_binary or tcc_unary.
33100 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
33103 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
33105 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
33106 (output_loc_operands): Handle DW_OP_call_ref and
33107 DW_OP_GNU_variable_value.
33108 (struct variable_value_struct): New type.
33109 (struct variable_value_hasher): Likewise.
33110 (variable_value_hash): New variable.
33111 (string_types): Remove.
33112 (copy_loc_descr): New function.
33113 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
33114 (prepend_loc_descr_to_each): New function.
33115 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
33116 instead of add_loc_descr_to_each if the first argument is single
33117 location list and the second has multiple.
33118 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
33119 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
33120 when looking for variable value which doesn't have other location info.
33121 (loc_list_from_tree): Formatting fix.
33122 (gen_array_type_die): Simplify DW_AT_string_length handling.
33123 (adjust_string_types): Remove.
33124 (gen_subprogram_die): Don't call adjust_string_types nor test/set
33125 string_types. Call resolve_variable_values.
33126 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
33127 (resolve_addr_in_expr): Likewise. Add A argument.
33128 (copy_deref_exprloc): Remove deref argument. Adjust for the
33129 original expression being DW_OP_GNU_variable_value with optionally
33130 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
33131 optionally after it.
33132 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
33133 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
33134 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
33135 (variable_value_hasher::hash, variable_value_hasher::equal): New
33137 (resolve_variable_value_in_expr, resolve_variable_value,
33138 resolve_variable_values, note_variable_value_in_expr,
33139 note_variable_value): New functions.
33140 (dwarf2out_early_finish): Call note_variable_value on all toplevel
33143 2017-02-24 Jakub Jelinek <jakub@redhat.com>
33146 * opts.h (handle_generated_option): Add GENERATED_P argument.
33147 * opts-common.c (handle_option): Adjust function comment.
33148 (handle_generated_option): Add GENERATED_P argument, pass it to
33150 (control_warning_option): Pass false to handle_generated_option
33152 * opts.c (maybe_default_option): Pass true to handle_generated_option
33154 * optc-gen.awk: Likewise.
33156 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33158 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
33159 a REG, look at the REG it is a SUBREG of.
33160 (splitter for cmpeqsi_t): Ditto.
33162 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33164 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
33165 the special USEs with the pattern of the insn, not the insn itself.
33167 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
33170 * doc/invoke.texi: Document -mload-store-pairs.
33172 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33173 Sandra Loosemore <sandra@codesourcery.com>
33175 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
33176 argument isn't a CONST_INT.
33177 (nios2_alternate_compare_const): Assert op is a CONST_INT.
33178 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
33179 (nios2_validate_compare): Bypass alternate compare logic if *op2
33180 is not a CONST_INT.
33181 (ldstwm_operation_p): Return false if first_base is not a REG or
33182 if first_offset is not a CONST_INT.
33184 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33186 * config/cris/cris.md: Use correct operand in a define_peephole2.
33188 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33190 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
33192 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33194 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
33195 this_insn if it is an INSN or JUMP_INSN.
33196 (force_offsettable): Look at base, not at addr.
33197 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
33198 on things that aren't necessarily CONST_INTs.
33200 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
33202 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
33203 -mfpmath=sse is the default also for x86-32 targets with SSE2
33204 instruction set when @option{-ffast-math} is enabled
33206 2017-02-24 Jeff Law <law@redhat.com>
33208 PR rtl-optimizatoin/79286
33209 * ira.c (update_equiv_regs): Drop may_trap_p exception to
33212 2017-02-24 Richard Biener <rguenther@suse.de>
33214 PR tree-optimization/79389
33215 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
33218 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
33220 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
33221 function comment to reflect reality.
33222 (loop_exits_before_overflow): Fix typo in function description.
33224 2017-02-24 Richard Biener <rguenther@suse.de>
33226 PR tree-optimization/79389
33227 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
33228 properly that a threading opportunity exists. Detect conditional
33229 copy/constant propagation opportunities.
33231 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
33233 * config/visium/visium.md (type): Add trap.
33234 (b): New mode attribute.
33235 (*btst): Rename into...
33236 (*btst<mode>): ...this and adjust.
33237 (*cbranchsi4_btst_insn): Rename into...
33238 (*cbranch<mode>4_btst_insn): ...this and adjust.
33239 (trap): New define_insn.
33241 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33243 PR tree-optimization/79389
33244 * ifcvt.c (struct noce_if_info): Add rev_cond field.
33245 (noce_reversed_cond_code): New function.
33246 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
33247 reversed_comparison_code. Formatting fix.
33248 (noce_try_store_flag): Test rev_cond != NULL in addition to
33249 reversed_comparison_code.
33250 (noce_try_store_flag_constants): Likewise.
33251 (noce_try_store_flag_mask): Likewise.
33252 (noce_try_addcc): Use rev_cond if non-NULL instead of
33253 reversed_comparison_code.
33254 (noce_try_cmove_arith): Likewise. Formatting fixes.
33255 (noce_try_minmax, noce_try_abs): Clear rev_cond.
33256 (noce_find_if_block): Initialize rev_cond.
33257 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
33258 instead of false as last argument never attempt to reverse it
33261 2017-02-23 Bin Cheng <bin.cheng@arm.com>
33263 PR tree-optimization/79663
33264 * tree-predcom.c (combine_chains): Process refs in reverse order
33265 only for ZERO length chains, and add explaining comment.
33267 2017-02-23 Jeff Law <law@redhat.com>
33269 PR tree-optimization/79578
33270 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
33271 in call to operand_equal_p.
33273 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
33276 * config/i386/cpuid.h: Fix another undefined behavior.
33278 2017-02-23 Richard Biener <rguenther@suse.de>
33280 PR tree-optimization/79683
33281 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
33282 vector types for data-refs.
33284 2017-02-23 Martin Liska <mliska@suse.cz>
33286 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
33288 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33290 PR middle-end/79665
33291 * internal-fn.c (get_range_pos_neg): Moved to ...
33292 * tree.c (get_range_pos_neg): ... here. No longer static.
33293 * tree.h (get_range_pos_neg): New prototype.
33294 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
33295 are known to be in between 0 and signed maximum inclusive, try to
33296 expand both unsigned and signed divmod and use the cheaper one from
33299 2017-02-22 Jeff Law <law@redhat.com>
33301 PR tree-optimization/79578
33302 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
33303 to compare base operands.
33305 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
33308 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
33309 gpc_reg_operand instead of fpr_reg_operand.
33311 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
33313 * config/mips/mips.c (mips_return_in_memory): Force FP
33314 vector types to be returned in memory for o32 ABI.
33316 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33318 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
33319 instead of DW_TAG_member for static data member declarations and don't
33320 set no_linkage_name for static inline data members.
33321 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
33324 2017-02-22 Martin Liska <mliska@suse.cz>
33326 * doc/invoke.texi: Replace inequality signs with square brackets
33329 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
33332 * lra-constraints.c (simplify_operand_subreg): Handle
33333 WORD_REGISTER_OPERATIONS targets.
33335 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33338 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
33339 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
33340 elimination by swapping fld*.
33342 2017-02-22 Richard Biener <rguenther@suse.de>
33344 PR tree-optimization/79673
33345 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
33346 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
33347 irrelevant address-space qualifiers and avoiding a
33348 ADDR_SPACE_CONVERT_EXPR from fold_convert.
33350 2017-02-22 Richard Biener <rguenther@suse.de>
33352 PR tree-optimization/79666
33353 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
33354 to not symbolically negate if that may introduce undefined
33357 2017-02-22 Martin Liska <mliska@suse.cz>
33360 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
33361 * data-streamer-out.c (streamer_write_gcov_count_stream):
33363 * value-prof.c (stream_out_histogram_value): Make assert more
33364 precise based on type of counter.
33366 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
33369 * config/i386/i386.md (standard_x87sse_constant_load splitter):
33370 Use nonimmediate_operand instead of memory_operand for operand 1.
33371 (float-extend standard_x87sse_constant_load splitter): Ditto.
33373 2017-02-21 Jeff Law <law@redhat.com>
33375 PR tree-optimization/79621
33376 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
33377 blocks with edges to themselves.
33379 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33382 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
33383 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
33384 Use gimple_call_builtin_p.
33387 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
33388 on temporarily removed DEBUG_INSNs.
33390 PR tree-optimization/79649
33391 * tree-loop-distribution.c (classify_partition): Give up on
33392 non-generic address space loads/stores.
33394 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
33396 * doc/loop.texi (Loop manipulation): Remove nonexistent
33397 tree_ssa_loop_version from the documentation.
33398 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
33400 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33403 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
33404 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
33405 * config/rs6000/rs6000.c: Include except.h.
33406 (rs6000_expand_split_stack_prologue): Call
33407 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
33409 2017-02-21 Martin Jambor <mjambor@suse.cz>
33412 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
33413 have been analyzed.
33415 2017-02-21 Martin Jambor <mjambor@suse.cz>
33417 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
33418 for backward compatibility only.
33419 * doc/invoke.texi (Option Summary): Remove all references to
33420 -fipa-cp-alignment.
33422 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
33426 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33428 * lra-constraints.c (curr_insn_transform): Handle
33429 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33431 2017-02-21 Martin Liska <mliska@suse.cz>
33433 * config/i386/i386.opt: Replace -masm-dialect with -masm.
33435 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
33437 PR translation/79638
33438 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
33440 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
33443 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
33444 (arm_function_ok_for_sibcall): Return false for an indirect call by
33445 descriptor if all the argument registers are used.
33446 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
33447 alignment of the function.
33449 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33451 PR tree-optimization/61441
33452 * simplify-rtx.c (simplify_const_unary_operation): For
33453 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
33454 the sNaN unmodified.
33456 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33458 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33459 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33460 instead of SYSTEM_HEADER_DIR.
33462 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
33463 Martin Liška <mliska@suse.cz>
33465 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
33466 Fix typos and grammar, use active voice, and clarify.
33468 2017-02-20 Marek Polacek <polacek@redhat.com>
33470 PR middle-end/79537
33471 * gimplify.c (gimplify_expr): Handle unused *&&L;.
33474 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
33476 2017-02-20 Jakub Jelinek <jakub@redhat.com>
33479 * config/i386/i386.c (ix86_expand_builtin): Handle
33480 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
33481 ix86_builtins_isa[fcode].isa as a requirement of those
33482 flags and any other flag in the bitmask.
33483 (ix86_init_mmx_sse_builtins): Use 0 instead of
33484 ~OPTION_MASK_ISA_64BIT as mask.
33485 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
33486 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
33487 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
33488 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
33490 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33493 * lra-constraints.c (split_reg): Check requested split mode
33494 is supported by the register.
33496 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33498 * lra-constraints.c (simplify_operand_subreg): Remove early
33501 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33504 * lra-constraints.c (curr_insn_transform): Tighten condition
33505 for converting SUBREG reloads from OP_OUT to OP_INOUT.
33507 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33510 * lra-constraints.c (curr_insn_transform): Handle
33511 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33513 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
33516 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
33518 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
33520 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
33523 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
33526 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33528 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
33531 2017-02-18 Jakub Jelinek <jakub@redhat.com>
33533 * final.c (last_columnnum, override_columnnum): New variables.
33534 (final_start_function): Set last_columnnum, pass it to begin_prologue
33535 hook and pass 0 to dwarf2out_begin_prologue.
33536 (final_scan_insn): Update override_columnnum. Pass last_columnnum
33537 to source_line debug hook.
33538 (notice_source_line): Compute last_columnnum and for debug_column_info
33539 return true on column changes.
33540 * debug.h (struct gcc_debug_hooks): Add column argument to
33541 source_line and begin_prologue hooks.
33542 (debug_nothing_int_charstar_int_bool): Remove prototype.
33543 (debug_nothing_int_int_charstar,
33544 debug_nothing_int_int_charstar_int_bool): New prototypes.
33545 (dwarf2out_begin_prologue): Add column argument.
33546 * debug.c (do_nothing_debug_hooks): Adjust source_line and
33547 begin_prologue hooks.
33548 (debug_nothing_int_charstar_int_bool): Remove.
33549 (debug_nothing_int_int_charstar,
33550 debug_nothing_int_int_charstar_int_bool): New functions.
33551 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
33552 through to dwarf2out_source_line.
33553 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
33554 (dwarf2out_source_line): Add column argument, emit it if requested.
33555 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
33557 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33558 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33559 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
33560 through to dwarf2out_begin_prologue.
33561 (vmsdbgout_source_line): Add column argument, pass it through to
33562 dwarf2out_source_line.
33563 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
33564 dbxout_source_line caller.
33565 (dbxout_source_line): Add column argument.
33567 * common.opt (gno-column-info, gcolumn-info): New options.
33568 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
33569 (check_die): Also test for multiple DW_AT_decl_column attributes.
33570 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
33571 DW_AT_decl_column if requested.
33572 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
33574 (gen_variable_die): Likewise.
33575 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
33576 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
33579 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
33580 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
33581 (ix86_handle_option): Handle OPT_m3dnowa.
33582 * doc/invoke.texi (-m3dnowa): Document.
33583 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
33584 -m3dnowa instead of -m3dnow -march=athlon.
33587 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
33588 instead of gcc_assert for K, r and R code checks. Formatting fixes.
33590 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33593 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
33594 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
33595 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
33596 generator for vsx_xxpermdi_<mode>_be.
33597 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
33598 force big-endian semantics.
33599 (vsx_xxpermdi_<mode>_be): New define_expand with same
33600 implementation as previous version of vsx_xxpermdi_<mode>.
33602 2017-02-17 Jakub Jelinek <jakub@redhat.com>
33604 PR tree-optimization/79327
33605 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
33606 variable, its initialization and use.
33608 2017-02-17 Julia Koval <julia.koval@intel.com>
33610 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
33611 (OPTION_MASK_ISA_PKU_UNSET): New.
33612 (ix86_handle_option): Handle -mrdpid.
33613 * config/i386/cpuid.h (bit_RDPID): New.
33614 * config/i386/driver-i386.c (host_detect_local_cpu):
33615 Detect RDPID feature.
33616 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
33617 * config/i386/i386-c.c (ix86_target_macros_internal):
33619 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
33620 (ix86_valid_target_attribute_inner_p): Add "rdpid".
33621 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
33622 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
33623 * config/i386/i386.md (define_insn "rdpid"): New.
33624 * config/i386/i386.opt Add -mrdpid.
33625 * config/i386/immintrin.h (_rdpid_u32): New.
33627 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
33629 PR rtl-optimization/79541
33630 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
33631 instead of transforming it into USE.
33633 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
33635 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
33636 If HONOR_SNANS (SFmode) force the input to a register.
33637 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
33638 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
33639 an frsp or similar insn.
33641 2017-02-17 Martin Liska <mliska@suse.cz>
33643 PR rtl-optimization/79577
33644 * params.def (selsched-max-sched-times): Increase minimum to 1.
33646 2017-02-17 Martin Liska <mliska@suse.cz>
33648 PR rtl-optimization/79574
33649 * gcse.c (want_to_gcse_p): Prevent integer overflow.
33651 2017-02-17 Martin Liska <mliska@suse.cz>
33653 PR tree-optimization/79529
33654 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
33655 ssa_defined_default_def_p to handle cases which are implicitly
33657 * tree-ssa.c (ssa_defined_default_def_p): New function.
33658 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
33659 which are implicitly defined.
33660 * tree-ssa.h (ssa_defined_default_def_p): Declare.
33662 2017-02-17 Richard Biener <rguenther@suse.de>
33664 PR middle-end/79576
33665 * params.def (max-ssa-name-query-depth): Limit to 10.
33667 2017-02-17 Richard Biener <rguenther@suse.de>
33669 PR tree-optimization/79552
33670 * tree-ssa-structalias.c (visit_loadstore): Properly verify
33673 2017-02-17 Richard Biener <rguenther@suse.de>
33676 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
33678 2017-02-17 Marek Polacek <polacek@redhat.com>
33680 PR middle-end/79536
33681 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
33682 (fold_negate_expr): New wrapper.
33684 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
33686 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
33687 Correct terminology and de-emphasize pre-standard behavior.
33689 2017-02-16 Alan Modra <amodra@gmail.com>
33691 PR rtl-optimization/79286
33692 * ira.c (def_dominates_uses): New function.
33693 (update_equiv_regs): Don't create an equivalence for insns that
33694 may trap where the register def does not dominate the use.
33696 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
33698 PR rtl-optimization/78127
33699 * lra.c (lra): Call lra_eliminate before finish the loop after
33702 2017-02-16 Richard Biener <rguenther@suse.de>
33704 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
33706 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
33707 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
33708 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
33709 (isl_val_int_from_wi): New function.
33710 (extract_affine_gmp): Rename to ...
33711 (extract_affine_wi): ... this, take a widest_int.
33712 (extract_affine_int): Just wrap extract_affine_wi.
33713 (add_param_constraints): Use isl_val_int_from_wi.
33714 (add_loop_constraints): Likewise, and extract_affine_wi.
33716 2017-02-15 Jeff Law <law@redhat.com>
33718 PR middle-end/79521
33719 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
33720 ira_init_register_move_cost_if_necessary.
33722 2017-02-15 Martin Sebor <msebor@redhat.com>
33724 PR middle-end/32003
33725 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
33726 removed in a prior commit.
33728 2017-02-15 Bin Cheng <bin.cheng@arm.com>
33730 PR tree-optimization/79347
33731 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
33732 counters during peeling.
33734 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
33736 * Makefile.in (site.exp): Remove "set ISLVER".
33738 2017-02-15 Jakub Jelinek <jakub@redhat.com>
33741 * real.c (real_from_integer): Call real_convert even for decimal.
33743 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
33746 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
33748 2017-02-14 Andrew Pinski <apinski@cavium.com>
33750 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
33751 cores and change the partno/implementer to be correct.
33752 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
33753 the 'B" as the implementer.
33754 * config/aarch64/aarch64-tune.md: Regenerate.
33756 2017-02-14 Carl Love <cel@us.ibm.com>
33758 * config/rs6000/rs6000.c: Add case statement entry to make the
33759 xvcvuxdsp built-in argument unsigned.
33760 * config/rs6000/vsx.md: Fix the source and return operand types so they
33761 match the instruction definitions from the ISA document. Fix typo
33762 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
33765 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
33768 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
33769 member early_clobber_alts.
33770 * lra-lives.c (reg_early_clobber_p): New.
33771 (process_bb_lives): Use it.
33772 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
33773 (debug_operand_data): Initialize early_clobber_alts.
33774 (setup_operand_alternative): Set up early_clobber_alts.
33775 (collect_non_operand_hard_regs): Ditto. Pass early clobber
33776 alternatives to new_insn_reg.
33777 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
33779 (lra_update_insn_regno_info): Pass the new arg.
33781 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33783 PR middle-end/79505
33784 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
33785 (new_oacc_loop_raw): Don't clear already cleared fields.
33788 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
33789 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
33790 _mm512_prefetch_i64gather_ps): New inline functions and macros.
33792 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
33795 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
33797 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
33800 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
33801 the extra instruction to the right place to store 128-bit constant
33804 2017-02-14 Martin Sebor <msebor@redhat.com>
33806 PR middle-end/79448
33807 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
33808 warning for strings of unknown length.
33810 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
33812 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
33814 2017-02-14 Jeff Law <law@redhat.com>
33817 * ira-costs.c (scan_one_insn): Initialize register move costs
33818 for pseudos seen in USE/CLOBBER insns.
33820 PR tree-optimization/79095
33821 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
33822 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
33823 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
33824 if the operands are known to be not equal, then the resulting range
33826 (intersect_ranges): If the new range is ~[0,0] and the old range is
33827 wide, then prefer ~[0,0].
33828 * tree-vrp.c (overflow_comparison_p_1): New function.
33829 (overflow_comparison_p): New function.
33830 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
33831 if NAME is used in an overflow test.
33832 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
33833 overflow check that can be expressed as an equality test, then adjust
33834 ops to be that equality test.
33836 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33838 * config/s390/s390-builtin-types.def: Remove flags argument.
33839 * config/s390/s390.c (s390_init_builtins): Likewise.
33841 2017-02-14 Martin Liska <mliska@suse.cz>
33843 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
33844 vector. Fix trailing white spaces.
33846 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
33848 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
33851 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33853 PR rtl-optimization/68664
33854 * config/arm/arm.c (arm_sched_can_speculate_insn):
33855 New function. Declare prototype.
33856 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33858 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33860 PR rtl-optimization/68664
33861 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
33863 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33865 2017-02-14 Amit Pawar <amit.pawar@amd.com>
33867 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
33868 max skip bytes for function, loop and jump.
33870 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33872 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
33873 ABS_EXPR for gimple dump.
33875 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33878 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
33880 PR tree-optimization/79408
33881 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
33882 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
33883 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
33884 also if rhs1 is INTEGER_CST.
33886 2017-02-14 Richard Biener <rguenther@suse.de>
33888 PR middle-end/79432
33889 * tree-into-ssa.c (insert_phi_nodes): When the function can
33890 have abnormal edges rewrite SSA names with broken use-def
33891 dominance out of SSA and register them for PHI insertion.
33893 2017-02-13 Martin Sebor <msebor@redhat.com>
33895 PR middle-end/79496
33896 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
33897 clearing info.nowrite flag when snprintf size argument is a range.
33899 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33901 * cprop.c (cprop_jump): Add missing space in string literal.
33902 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
33903 (get_constraint_for_component_ref): Likewise.
33904 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
33905 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
33906 * lra-constraints.c (process_alt_operands): Likewise.
33907 * ipa-inline.c (inline_small_functions): Likewise.
33908 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
33909 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
33910 * trans-mem.c (diagnose_tm_1_op): Likewise.
33911 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
33912 (grid_parallel_clauses_gridifiable): Likewise.
33914 * config/nvptx/mkoffload.c (process): Add space in between
33917 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
33918 "MOD4_SSE_REGS" and "ALL_REGS".
33920 * spellcheck.c (test_data): Add , in between "foo" and "food".
33922 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33925 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
33926 boundary crossing check and subsequent code generation agree.
33928 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33930 * config/aarch64/aarch64.c (has_memory_op): Delete.
33931 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
33934 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33936 PR rtl-optimization/79388
33937 PR rtl-optimization/79450
33938 * combine.c (distribute_notes): When removing TEM_INSN for which
33939 corresponding dest has last value recorded, invalidate that last
33942 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33944 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
33945 of explicit '@'. Add missing assembly comment marker on branch costs
33948 2017-02-13 Nathan Sidwell <nathan@acm.org>
33950 * gengtype-lex.l (<in_struct>): Add '/'.
33952 2017-02-13 Martin Liska <mliska@suse.cz>
33955 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
33957 2017-02-13 Richard Biener <rguenther@suse.de>
33959 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
33961 * configure: Re-generate.
33962 * config.in: Likewise.
33963 * graphite-dependences.c: Simplify as if
33964 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
33965 * graphite-isl-ast-to-gimple.c: Likewise.
33966 * graphite-optimize-isl.c: Likewise.
33967 * graphite-poly.c: Likewise.
33968 * graphite-sese-to-poly.c: Likewise.
33969 * graphite.h: Likewise.
33970 * toplev.c: Include isl/version.h and use isl_version () for
33971 printing the ISL version.
33972 * doc/install.texi: Update ISL requirement.
33974 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33976 * doc/standards.texi (Standards): Update reference to
33979 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33981 * doc/extend.texi (Named Address Spaces): sourceware.org now
33983 * doc/install.texi (Binaries): Ditto.
33986 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33988 * doc/cpp.texi: Replace "stringify"/"stringification" with C
33989 standard terminology "stringize"/"stringizing" throughout.
33990 * doc/cppinternals.texi: Likewise.
33992 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33994 * doc/extend.texi: Fix some spelling mistakes and typos.
33995 * doc/invoke.texi: Likewise.
33997 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34000 * params.def (inline-min-speedup) Change from 10 to 8.
34002 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34004 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34007 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34010 * ipa-inline-analysis.c (get_minimal_bb): New function.
34011 (record_modified): Use it.
34012 (remap_edge_change_prob): Handle also ancestor functions.
34014 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34016 * doc/contrib.texi (Contributors): Remove broken link into
34017 the Mauve CVS repository.
34019 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34021 PR middle-end/79454
34022 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
34023 result computation whenever lhs doesn't have vector mode, not
34024 just when it has BLKmode.
34026 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34028 * doc/makefile.texi (profiledbootstrap): Refer to the
34029 installation instructions only in textual form.
34031 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34034 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
34036 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34038 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
34039 (Specific): Update mingw-w64 reference.
34041 (Specific): Remove broken link to Renesas RX processor.
34043 2017-02-10 Richard Biener <rguenther@suse.de>
34045 * toplev.c (process_options): Do not mention obsolete graphite
34046 options when printing sorry message about missing graphite support.
34047 Mention -floop-nest-optimize.
34049 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
34051 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
34052 (vtst_p16): Likewise.
34053 (vtstq_p8): Likewise.
34054 (vtstq_p16): Likewise.
34056 (vtstq_p64): Likewise.
34057 * config/arm/arm_neon.h (vgetq_lane_p64): New.
34058 (vset_lane_p64): New.
34059 (vsetq_lane_p64): New.
34061 2017-02-10 Jakub Jelinek <jakub@redhat.com>
34063 PR tree-optimization/79411
34064 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
34065 stmt operands are SSA_NAMEs used in abnormal phis.
34066 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
34069 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34072 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
34075 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34077 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
34079 2017-02-09 Jakub Jelinek <jakub@redhat.com>
34081 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
34085 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
34086 not arbitrary TREE_CONSTANT.
34089 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
34090 "omp declare target link" attribute unless is_global_var.
34091 * omp-offload.c (find_link_var_op): Likewise.
34093 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
34094 Chung-Lin Tang <cltang@codesourcery.com>
34096 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
34098 (gimplify_adjust_omp_clauses): Don't delete TILE.
34099 (gimplify_omp_for): Deal with TILE.
34100 * internal-fn.c (expand_GOACC_TILE): New function.
34101 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
34103 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
34104 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
34106 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
34107 avoid DIV for outermost collapse var.
34108 (expand_oacc_for): Insert tile element loop as needed. Adjust.
34109 Remove out of date comments, fix whitespace.
34110 * omp-general.c (omp_extract_for_data): Deal with tiling.
34111 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
34112 adjust OLF_DIM_BASE value.
34113 (struct omp_for_data): Add tiling field.
34114 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
34115 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
34116 for auto loops. Remove default auto determining, moved to
34117 oacc_loop_fixed_partitions.
34118 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
34119 stmts, add e_mask field.
34120 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
34121 (oacc_thread_numbers): Use oacc_dim_call.
34122 (oacc_xform_tile): New.
34123 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
34124 (finish_oacc_loop): Adjust for ifns vector.
34125 (oacc_loop_discover_walk): Append loop abstraction sites to list,
34126 add case for GOACC_TILE fns.
34127 (oacc_loop_xform_loop): Delete.
34128 (oacc_loop_process): Iterate over call list directly, and add
34129 handling for GOACC_TILE fns.
34130 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
34132 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
34133 vector partitioning to outer loops. Assign 2 partitions to loops
34134 when available. Add TILE handling.
34135 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
34136 (execite_oacc_device_lower): Process GOACC_TILE fns,
34137 ignore unknown specs.
34138 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
34139 * tree.c (omp_clause_num_ops): Adjust TILE ops.
34140 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
34142 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
34144 * configure.ac (ACX_BUGURL): Update.
34145 * configure: Regenerate.
34147 2017-02-09 Richard Biener <rguenther@suse.de>
34149 PR tree-optimization/69823
34150 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
34151 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
34153 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34155 * config/arc/arc-c.def: Add __NPS400__ definition.
34156 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
34157 (TARGET_NPS400): Define.
34159 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34161 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
34163 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
34164 pointer, arch_info.
34165 (arc_cpu_types): Fill the arch_info field with a pointer into the
34166 arc_arch_types table.
34167 (arc_selected_cpu): Declare.
34168 * config/arc/arc.c (arc_selected_cpu): Make global.
34169 (arc_selected_arch): Delete.
34170 (arc_base_cpu): Delete.
34171 (arc_override_options): Remove references to deleted variables,
34172 update access to arch information.
34173 (ARC_OPT): Update access to arch information.
34174 (ARC_OPTX): Likewise.
34175 * config/arc/arc.h (arc_base_cpu): Remove declaration.
34176 (TARGET_ARC600): Update access to arch information.
34177 (TARGET_ARC601): Likewise.
34178 (TARGET_ARC700): Likewise.
34179 (TARGET_EM): Likewise.
34180 (TARGET_HS): Likewise.
34181 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
34184 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
34187 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
34188 condition/operands for integer GE/LE/GEU/LEU operations.
34190 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
34192 PR translation/79397
34193 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
34196 2017-02-08 Martin Jambor <mjambor@suse.cz>
34199 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
34200 whether allocation happened.
34201 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
34202 nothing was allocated.
34204 2017-02-08 Jakub Jelinek <jakub@redhat.com>
34206 PR tree-optimization/79408
34207 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
34208 constant, but SSA_NAME with a known integer range, use the minimum
34209 of that range instead of op1 to determine if modulo can be replaced
34210 with its first operand.
34212 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34214 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
34216 2017-02-08 Richard Biener <rguenther@suse.de>
34218 PR tree-optimization/71824
34219 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34220 Check all loops contained in the merged region.
34222 2017-02-07 Andrew Pinski <apinski@cavium.com>
34224 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
34226 2017-02-07 Andrew Pinski <apinski@cavium.com>
34228 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
34229 (thunderxt88): Likewise.
34230 (thunderxt81): Disable LSE and change v8.1 to v8.
34231 (thunderxt83): Likewise.
34233 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34234 Richard Biener <rguenther@suse.de>
34236 PR middle-end/79399
34237 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
34238 type from int to size_t.
34239 * ira-costs.c (struct_costs_size): Change type from int to size_t.
34241 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34243 PR rtl-optimization/79386
34244 * cprop.c (bypass_conditional_jumps): Initialize
34245 bypass_last_basic_block already before splitting bbs after
34246 unconditional traps...
34247 (bypass_conditional_jumps): ... rather than here.
34250 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
34251 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
34252 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
34253 fix -masm=intel patterns.
34255 2017-02-07 Richard Biener <rguenther@suse.de>
34257 PR tree-optimization/79256
34258 PR middle-end/79278
34259 * builtins.c (get_object_alignment_2): Use min_align_of_type
34260 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
34261 and ADJUST_FIELD_ALIGN.
34263 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
34265 * doc/tm.texi: Regenerate.
34266 * stor-layout.c (layout_decl): Adjust.
34267 (update_alignment_for_field): Likewise.
34268 (place_field): Likewise.
34269 (min_align_of_type): Likewise.
34270 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
34271 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
34272 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
34273 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
34274 * config/frv/frv.c (frv_adjust_field_align): Likewise.
34275 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
34276 * config/i386/i386.c (x86_field_alignment): Likewise.
34277 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
34278 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
34279 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
34280 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
34281 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
34282 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
34286 2017-01-30 Richard Biener <rguenther@suse.de>
34288 PR tree-optimization/79256
34289 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34290 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34293 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
34295 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
34296 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
34297 builtins to SImode and emit a zero-extend, if necessary.
34299 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34301 * docs/invoke.texi (RISC-V Options): Alphabetize.
34303 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34305 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
34308 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34310 * config/riscv/riscv.c: New file.
34311 * common/config/riscv/riscv-common.c: Likewise.
34312 * config.gcc: Likewise.
34313 * config/riscv/constraints.md: Likewise.
34314 * config/riscv/elf.h: Likewise.
34315 * config/riscv/generic.md: Likewise.
34316 * config/riscv/linux.h: Likewise.
34317 * config/riscv/multilib-generator: Likewise.
34318 * config/riscv/peephole.md: Likewise.
34319 * config/riscv/pic.md: Likewise.
34320 * config/riscv/predicates.md: Likewise.
34321 * config/riscv/riscv-builtins.c: Likewise.
34322 * config/riscv/riscv-c.c: Likewise.
34323 * config/riscv/riscv-ftypes.def: Likewise.
34324 * config/riscv/riscv-modes.def: Likewise.
34325 * config/riscv/riscv-opts.h: Likewise.
34326 * config/riscv/riscv-protos.h: Likewise.
34327 * config/riscv/riscv.h: Likewise.
34328 * config/riscv/riscv.md: Likewise.
34329 * config/riscv/riscv.opt: Likewise.
34330 * config/riscv/sync.md: Likewise.
34331 * config/riscv/t-elf-multilib: Likewise.
34332 * config/riscv/t-linux: Likewise.
34333 * config/riscv/t-linux-multilib: Likewise.
34334 * config/riscv/t-riscv: Likewise.
34335 * configure.ac: Likewise.
34336 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
34337 Waterman as RISC-V maintainers.
34338 * doc/install.texi: Add RISC-V entries.
34339 * doc/invoke.texi: Add RISC-V options section.
34340 * doc/md.texi: Add RISC-V constraints section.
34341 * configure: Regenerated.
34343 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
34346 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
34347 false values to be constant vectors with all 0 or all 1 bits set.
34348 (vcondu<mode><mode>): Likewise.
34349 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
34351 (fpmask_comparison_operator): Update comment.
34352 (vecint_comparison_operator): New predicate.
34353 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
34354 vector conditionals when the true and false values are constant
34355 vectors with all 0 bits or all 1 bits set.
34357 2017-02-06 Martin Sebor <msebor@redhat.com>
34359 PR tree-optimization/79376
34360 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
34362 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
34364 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
34365 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
34366 to simplify split condition.
34368 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34370 * omp-expand.c (oxpand_omp_atomic_fetch_op,
34371 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
34374 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
34376 PR rtl-optimization/68664
34377 * target.def (can_speculate_insn): New hook.
34378 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
34379 * doc/tm.texi: Regenerate.
34380 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
34381 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
34382 (rs6000_sched_can_speculate_insn): New function.
34384 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34386 PR tree-optimization/79284
34387 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
34388 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
34389 vectorizable_mask_load_store, vectorizable_operation,
34390 vect_is_simple_cond, get_same_sized_vectype): Use it instead
34391 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
34392 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
34393 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
34394 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
34395 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
34396 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
34397 is_gimple_assign (stmt). Replace another such test with
34398 is_gimple_assign (stmt).
34400 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34403 * config/avr/avr.c (rtl-iter.h): Include it.
34404 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
34405 (avr_legitimate_combined_insn): ...and implementation.
34407 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
34409 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
34410 * config/s390/s390.c (s390_const_operand_ok)
34411 (s390_canonicalize_comparison, s390_extract_part)
34412 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
34413 (s390_contiguous_bitmask_p, s390_rtx_costs)
34414 (legitimize_pic_address): Likewise.
34415 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
34416 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
34417 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
34418 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
34419 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
34421 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34423 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
34424 REGNO($0) == REGNO($1).
34426 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34428 * config/s390/linux.h(SIZE_TYPE): Add comment.
34430 2017-02-06 Julian Brown <julian@codesourcery.com>
34431 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34432 Virendra Pathak <virendra.pathak@broadcom.com>
34434 * config/aarch64/aarch64-cores.def: Change the scheduler
34436 * config/aarch64/aarch64.md: Include thunderx2t99.md.
34437 * config/aarch64/thunderx2t99.md: New file.
34439 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34441 * doc/standards.texi (Go Language): Update link to language
34444 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34446 * tree-eh.c (lower_resx): Sanitize profile.
34447 (cleanup_empty_eh_move_lp): Likewise.
34449 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34452 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
34454 * cfgloopmanip.h (loop_version): Update prototype.
34455 * modulo-sched.c (sms_schedule): Update call of loop_version.
34456 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
34457 * tree-parloops.c (gen_parallel_loop): Likewise.
34458 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
34459 * tree-ssa-loop-split.c (split_loop): Likewise.
34460 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
34461 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
34463 2017-02-05 Martin Liska <mliska@suse.cz>
34466 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
34468 (print_operand_address): Initialize a struct to zero.
34470 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34472 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
34473 garbage collector only in textual form.
34475 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34477 * doc/extend.texi (x86 specific memory model extensions for
34478 transactional memory): Simplify a phrase.
34480 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
34483 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
34484 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
34485 (atomic_storedi_1): Likewise.
34487 2017-02-04 Jakub Jelinek <jakub@redhat.com>
34489 PR tree-optimization/79338
34490 * tree-parloops.c (gather_scalar_reductions): Don't call
34491 vect_analyze_loop_form for loop->inner before destroying loop's
34494 2017-02-03 Martin Sebor <msebor@redhat.com>
34496 PR tree-optimization/79327
34497 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
34498 when precision has resulted in leading zeros.
34499 (format_integer): Adjust the likely counter to assume an unknown
34500 argument that may be zero is non-zero.
34502 2017-02-03 Jason Merrill <jason@redhat.com>
34505 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
34506 avoid copying non-taken branch.
34508 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34510 PR tree-optimization/79340
34511 * tree-vect-loop.c (vectorizable_reduction): Release
34512 vec_defs elements after safe_splicing them into other vectors.
34515 PR tree-optimization/79327
34516 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
34517 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
34519 (format_integer): Use wide_int_to_tree instead of build_int_cst
34520 + to_?hwi. If argmin is NULL, just set argmin and argmax to
34521 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
34522 of shortest and longest sequence.
34524 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
34526 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
34527 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
34529 2017-02-03 Walter Lee <walt@tilera.com>
34532 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
34533 after initial stackframe link reg save.
34534 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
34536 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34539 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
34540 wu for stxssp alternative.
34542 2017-02-03 Martin Sebor <msebor@redhat.com>
34544 PR tree-optimization/79352
34545 * gimple-fold.c (get_range_strlen): Add argument.
34546 (get_range_strlen): Change return type to bool.
34547 (get_maxval_strlen): Pass in a dummy argument.
34548 * gimple-fold.h (get_range_strlen): Change return type to bool.
34549 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
34550 * tree.h (array_at_struct_end_p): Add argument.
34551 * tree.c (array_at_struct_end_p): Handle it.
34553 2017-02-03 Martin Liska <mliska@suse.cz>
34556 * multiple_target.c (create_dispatcher_calls): Redirect edge
34557 from a caller of a dispatcher.
34558 (expand_target_clones): Make the clones local.
34559 (ipa_target_clone): Do both target clones and resolvers.
34560 (ipa_dispatcher_calls): Remove the pass.
34561 (pass_dispatcher_calls::gate): Likewise.
34562 (make_pass_dispatcher_calls): Likewise.
34563 * passes.def (pass_target_clone): Put as very first IPA early
34566 2017-02-03 Martin Liska <mliska@suse.cz>
34568 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
34569 in case of a function with ifunc attribute.
34571 2017-02-03 Martin Liska <mliska@suse.cz>
34573 * cgraph.c (cgraph_node::dump): Dump function version info.
34574 * symtab.c (symtab_node::dump_base): Add missing new line.
34576 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34578 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
34579 (ifcombine_ifandif): Use it.
34581 2017-02-03 Martin Liska <mliska@suse.cz>
34583 * doc/invoke.texi: Document default value for
34584 use-after-scope-direct-emission-threshold.
34586 2017-02-03 Martin Liska <mliska@suse.cz>
34588 PR tree-optimization/79339
34589 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
34590 (format_floating): Likewise.
34592 2017-02-03 Martin Liska <mliska@suse.cz>
34595 * ipa-prop.c (ipa_node_params_t::insert): Remove current
34597 (ipa_node_params_t::remove): Likewise.
34598 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
34599 initialization from removed ipa_node_params_t::insert.
34600 (ipa_node_params::~ipa_node_params): Move from removed
34601 ipa_node_params_t::release.
34602 * symbol-summary.h (symbol_summary::m_released): New member.
34603 Do not release a summary twice. Do not allow to call finalizer
34604 for types of a summary that live in GGC memory.
34606 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34608 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
34611 2017-02-02 Martin Sebor <msebor@redhat.com>
34613 PR middle-end/79275
34614 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
34615 (format_string): Tighten up the range of output for non-constant
34616 strings and correct the expected range for wide non-constant strings.
34618 2017-02-02 Martin Sebor <msebor@redhat.com>
34620 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
34622 PR middle-end/32003
34623 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
34625 (-fdump-tree-@var): Add to index and document how to come up
34626 with pass-specific option and dump file names.
34627 (-fdump-passes): Clarify where to look for output.
34629 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34631 PR middle-end/77445
34632 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
34633 statistics of the analyzed path; allow threading for speed when
34634 any of BBs along the path are optimized for speed.
34636 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
34638 PR middle-end/78468
34639 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
34640 settings of the virtual registers.
34643 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
34645 * explow.c (get_dynamic_stack_size): Take known alignment of stack
34646 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
34649 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34651 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
34652 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
34654 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34656 * config/s390/s390.md: Add missing comments with the expanded
34658 * config/s390/vector.md: Likewise.
34659 * config/s390/vx-builtins.md: Likewise.
34661 2017-02-02 Jakub Jelinek <jakub@redhat.com>
34664 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
34665 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
34666 conditions on a single line.
34668 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34670 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34671 __S390_VX__ to __VX__.
34673 2017-02-01 Andrew Pinski <apinski@cavium.com>
34675 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
34676 stmt_info to record_stmt_cost.
34677 (vect_get_known_peeling_cost): Pass stmt_info if known to
34679 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
34680 cpu_vector_cost field into
34681 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
34682 field into vec_int_stmt_cost and vec_fp_stmt_cost.
34683 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
34684 splitting of scalar_stmt_cost and vec_stmt_cost.
34685 (thunderx_vector_cost): Likewise.
34686 (cortexa57_vector_cost): LIkewise.
34687 (exynosm1_vector_cost): Likewise.
34688 (xgene1_vector_cost): Likewise.
34689 (thunderx2t99_vector_cost): Improve after the splitting of the two
34691 (aarch64_builtin_vectorization_cost): Update for the splitting of
34692 scalar_stmt_cost and vec_stmt_cost.
34694 2017-02-01 Torvald Riegel <triegel@redhat.com>
34695 Richard Henderson <rth@redhat.com>
34697 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
34698 conditional on existance of a fast atomic load.
34699 * optabs-query.c (can_atomic_load_p): New function.
34700 * optabs-query.h (can_atomic_load_p): Declare it.
34701 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
34702 no fast atomic load is available for the particular size of access.
34703 (expand_atomic_compare_and_swap): Likewise.
34704 (expand_atomic_load): Likewise.
34705 (expand_atomic_store): Likewise.
34706 (expand_atomic_fetch_op): Likewise.
34707 * testsuite/lib/target-supports.exp
34708 (check_effective_target_sync_int_128): Remove x86 because it provides
34709 no fast atomic load.
34710 (check_effective_target_sync_int_128_runtime): Likewise.
34712 2017-02-01 Richard Biener <rguenther@suse.de>
34714 * graphite.c: Include tree-vectorizer.h for find_loop_location.
34715 (graphite_transform_loops): Provide opt-info for optimized nests.
34716 * tree-parloop.c (parallelize_loops): Provide opt-info for
34717 parallelized loops.
34719 2017-02-01 Richard Biener <rguenther@suse.de>
34721 PR middle-end/79315
34722 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
34723 was not set before.
34725 2017-02-01 Richard Biener <rguenther@suse.de>
34727 PR tree-optimization/71824
34728 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34729 Verify the loops are valid in the merged SESE region.
34730 (scop_detection::can_represent_loop_1): Check analyzing the
34731 evolution of the number of iterations in the region succeeds.
34733 2017-01-31 Ian Lance Taylor <iant@golang.org>
34735 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
34736 REG_ARGS_SIZE note to 32-bit push insns and call insn.
34738 2017-01-31 David Malcolm <dmalcolm@redhat.com>
34740 PR preprocessor/79210
34741 * input.c (get_substring_ranges_for_loc): Replace line_width
34742 assertion with error-handling.
34744 2017-01-31 Richard Biener <rguenther@suse.de>
34746 PR tree-optimization/77318
34747 * graphite-sese-to-poly.c (extract_affine): Fix assert.
34748 (create_pw_aff_from_tree): Take loop parameter.
34749 (add_condition_to_pbb): Pass loop of the condition to
34750 create_pw_aff_from_tree.
34752 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34754 * config/s390/s390.c (s390_asan_shadow_offset): New function.
34755 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
34757 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
34761 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
34763 (convert_int_to_float128): Likewise.
34764 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
34765 (convert_int_to_float128): Likewise.
34766 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
34767 (UNSPEC_IEEE128_CONVERT): Likewise.
34768 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
34769 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
34770 Use local variables for IBM extended format.
34771 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
34772 (fix_trunc<mode>si2_fprs): Likewise.
34773 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
34774 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
34775 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
34776 to know that we can now have integers of all sizes in vector
34778 (fix<uns>_<mode>di2_hw): Likewise.
34779 (float<uns>_<mode>si2_hw): Likewise.
34780 (fix_<mode>si2_hw): Likewise.
34781 (fixuns_<mode>si2_hw): Likewise.
34782 (float<uns>_<mode>di2_hw): Likewise.
34783 (float_<mode>di2_hw): Likewise.
34784 (float_<mode>si2_hw): Likewise.
34785 (floatuns_<mode>di2_hw): Likewise.
34786 (floatuns_<mode>si2_hw): Likewise.
34787 (xscvqp<su>wz_<mode>): Delete, no longer used.
34788 (xscvqp<su>dz_<mode>): Likewise.
34789 (xscv<su>dqp_<mode>): Likewise.
34790 (ieee128_mfvsrd_64bit): Likewise.
34791 (ieee128_mfvsrd_32bit): Likewise.
34792 (ieee128_mfvsrwz): Likewise.
34793 (ieee128_mtvsrw): Likewise.
34794 (ieee128_mtvsrd_64bit): Likewise.
34795 (ieee128_mtvsrd_32bit): Likewise.
34797 2017-01-31 Martin Liska <mliska@suse.cz>
34800 * ipa-prop.c (ipa_free_all_node_params): Call release method
34801 instead of ~sumbol_summary to not to trigger double times
34804 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
34806 PR tree-optimization/71691
34807 * bitmap.h (class auto_bitmap): New.
34808 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
34809 is_maybe_undefined instead of ssa_undefined_value_p.
34811 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34813 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34814 __S390_ARCH_LEVEL__ to __ARCH__.
34816 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34818 PR tree-optimization/79267
34819 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
34820 if should_remove_lhs_p is true.
34822 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
34825 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
34826 (add_alignment_attribute): New.
34827 (base_type_die): Add alignment attribute.
34828 (subrange_type_die): Likewise.
34829 (modified_type_die): Likewise.
34830 (gen_array_type_die): Likewise.
34831 (gen_descr_array_type_die: Likewise.
34832 (gen_enumeration_type_die): Likewise.
34833 (gen_subprogram_die): Likewise.
34834 (gen_variable_die): Likewise.
34835 (gen_field_die): Likewise.
34836 (gen_ptr_to_mbr_type_die): Likewise.
34837 (gen_struct_or_union_type_die): Likewise.
34838 (gen_subroutine_type_die): Likewise.
34839 (gen_typedef_die): Likewise.
34840 (base_type_cmp): Compare alignment attribute.
34842 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34845 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
34846 (setb_unsigned) New pattern for setb with CCUNS.
34847 * config/rs6000/rs6000.c (expand_block_compare): Use a different
34848 subfc./subfe sequence to avoid overflow problems. Generate a
34849 shorter sequence with cmpld/setb for power9.
34850 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
34851 for generating subfc. instruction.
34852 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
34853 now uses this instruction.
34855 2017-01-30 Ian Lance Taylor <iant@google.com>
34858 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
34859 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
34861 2017-01-30 Martin Sebor <msebor@redhat.com>
34863 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
34864 Move constant to the right of a relational operator.
34865 (get_mpfr_format_length, format_character, format_string): Ditto.
34866 (should_warn_p, maybe_warn): Same.
34868 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
34870 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
34873 * asan.c (get_translation_unit_decl): Remove function.
34874 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
34876 2017-01-30 Martin Liska <mliska@suse.cz>
34878 PR gcov-profile/79259
34879 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
34880 -fprofile-generate.
34882 2017-01-30 Martin Liska <mliska@suse.cz>
34885 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
34886 Initialize variables with NULL value.
34888 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
34891 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
34893 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
34895 2017-01-30 Richard Biener <rguenther@suse.de>
34897 PR tree-optimization/79276
34898 * tree-vrp.c (process_assert_insertions): Properly adjust common
34899 when removing a duplicate.
34901 2017-01-30 Richard Biener <rguenther@suse.de>
34903 PR tree-optimization/79256
34904 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34905 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34907 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
34909 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
34912 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
34913 ("*r<noxa>sbg_<mode>_sll_bitmask")
34914 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
34915 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
34916 Use contiguous_bitmask_nowrap_operand.
34918 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34921 * config/rs6000/altivec.h (vec_xl): Revise #define.
34922 (vec_xst): Likewise.
34924 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
34926 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
34928 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
34930 PR rtl-optimization/79194
34931 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
34932 traps before call to bypass_conditional_jumps.
34934 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34936 PR tree-optimization/71374
34937 * lra-constraints.c (check_conflict_input_operands): New.
34938 (match_reload): Use it.
34940 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34943 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
34944 account to calculate conflict_set.
34946 2017-01-27 Bin Cheng <bin.cheng@arm.com>
34948 PR rtl-optimization/78559
34949 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
34950 other_insn in combine.
34952 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
34954 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
34955 uint16_type_node for BT_UINT16.
34957 2017-01-27 David Malcolm <dmalcolm@redhat.com>
34959 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
34960 "RTL Tests" to menu.
34961 (GIMPLE Tests): New node.
34962 (RTL Tests): New node.
34964 2017-01-27 Richard Biener <rguenther@suse.de>
34966 PR tree-optimization/79245
34967 * tree-loop-distribution.c (distribute_loop): Apply cost
34968 modeling also to detected patterns.
34970 2017-01-27 Richard Biener <rguenther@suse.de>
34972 PR tree-optimization/71433
34973 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
34974 (compare_assert_loc): New function.
34975 (process_assert_insertions): Sort and optimize assert locations
34976 to remove duplicates and push down identical assertions on
34977 edges to their destination block.
34979 2017-01-27 Richard Biener <rguenther@suse.de>
34981 PR tree-optimization/79244
34982 * tree-vrp.c (remove_range_assertions): Forcefully propagate
34983 out SSA names even if abnormal.
34985 2017-01-27 Jakub Jelinek <jakub@redhat.com>
34987 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
34988 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
34989 instead of MPFR_RNDN.
34991 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
34994 * arm.c (arm_option_override): Don't call build_target_option_node
34995 until after doing all option overrides.
34996 (arm_valid_target_attribute_tree): Likewise.
34998 2017-01-27 Martin Liska <mliska@suse.cz>
35000 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35001 for {cd}tors and C++ {cd}tors.
35003 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35005 * config/s390/s390.md ("*setmem_long_and")
35006 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35008 2017-01-26 Martin Sebor <msebor@redhat.com>
35010 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35013 2017-01-26 Martin Sebor <msebor@redhat.com>
35015 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35016 HAVE_DFmode before using XFmode or DFmode.
35017 (parse_directive): Avoid using the z length modifier to avoid
35018 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
35020 PR middle-end/78703
35021 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
35022 to accept adjustment as an array.
35023 (get_int_range): New function.
35024 (struct directive): Make width and prec arrays.
35025 (directive::set_width, directive::set_precision): Call get_int_range.
35026 (format_integer, format_floating): Handle width and precision ranges.
35027 (format_string, parse_directive): Same.
35029 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35032 * dwarf2out.c (generate_skeleton_bottom_up): For children with
35033 comdat_type_p set, just clone them, but keep the children in the
35037 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
35038 which have direct callers with -fvar-tracking-assignments enabled
35040 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
35041 inside of type units.
35043 2017-01-26 Martin Sebor <msebor@redhat.com>
35045 PR middle-end/78703
35046 * gimple-ssa-sprintf.c (struct result_range): Add likely and
35048 (struct format_result): Replace number_chars, number_chars_min,
35049 and number_chars_max with a single member of struct result_range.
35051 (format_result::operator+=): Adjust.
35052 (struct fmtresult): Remove bounded. Handle likely and unlikely
35054 (fmtresult::adjust_for_width_or_precision): New function.
35055 (fmtresult:type_max_digits): New function.
35056 (bytes_remaining): Handle likely and unlikely counters.
35057 (min_bytes_remaining): Remove.
35058 (format_percent): Simplify.
35059 (format_integer, format_floating): Set likely and unlikely counters.
35060 (get_string_length, format_character, format_string): Same.
35061 (format_plain, should_warn_p): New function.
35062 (maybe_warn): Call should_warn_p. Update diagnostic messages
35063 and handle those for all directives, including plain strings.
35064 (format_directive): Handle likely and unlikely counters.
35065 Remove unnecessary quoting from diagnostics. Add an informational
35067 (add_bytes): Remove.
35068 (pass_sprintf_length::compute_format_length): Simplify.
35069 (try_substitute_return_value): Handle likely and unlikely counters.
35071 2017-01-26 Carl Love <cel@us.ibm.com>
35073 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
35074 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
35076 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
35079 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
35080 endianess for subregs into account.
35081 * lra-constraints.c (lra_constraints): Do risky transformations
35082 always on the first iteration.
35083 * lra-lives.c (check_pseudos_live_through_calls): Add arg
35084 last_call_used_reg_set.
35085 (process_bb_lives): Define and use last_call_used_reg_set.
35086 * lra.c (lra): Always continue after lra_constraints on the first
35089 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
35091 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
35093 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
35095 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35097 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
35098 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
35099 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
35100 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
35101 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
35102 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
35103 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
35104 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
35105 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
35107 2017-01-26 Marek Polacek <polacek@redhat.com>
35110 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
35111 for the third operand.
35113 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35115 PR middle-end/79236
35116 * omp-low.c (struct omp_context): Add simt_stmt field.
35117 (scan_omp_for): Return omp_context *.
35118 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
35119 context to the _simt_ SIMD stmt.
35120 (lower_omp_for): For combined SIMD with sibling _simt_
35121 SIMD, make sure to use the same decls in _looptemp_
35122 clauses as in the sibling.
35124 2017-01-26 David Sherwood <david.sherwood@arm.com>
35126 PR middle-end/79212
35127 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
35130 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35133 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
35134 emit fld b; fld a; if possible.
35136 * brig-builtins.def: Update copyright years.
35137 * config/arm/arm_acle_builtins.def: Update copyright years.
35139 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
35142 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
35143 constraint instead of o for the stxsd instruction.
35145 2017-01-25 Carl Love <cel@us.ibm.com>
35147 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
35148 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
35150 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
35152 * doc/invoke.texi (C++ Dialect Options): Fix typo.
35154 2017-01-25 Richard Biener <rguenther@suse.de>
35156 PR tree-optimization/69264
35157 * target.def (vector_alignment_reachable): Improve documentation.
35158 * doc/tm.texi: Regenerate.
35159 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
35161 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
35162 earlier changes with respect to TYPE_USER_ALIGN.
35163 (vector_alignment_reachable_p): Likewise. Improve dumping.
35165 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35168 * config/arm/arm.md (xordi3): Force constant operand into a register
35171 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35173 * doc/invoke.texi (-fstore-merging): Correct default optimization
35174 levels at which it is enabled.
35175 (-O): Move -fstore-merging from list to...
35178 2017-01-25 Richard Biener <rguenther@suse.de>
35181 * omp-expand.c: Include debug.h.
35182 (expand_omp_taskreg): Make sure to generate early debug before
35183 outlining anything from a function.
35184 (expand_omp_target): Likewise.
35185 (grid_expand_target_grid_body): Likewise.
35187 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
35190 * asan.c (get_translation_unit_decl): New function.
35191 (asan_add_global): Extract modules file name from globals
35192 TRANSLATION_UNIT_DECL name.
35194 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
35197 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
35198 for long calls with APCS frame and VFP.
35200 2017-01-24 David Malcolm <dmalcolm@redhat.com>
35202 * cfg.c (original_copy_tables_initialized_p): New function.
35203 * cfg.h (original_copy_tables_initialized_p): New decl.
35204 * cfgrtl.c (relink_block_chain): Guard the call to
35205 free_original_copy_tables with a call to
35206 original_copy_tables_initialized_p.
35207 * cgraph.h (symtab_node::native_rtl_p): New decl.
35208 * cgraphunit.c (symtab_node::native_rtl_p): New function.
35209 (symtab_node::needed_p): Don't assert for early assembly output
35210 for __RTL functions.
35211 (cgraph_node::finalize_function): Set "force_output" for __RTL
35213 (cgraph_node::analyze): Bail out early for __RTL functions.
35214 (analyze_functions): Update assertion to support __RTL functions.
35215 (cgraph_node::expand): Bail out early for __RTL functions.
35216 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
35218 * function.h (struct function): Update comment for field
35220 * gimple-expr.c: Include "tree-pass.h".
35221 (gimple_has_body_p): Return false for __RTL functions.
35222 * Makefile.in (OBJS): Add run-rtl-passes.o.
35223 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
35225 (gcc::pass_manager::get_clean_slate): New accessor.
35226 * passes.c: Include "insn-addr.h".
35227 (should_skip_pass_p): Add logging. Update logic for running
35228 "expand" to be compatible with both __GIMPLE and __RTL. Guard
35229 property-provider override so it is only done for gimple passes.
35231 (skip_pass): New function.
35232 (execute_one_pass): Call skip_pass when skipping passes.
35233 * read-md.c (md_reader::read_char): Support filtering
35234 the input to a subset of line numbers.
35235 (md_reader::md_reader): Initialize fields
35236 m_first_line and m_last_line.
35237 (md_reader::read_file_fragment): New function.
35238 * read-md.h (md_reader::read_file_fragment): New decl.
35239 (md_reader::m_first_line): New field.
35240 (md_reader::m_last_line): New field.
35241 * read-rtl-function.c (function_reader::create_function): Only
35242 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
35243 curr_properties. Set DECL_INITIAL to a dummy block.
35244 (read_rtl_function_body_from_file_range): New function.
35245 * read-rtl-function.h (read_rtl_function_body_from_file_range):
35247 * run-rtl-passes.c: New file.
35248 * run-rtl-passes.h: New file.
35250 2017-01-24 Jeff Law <law@redhat.com>
35252 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
35255 2017-01-24 Bin Cheng <bin.cheng@arm.com>
35257 PR tree-optimization/79159
35258 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
35259 (record_nonwrapping_iv): Improve boundary using above function if no
35260 value range information.
35262 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
35263 Martin Jambor <mjambor@suse.cz>
35265 * brig-builtins.def: New file.
35266 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
35267 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
35268 (DEF_HSAIL_SAT_BUILTIN): Likewise.
35269 (DEF_HSAIL_INTR_BUILTIN): Likewise.
35270 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
35271 * builtin-types.def (BT_INT8): New.
35272 (BT_INT16): Likewise.
35273 (BT_UINT8): Likewise.
35274 (BT_UINT16): Likewise.
35275 (BT_FN_ULONG): Likewise.
35276 (BT_FN_UINT_INT): Likewise.
35277 (BT_FN_UINT_ULONG): Likewise.
35278 (BT_FN_UINT_LONG): Likewise.
35279 (BT_FN_UINT_PTR): Likewise.
35280 (BT_FN_ULONG_PTR): Likewise.
35281 (BT_FN_INT8_FLOAT): Likewise.
35282 (BT_FN_INT16_FLOAT): Likewise.
35283 (BT_FN_UINT32_FLOAT): Likewise.
35284 (BT_FN_UINT16_FLOAT): Likewise.
35285 (BT_FN_UINT8_FLOAT): Likewise.
35286 (BT_FN_UINT64_FLOAT): Likewise.
35287 (BT_FN_UINT16_UINT32): Likewise.
35288 (BT_FN_UINT32_UINT16): Likewise.
35289 (BT_FN_UINT16_UINT16_UINT16): Likewise.
35290 (BT_FN_INT_PTR_INT): Likewise.
35291 (BT_FN_UINT_PTR_UINT): Likewise.
35292 (BT_FN_LONG_PTR_LONG): Likewise.
35293 (BT_FN_ULONG_PTR_ULONG): Likewise.
35294 (BT_FN_VOID_UINT64_UINT64): Likewise.
35295 (BT_FN_UINT8_UINT8_UINT8): Likewise.
35296 (BT_FN_INT8_INT8_INT8): Likewise.
35297 (BT_FN_INT16_INT16_INT16): Likewise.
35298 (BT_FN_INT_INT_INT): Likewise.
35299 (BT_FN_UINT_FLOAT_UINT): Likewise.
35300 (BT_FN_FLOAT_UINT_UINT): Likewise.
35301 (BT_FN_ULONG_UINT_UINT): Likewise.
35302 (BT_FN_ULONG_UINT_PTR): Likewise.
35303 (BT_FN_ULONG_ULONG_ULONG): Likewise.
35304 (BT_FN_UINT_UINT_UINT): Likewise.
35305 (BT_FN_VOID_UINT_PTR): Likewise.
35306 (BT_FN_UINT_UINT_PTR: Likewise.
35307 (BT_FN_UINT32_UINT64_PTR): Likewise.
35308 (BT_FN_INT_INT_UINT_UINT): Likewise.
35309 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
35310 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
35311 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
35312 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
35313 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
35314 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
35315 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
35316 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
35317 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
35318 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
35319 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
35320 * doc/frontends.texi: List BRIG FE.
35321 * doc/install.texi (Testing): Add BRIG tesring requirements.
35322 * doc/invoke.texi (Overall Options): Mention BRIG.
35323 * doc/standards.texi (Standards): Doucment BRIG HSA version.
35325 2017-01-24 Richard Biener <rguenther@suse.de>
35327 PR translation/79208
35328 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
35330 2017-01-24 Martin Jambor <mjambor@suse.cz>
35333 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
35334 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
35335 and known_contexts.
35337 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
35339 PR middle-end/79123
35340 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
35341 casts from signed to unsigned really don't have a range.
35343 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
35345 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
35346 GMP_RNDx for compatiblity.
35348 2017-01-24 Martin Liska <mliska@suse.cz>
35351 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
35352 that would prevent us to call alloca with -1 as argument.
35354 2017-01-24 Jakub Jelinek <jakub@redhat.com>
35356 * dwarf2out.c (output_compilation_unit_header, output_file_names):
35357 Avoid -Wformat-security warning.
35359 2017-01-23 Andrew Pinski <apinski@cavium.com>
35361 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
35364 2017-01-23 Martin Sebor <msebor@redhat.com>
35366 PR middle-end/78703
35367 * gimple-ssa-sprintf.c (warn_level): New global.
35368 (format_integer): Use it here and throughout the rest of the file.
35369 Use the same switch to compute sign as base.
35370 (maybe_warn): New function.
35371 (format_directive): Factor out warnings into maybe_warn.
35372 Add debugging output. Use warn_level.
35373 (add_bytes): Use warn_level.
35374 (pass_sprintf_length::compute_format_length): Add debugging output.
35375 (try_substitute_return_value): Same.
35376 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
35378 PR middle-end/78703
35379 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
35380 (struct fmtresult, format_integer, format_floating): Adjust.
35381 (fmtresult::fmtresult): Set max correctly in two argument ctor.
35382 (get_string_length, format_string,format_directive): Same.
35383 (pass_sprintf_length::compute_format_length): Same.
35384 (try_substitute_return_value): Simplify slightly.
35386 PR middle-end/78703
35387 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
35388 (fmtresult::operator+=): Outlined.
35389 (struct fmtresult): Add ctors.
35390 (struct conversion_spec): Rename...
35391 (struct directive): ...to this. Add and remove data members.
35392 (directive::set_width, directive::set_precision): New functions.
35393 (format_percent): Use fmtresult ctor.
35394 (get_width_and_precision): Remove.
35395 (format_integer): Make naming changes. Avoid computing width and
35397 (format_floating): Same. Adjust indentation.
35398 (format_character, format_none): New functions.
35399 (format_string): Moved character handling to format_character.
35400 (format_directive): Remove arguments, change return type.
35401 (parse_directive): New function.
35402 (pass_sprintf_length::compute_format_length): Move directive
35403 parsing to parse_directive.
35405 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35407 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
35408 (assign_assembler_name_if_needed): ... this.
35409 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
35410 (assign_assembler_name_if_needed): ... this.
35411 (free_lang_data_in_cgraph): Adjust callers.
35412 * cgraphunit.c (cgraph_node::analyze): Likewise.
35413 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
35416 2017-01-23 Richard Biener <rguenther@suse.de>
35418 PR tree-optimization/79088
35419 PR tree-optimization/79188
35420 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
35421 resetting loop bounds after last path deletion. Reset loop
35422 bounds of the target loop, make code match the comments.
35423 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
35424 Make sure loops need no fixups.
35426 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35428 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
35429 exponent support with double type for first argument.
35430 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
35431 type returned by __builtin_vec_extract_sig,
35432 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
35433 functions from "vector int" to "vector unsigned int" or from
35434 "vector long long int" to "vector unsigned long long int".
35435 Changed type returned by __builtin_vec_extract_exp,
35436 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
35437 functions from "vector int" to "vector unsigned int" or from
35438 "vector long long int" to "vector unsigned long long int".
35439 Changed return type of __builtin_vec_test_data_class,
35440 __builtin_vec_test_data_class_sp, and
35441 __builtin_vec_test_data_class_dp from "vector int" to
35442 "vector bool int" or from "vector long long int" to "vector bool
35443 long long int" and changed second argument type from "unsigned
35444 int" to "int". Added new overloaded function forms "vector float
35445 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
35446 "vector float __builtin_vec_insert_exp_sp (vector float, vector
35447 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
35448 double, vector unsigned long long int)" and "vector double
35449 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
35450 long int)". Changed return type of
35451 __builtin_scalar_test_data_class and
35452 __builtin_scalar_test_data_class_sp and
35453 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
35454 int" and changed second argument from "unsigned int" to "int".
35455 Changed type returned by __builtin_scalar_test_neg,
35456 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
35457 from "int" to "bool int". Added new overloaded function form
35458 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
35459 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
35460 exponent double-precision with floating point first argument.
35461 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
35462 documentation of scalar_test_data_class, scalar_test_neg,
35463 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
35464 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
35465 vec_test_data_class built-in functions to reflect refinements in
35466 their type signatures.
35468 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
35470 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
35472 (aarch64_elf_asm_destructor): Likewise.
35474 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
35476 PR rtl-optimization/78634
35477 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
35478 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
35479 * ifcvt.c (noce_try_cmove): Add missing cost check.
35481 PR rtl-optimization/71724
35482 * combine.c (if_then_else_cond): Look for situations where it is
35483 beneficial to undo the work of one of the recursive calls.
35485 2017-01-23 Bin Cheng <bin.cheng@arm.com>
35487 PR tree-optimization/70754
35488 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
35489 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
35490 combined stmt before it if not NULL.
35491 (combine_chains): Process refs reversely and compute dominance point
35494 2017-01-23 Martin Liska <mliska@suse.cz>
35496 PR tree-optimization/79196
35497 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
35498 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
35500 (strlen_optimize_stmt): Call the renamed function.
35502 2017-01-23 Michael Matz <matz@suse.de>
35504 PR tree-optimization/78384
35505 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
35507 2017-01-23 Richard Biener <rguenther@suse.de>
35509 PR tree-optimization/79186
35510 * tree-vrp.c (register_new_assert_for): Make sure we've seen
35511 both incoming edges before moving an assert.
35513 2017-01-23 Martin Jambor <mjambor@suse.cz>
35515 * ipa-prop.c (load_from_param_1): Removed.
35516 (load_from_unmodified_param): Bits from load_from_param_1 put back
35518 (load_from_param): Removed.
35519 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
35520 with stmt. Reverted back to use of load_from_unmodified_param.
35522 2017-01-23 Martin Jambor <mjambor@suse.cz>
35525 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
35526 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
35527 field a pointer to garbage collected vector, mark lattices and
35528 ipcp_orig_node with GTY((skip)).
35529 (ipa_get_param_count): Adjust to descriptors being a pointer.
35530 (ipa_get_param): Likewise.
35531 (ipa_get_type): Likewise.
35532 (ipa_get_param_move_cost): Likewise.
35533 (ipa_set_param_used): Likewise.
35534 (ipa_get_controlled_uses): Likewise.
35535 (ipa_set_controlled_uses): Likewise.
35536 (ipa_is_param_used): Likewise.
35537 (ipa_node_params_t): Move into garbage collector. New methods insert
35539 (ipa_node_params_sum): Annotate wth GTY(()).
35540 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
35542 (ipa_load_from_parm_agg): Adjust declaration.
35543 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
35544 * ipa-profile.c (ipa_profile): Likewise.
35545 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
35546 (ipa_populate_param_decls): Make descriptors parameter garbage
35548 (ipa_dump_param): Adjust to descriptors being a pointer.
35549 (ipa_alloc_node_params): Likewise.
35550 (ipa_initialize_node_params): Likewise.
35551 (load_from_param_1): Make descriptors parameter garbage collected.
35552 (load_from_unmodified_param): Likewise.
35553 (load_from_param): Likewise.
35554 (ipa_load_from_parm_agg): Likewise.
35555 (ipa_node_params::~ipa_node_params): Removed.
35556 (ipa_free_all_node_params): Remove call to delete operator.
35557 (ipa_node_params_t::insert): New.
35558 (ipa_node_params_t::remove): Likewise.
35559 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
35560 copy known_csts and known_contexts vectors.
35561 (ipa_read_node_info): Adjust to descriptors being a pointer.
35562 (ipcp_modif_dom_walker): Make m_descriptors field garbage
35564 (ipcp_transform_function): Make descriptors variable garbage
35567 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
35569 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
35570 * config/i386/avx512dqintrin.h: Ditto.
35571 * config/i386/avx512fintrin.h: Ditto.
35572 * config/i386/i386.c: Handle new builtins.
35573 * config/i386/i386-builtin.def: Add new builtins.
35574 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
35575 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
35577 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35578 Martin Liska <mliska@suse.cz>
35580 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
35581 * asan.c (asan_expand_poison_ifn): Support stores and use
35582 appropriate ASAN report function.
35583 * internal-fn.c (expand_ASAN_POISON_USE): New function.
35584 * internal-fn.def (ASAN_POISON_USE): Declare.
35585 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
35586 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
35587 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
35588 ASAN_POISON calls w/o LHS.
35589 * tree-ssa.c (execute_update_addresses_taken): Create clobber
35590 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
35591 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
35592 * gimplify.c (asan_poison_variables): Add attribute
35593 use_after_scope_memory to variables that really needs to live
35595 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
35596 having the attribute.
35598 2017-01-23 Martin Liska <mliska@suse.cz>
35600 * asan.c (create_asan_shadow_var): New function.
35601 (asan_expand_poison_ifn): Likewise.
35602 * asan.h (asan_expand_poison_ifn): New declaration.
35603 * internal-fn.c (expand_ASAN_POISON): Likewise.
35604 * internal-fn.def (ASAN_POISON): New builtin.
35605 * sanopt.c (pass_sanopt::execute): Expand
35606 asan_expand_poison_ifn.
35607 * tree-inline.c (copy_decl_for_dup_finish): Make function
35609 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
35610 * tree-ssa.c (is_asan_mark_p): New function.
35611 (execute_update_addresses_taken): Rewrite local variables
35612 (identified just by use-after-scope as addressable) into SSA.
35614 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35616 * doc/install.texi (Specific): opensource.apple.com uses https
35617 now. Remove trailing slash.
35619 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35621 * README.Portability: Remove note on an Irix compatibility issue.
35623 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
35625 * gcov.c (INCLUDE_ALGORITHM): Define.
35626 (INCLUDE_VECTOR): Define.
35627 No longer include <vector> and <algorithm> directly.
35629 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
35631 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
35633 * doc/invoke.texi (Code Gen Options): Ditto.
35635 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
35638 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
35640 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
35642 rtl-optimization/79125
35643 * cprop.c (local_cprop_pass): Handle cases where we make an
35644 unconditional trap.
35646 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
35650 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
35651 read from, for big endian.
35653 2017-01-20 Jiong Wang <jiong.wang@arm.com>
35655 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
35656 register pauth builtins for LP64 only.
35658 2017-01-20 Marek Polacek <polacek@redhat.com>
35661 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
35664 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35666 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
35668 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
35669 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
35670 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
35672 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35675 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
35676 in big-endian mode when they are not a single duplicated value.
35678 2017-01-20 Richard Biener <rguenther@suse.de>
35680 * BASE-VER: Bump to 7.0.1.
35682 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35684 * omp-low.c (omplow_simd_context): New struct. Use it...
35685 (lower_rec_simd_input_clauses): ...here and...
35686 (lower_rec_input_clauses): ...here to hold common data. Adjust all
35687 references to idx, lane, max_vf, is_simt.
35689 2017-01-20 Graham Markall <graham.markall@embecosm.com>
35691 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
35694 2017-01-20 Martin Jambor <mjambor@suse.cz>
35696 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
35697 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
35699 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
35700 (GTFILES): Rename hsa.c to hsa-common.c.
35701 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
35702 * hsa-dump.c: Likewise.
35703 * hsa-gen.c: Likewise.
35704 * hsa-regalloc.c: Likewise.
35705 * ipa-hsa.c: Likewise.
35706 * omp-expand.c: Likewise.
35707 * omp-low.c: Likewise.
35708 * toplev.c: Likewise.
35710 2017-01-20 Marek Polacek <polacek@redhat.com>
35713 * doc/invoke.texi: Document -Wduplicated-branches.
35714 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
35715 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
35716 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
35717 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
35718 return 0 only when not OEP_LEXICOGRAPHIC.
35719 (fold_build_cleanup_point_expr): Use the expression
35720 location when building CLEANUP_POINT_EXPR.
35721 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
35722 * tree.c (add_expr): Handle error_mark_node.
35724 2017-01-20 Martin Liska <mliska@suse.cz>
35727 * tree-profile.c (init_ic_make_global_vars): Do not call
35729 (gimple_init_gcov_profiler): Likewise.
35731 2017-01-20 Martin Liska <mliska@suse.cz>
35734 * cgraph.h (maybe_create_reference): Remove argument and
35736 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
35738 * ipa-cp.c (create_specialized_node): Likewise.
35739 * symtab.c (symtab_node::maybe_create_reference): Handle
35740 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
35742 2017-01-20 Martin Liska <mliska@suse.cz>
35744 * read-rtl-function.c (function_reader::create_function): Use
35745 build_decl instread of build_decl_stat.
35747 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
35749 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
35750 * config/i386/avx512dqintrin.h: Ditto.
35751 * config/i386/avx512fintrin.h: Ditto.
35752 * config/i386/i386-builtin-types.def: Add new types.
35753 * config/i386/i386.c: Handle new types.
35754 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
35755 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
35756 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
35757 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
35758 (__builtin_ia32_kshiftridi): New.
35759 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
35761 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
35765 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
35766 define to rs6000_init_stack_protect_guard.
35767 (rs6000_init_stack_protect_guard): New function.
35769 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35770 Yunqiang Su <yunqiang.su@imgtec.com>
35772 * config.gcc (supported_defaults): Add madd4.
35773 (with_madd4): Add validation.
35774 (all_defaults): Add madd4.
35775 * config/mips/mips.opt (mmadd4): New option.
35776 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35778 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35780 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
35781 (ISA_HAS_FUSED_MADD4): Likewise.
35782 * doc/invoke.texi (-mmadd4): Document the new option.
35783 * doc/install.texi (--with-madd4): Document the new option.
35785 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35787 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
35788 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
35789 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
35790 (aarch64_init_pauth_hint_builtins): New.
35791 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
35792 (aarch64_expand_builtin): Expand new builtins.
35794 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35796 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
35797 * combine-stack-adj.c (no_unhandled_cfa): Handle
35798 REG_CFA_TOGGLE_RA_MANGLE.
35799 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
35800 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
35801 info for return address signing.
35802 (aarch64_expand_epilogue): Likewise.
35804 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35806 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
35807 * config/aarch64/aarch64-protos.h
35808 (aarch64_return_address_signing_enabled): New declaration.
35809 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
35811 (aarch64_expand_prologue): Sign return address before it's pushed onto
35813 (aarch64_expand_epilogue): Authenticate return address fetched from
35815 (aarch64_override_options): Sanity check for ILP32 and ISA level.
35816 (aarch64_attributes): New function attributes for "sign-return-address".
35817 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
35818 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
35819 ("*do_return"): Generate combined instructions according to key index.
35820 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
35821 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
35823 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
35824 * config/aarch64/aarch64.opt (msign-return-address=): New.
35825 * doc/extend.texi (AArch64 Function Attributes): Documents
35826 "sign-return-address=".
35827 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
35829 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35831 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
35832 overall option summary.
35834 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35836 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
35837 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
35838 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
35839 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
35841 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
35843 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
35844 -mpower9-minmax by default for -mcpu=power9.
35845 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
35846 128-bit floating point.
35848 2017-01-20 Alan Modra <amodra@gmail.com>
35850 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
35851 optimizing for size.
35853 2017-01-20 Alan Modra <amodra@gmail.com>
35856 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
35857 for strcmp and strncmp from corresponding builtin decl.
35859 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35861 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
35862 instead of i386/rtems-64.h.
35863 * config/i386/rtems-64.h: Remove.
35865 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35869 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
35871 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
35873 2017-01-19 Tamar Christina <tamar.christina@arm.com>
35875 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
35876 Change int to HOST_WIDE_INT.
35877 * config/aarch64/aarch64-protos.h
35878 (aarch64_simd_gen_const_vector_dup): Likewise.
35879 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
35881 2017-01-19 David Malcolm <dmalcolm@redhat.com>
35883 * langhooks-def.h (lhd_type_for_size): New decl.
35884 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
35885 * langhooks.c (lhd_type_for_size): New function, taken from
35888 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
35890 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
35891 define_bypass for CR latency.
35892 (power9-cracked-alu): Update bypass latency and remove power9-branch.
35893 (power9-alu2): Add define_bypass for CR latency.
35895 (power9-mul): Update insn latency.
35896 (power9-mul-compare): Update insn latency, bypass latency and remove
35899 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35901 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
35903 * config/aarch64/aarch64.md
35904 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
35905 aarch64_nopcrelative_literal_loads.
35906 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
35908 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
35910 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
35911 TARGET_LOONGSON_3A.
35912 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
35914 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
35917 * config.gcc (supported_defaults): Add lxc1-sxc1.
35918 (with_lxc1_sxc1): Add validation.
35919 (all_defaults): Add lxc1-sxc1.
35920 * config/mips/mips.opt (mlxc1-sxc1): New option.
35921 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35923 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35924 __mips_no_lxc1_sxc1.
35925 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
35926 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
35927 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
35929 2017-01-19 Richard Biener <rguenther@suse.de>
35931 PR tree-optimization/72488
35932 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
35933 sure to restore SSA info.
35934 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
35936 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
35938 PR rtl-optimization/79121
35939 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
35940 of the inner type when shifting an extended value.
35942 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35945 * symtab.c (symtab_node::equal_address_to): Fix comparing of
35946 interposable aliases.
35948 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
35951 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
35952 Use the evmergelohi instruction.
35953 (mov_si<mode>_e500_subreg4_2_le): Likewise.
35954 (mov_sitf_e500_subreg8_2_be): Likewise.
35955 (mov_sitf_e500_subreg12_2_le): Likewise.
35956 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
35957 (mov_si<mode>_e500_subreg4_2_be): Likewise.
35958 (mov_sitf_e500_subreg8_2_le): Likewise.
35959 (mov_sitf_e500_subreg12_2_be): Likewise.
35961 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35963 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
35964 attribute from vecsimple to vecperm.
35965 (altivec_vbpermq2): Likewise.
35967 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35970 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
35972 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35973 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
35974 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
35975 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
35976 case where N arg is SIZE_MAX.
35977 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
35978 (cmpstrsi): Add pattern.
35980 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
35982 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
35983 __builtin_vec_revb builtins.
35984 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
35985 built-in functions to support generation of the ISA 3.0 XXBR<x>
35986 vector byte reverse instructions.
35987 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
35988 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
35989 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
35990 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
35991 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
35992 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
35993 (P9V_BUILTIN_VEC_REVB): Likewise.
35994 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
35995 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
35996 (p9_xxbrq_v16qi): Likewise.
35997 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
35998 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
35999 (p9_xxbrh_v8hi): Likewise.
36000 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36001 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36002 vec_revb built-in functions.
36004 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36006 PR rtl-optimization/78952
36007 * config/i386/i386.md (any_extract): New code iterator.
36008 (*insvqi_2): Use any_extract for source operand.
36009 (*insvqi_3): Use any_shiftrt for source operand.
36011 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36013 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36015 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36017 2017-01-18 Matthias Klose <doko@ubuntu.com>
36019 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
36021 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36023 * config/rs6000/altivec.h (vec_bperm): Change #define.
36024 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
36025 (altivec_vbpermq2): New define_insn.
36026 (altivec_vbpermd): Likewise.
36027 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
36028 function interface.
36029 (VBPERMD): Likewise.
36030 (VBPERM): New polymorphic function interface.
36031 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
36032 Add entries for P9V_BUILTIN_VEC_VBPERM.
36033 * doc/extend.texi: Add interfaces for vec_bperm.
36035 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36037 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
36038 first letter of error messages.
36039 (s390_resolve_overloaded_builtin): Likewise.
36040 * config/s390/s390.c (s390_expand_builtin): Likewise.
36041 (s390_invalid_arg_for_unprototyped_fn): Likewise.
36042 (s390_valid_target_attribute_inner_p): Likewise.
36043 * config/s390/s390.md ("tabort"): Likewise.
36045 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
36047 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
36048 (ISA_AVOID_DIV_HILO): New macro.
36049 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
36050 (ISA_HAS_DDIV): Likewise.
36052 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36054 * doc/invoke.texi (fabi-version): Correct number of occurrences.
36056 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36058 * doc/invoke.texi (fabi-version): Spelling fix.
36060 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36063 * doc/invoke.texi (fabi-version): Mention mangling fix for
36066 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36069 * doc/invoke.texi (fabi-version): Document discriminator mangling.
36071 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
36074 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
36075 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
36077 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
36079 (stack_protect_test): Ditto.
36080 * config/rs6000/rs6000.opt (mstack-protector-guard=,
36081 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
36083 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
36084 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
36085 -mstack-protector-guard-offset=.
36086 (RS/6000 and PowerPC Options): Ditto.
36088 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36090 * config/i386/i386.h (MASK_CLASS_P): New define.
36091 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
36092 there are no registers from different register sets also when
36093 mask registers are used. Update function comment.
36094 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
36095 to (*k/*r) and (*k/*km) alternatives.
36097 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
36099 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
36100 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
36101 (EH_RETURN_HANDLER_RTX): New define.
36102 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
36103 Force frame pointer in EH return functions.
36104 (aarch64_expand_epilogue): Add barrier for eh_return.
36105 (aarch64_final_eh_return_addr): Remove.
36106 (aarch64_eh_return_handler_rtx): New function.
36107 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
36109 (aarch64_eh_return_handler_rtx): New prototype.
36111 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36113 * config/rs6000/altivec.h (vec_rlmi): New #define.
36114 (vec_vrlnm): Likewise.
36115 (vec_rlnm): Likewise.
36116 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
36117 (UNSPEC_VRLNM): Likewise.
36118 (VIlong): New mode iterator.
36119 (altivec_vrl<VI_char>mi): New define_insn.
36120 (altivec_vrl<VI_char>nm): Likewise.
36121 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
36123 (VRLDNM): Likewise.
36124 (RLNM): New polymorphic function entry.
36125 (VRLWMI): New monomorphic function entry.
36126 (VRLDMI): Likewise.
36127 (RLMI): New polymorphic function entry.
36128 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
36129 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
36130 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
36133 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36136 * dwarf2out.c (field_byte_offset): Restore the
36137 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
36138 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
36139 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
36142 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36145 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
36147 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36150 * dwarf2out.c (add_data_member_location_attribute): For constant
36151 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
36152 instead of DW_AT_data_member_location, DW_AT_bit_offset and
36153 DW_AT_byte_size attributes.
36155 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36157 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
36158 after forcing to constant memory when the code model is medium.
36160 2017-01-17 Julia Koval <julia.koval@intel.com>
36163 * config/i386/avx512fintrin.h
36164 (_mm512_i32gather_ps): Change __addr type to void const*.
36165 (_mm512_mask_i32gather_ps): Ditto.
36166 (_mm512_i32gather_pd): Ditto.
36167 (_mm512_mask_i32gather_pd): Ditto.
36168 (_mm512_i64gather_ps): Ditto.
36169 (_mm512_mask_i64gather_ps): Ditto.
36170 (_mm512_i64gather_pd): Ditto.
36171 (_mm512_mask_i64gather_pd): Ditto.
36172 (_mm512_i32gather_epi32): Ditto.
36173 (_mm512_mask_i32gather_epi32): Ditto.
36174 (_mm512_i32gather_epi64): Ditto.
36175 (_mm512_mask_i32gather_epi64): Ditto.
36176 (_mm512_i64gather_epi32): Ditto.
36177 (_mm512_mask_i64gather_epi32): Ditto.
36178 (_mm512_i64gather_epi64): Ditto.
36179 (_mm512_mask_i64gather_epi64): Ditto.
36180 (_mm512_i32scatter_ps): Change __addr type to void*.
36181 (_mm512_mask_i32scatter_ps): Ditto.
36182 (_mm512_i32scatter_pd): Ditto.
36183 (_mm512_mask_i32scatter_pd): Ditto.
36184 (_mm512_i64scatter_ps): Ditto.
36185 (_mm512_mask_i64scatter_ps): Ditto.
36186 (_mm512_i64scatter_pd): Ditto.
36187 (_mm512_mask_i64scatter_pd): Ditto.
36188 (_mm512_i32scatter_epi32): Ditto.
36189 (_mm512_mask_i32scatter_epi32): Ditto.
36190 (_mm512_i32scatter_epi64): Ditto.
36191 (_mm512_mask_i32scatter_epi64): Ditto.
36192 (_mm512_i64scatter_epi32): Ditto.
36193 (_mm512_mask_i64scatter_epi32): Ditto.
36194 (_mm512_i64scatter_epi64): Ditto.
36195 (_mm512_mask_i64scatter_epi64): Ditto.
36196 * config/i386/avx512pfintrin.h
36197 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
36198 (_mm512_mask_prefetch_i32gather_ps): Ditto.
36199 (_mm512_mask_prefetch_i64gather_pd): Ditto.
36200 (_mm512_mask_prefetch_i64gather_ps): Ditto.
36201 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
36202 (_mm512_prefetch_i32scatter_ps): Ditto.
36203 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
36204 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
36205 (_mm512_prefetch_i64scatter_pd): Ditto.
36206 (_mm512_prefetch_i64scatter_ps): Ditto.
36207 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
36208 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
36209 * config/i386/avx512vlintrin.h
36210 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
36211 (_mm_mmask_i32gather_ps): Ditto.
36212 (_mm256_mmask_i32gather_pd): Ditto.
36213 (_mm_mmask_i32gather_pd): Ditto.
36214 (_mm256_mmask_i64gather_ps): Ditto.
36215 (_mm_mmask_i64gather_ps): Ditto.
36216 (_mm256_mmask_i64gather_pd): Ditto.
36217 (_mm_mmask_i64gather_pd): Ditto.
36218 (_mm256_mmask_i32gather_epi32): Ditto.
36219 (_mm_mmask_i32gather_epi32): Ditto.
36220 (_mm256_mmask_i32gather_epi64): Ditto.
36221 (_mm_mmask_i32gather_epi64): Ditto.
36222 (_mm256_mmask_i64gather_epi32): Ditto.
36223 (_mm_mmask_i64gather_epi32): Ditto.
36224 (_mm256_mmask_i64gather_epi64): Ditto.
36225 (_mm_mmask_i64gather_epi64): Ditto.
36226 (_mm256_i32scatter_ps): Change __addr type to void*.
36227 (_mm256_mask_i32scatter_ps): Ditto.
36228 (_mm_i32scatter_ps): Ditto.
36229 (_mm_mask_i32scatter_ps): Ditto.
36230 (_mm256_i32scatter_pd): Ditto.
36231 (_mm256_mask_i32scatter_pd): Ditto.
36232 (_mm_i32scatter_pd): Ditto.
36233 (_mm_mask_i32scatter_pd): Ditto.
36234 (_mm256_i64scatter_ps): Ditto.
36235 (_mm256_mask_i64scatter_ps): Ditto.
36236 (_mm_i64scatter_ps): Ditto.
36237 (_mm_mask_i64scatter_ps): Ditto.
36238 (_mm256_i64scatter_pd): Ditto.
36239 (_mm256_mask_i64scatter_pd): Ditto.
36240 (_mm_i64scatter_pd): Ditto.
36241 (_mm_mask_i64scatter_pd): Ditto.
36242 (_mm256_i32scatter_epi32): Ditto.
36243 (_mm256_mask_i32scatter_epi32): Ditto.
36244 (_mm_i32scatter_epi32): Ditto.
36245 (_mm_mask_i32scatter_epi32): Ditto.
36246 (_mm256_i32scatter_epi64): Ditto.
36247 (_mm256_mask_i32scatter_epi64): Ditto.
36248 (_mm_i32scatter_epi64): Ditto.
36249 (_mm_mask_i32scatter_epi64): Ditto.
36250 (_mm256_i64scatter_epi32): Ditto.
36251 (_mm256_mask_i64scatter_epi32): Ditto.
36252 (_mm_i64scatter_epi32): Ditto.
36253 (_mm_mask_i64scatter_epi32): Ditto.
36254 (_mm256_i64scatter_epi64): Ditto.
36255 (_mm256_mask_i64scatter_epi64): Ditto.
36256 (_mm_i64scatter_epi64): Ditto.
36257 (_mm_mask_i64scatter_epi64): Ditto.
36258 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
36259 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
36260 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
36261 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
36262 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
36263 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
36264 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
36265 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
36266 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
36267 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
36268 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
36269 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
36270 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
36271 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
36272 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
36273 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
36274 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
36275 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
36276 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
36277 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
36278 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
36279 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
36280 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
36281 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
36282 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
36283 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
36284 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
36285 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
36286 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
36287 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
36288 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
36289 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
36290 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
36291 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
36292 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
36293 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
36294 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
36295 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
36296 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
36297 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
36298 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
36299 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
36300 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
36301 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
36302 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
36303 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
36304 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
36305 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
36306 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
36307 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
36308 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
36309 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
36310 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
36311 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
36312 definitions accordingly.
36314 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
36315 Kuan-Lin Chen <kuanlinchentw@gmail.com>
36318 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
36321 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
36324 * ira-conflicts.c (ira_build_conflicts): Update total conflict
36325 hard regs for inner regno.
36327 2017-01-17 Martin Liska <mliska@suse.cz>
36330 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
36331 assumption and add comment.
36333 2017-01-17 Nathan Sidwell <nathan@acm.org>
36335 * ipa-visibility.c (localize_node): New function, broken out of ...
36336 (function_and_variable_visibility): ... here. Call it.
36338 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36340 PR middle-end/77445
36341 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
36342 correctly set frequency of oudgoing edge.
36343 (duplicate_thread_path): Fix profile updating.
36345 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36348 * configure.ac: Add GCC_BASE_VER.
36349 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
36350 version from BASE-VER file.
36351 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
36352 (gcc.o): Depend on $(BASEVER).
36353 * common.opt (dumpfullversion): New option.
36354 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
36355 * doc/invoke.texi: Document -dumpfullversion.
36356 * doc/install.texi: Document --with-gcc-major-version-only.
36357 * configure: Regenerated.
36359 2017-01-17 Richard Biener <rguenther@suse.de>
36361 PR tree-optimization/71433
36362 * tree-vrp.c (register_new_assert_for): Merge same asserts
36363 on all incoming edges.
36364 (process_assert_insertions_for): Handle insertions at the
36367 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
36369 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
36370 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
36372 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
36375 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
36378 2017-01-17 Alan Modra <amodra@gmail.com>
36381 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
36382 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
36383 symbolic stack limit when pic.
36385 2017-01-16 Martin Sebor <msebor@redhat.com>
36387 PR tree-optimization/78608
36388 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
36390 2017-01-16 Jeff Law <law@redhat.com>
36393 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
36394 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
36395 for several include directories that may be relative to sysroot.
36396 * config/i386/x-mingw32 (gplus_includedir): Define.
36397 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
36398 (native_system_includedir): Likewise.
36399 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
36400 override if TARGET_SYSTEM_ROOT is defined.
36401 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
36403 PR tree-optimization/79090
36404 PR tree-optimization/33562
36405 PR tree-optimization/61912
36406 PR tree-optimization/77485
36407 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
36408 and computed trims into the dump file.
36410 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36412 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
36414 2017-01-16 Jakub Jelinek <jakub@redhat.com>
36417 * gimplify.c (gimplify_init_constructor): If want_value and
36418 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
36422 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
36423 sequence. Formatting fixes.
36424 (doloop_optimize): Formatting fixes.
36427 * gcc.c (debug_level_greater_than_spec_func): New function.
36428 (static_spec_functions): Add debug-level-gt spec function.
36429 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
36431 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36432 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36433 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
36434 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
36435 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
36436 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
36438 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
36440 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
36441 QImode fixups to general and mask registers only.
36443 2017-01-16 Carl Love <cel@us.ibm.com>
36445 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
36446 for built-in functions
36447 vector signed char vec_nabs (vector signed char)
36448 vector signed short vec_nabs (vector signed short)
36449 vector signed int vec_nabs (vector signed int)
36450 vector signed long long vec_nabs (vector signed long long)
36451 vector float vec_nabs (vector float)
36452 vector double vec_nabs (vector double)
36453 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
36455 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
36456 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
36457 * doc/extend.texi: Update the documentation file for the new built-in
36460 2017-01-16 Martin Sebor <msebor@redhat.com>
36462 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
36465 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36467 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
36468 UNSPEC_VSX__XXSPLTD to require special splat handling.
36470 2017-01-16 David Malcolm <dmalcolm@redhat.com>
36473 * system.h: Poison strndup.
36475 2017-01-16 Alan Modra <amodra@gmail.com>
36478 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
36481 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
36483 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
36485 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
36487 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
36488 call recog here. Assert that INSN_CODE (insn) is non-negative.
36490 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36493 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
36495 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
36496 in the currently scheduled RTL fragment.
36498 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36500 PR rtl-optimization/78751
36501 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
36504 2017-01-14 Jeff Law <law@redhat.com>
36506 PR tree-optimization/79090
36507 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
36508 variable length stores.
36509 (compute_trims): Delete dead assignment to *trim_tail.
36510 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
36513 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
36515 PR rtl-optimization/78626
36516 PR rtl-optimization/78727
36517 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
36518 of a block, and split such blocks after everything else is finished.
36520 2017-01-14 Alan Modra <amodra@gmail.com>
36523 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
36524 target legitimate_combined_insn.
36525 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
36526 (rs6000_legitimate_combined_insn): New function.
36527 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
36529 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
36530 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
36531 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
36533 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
36535 * doc/frontends.texi (G++ and GCC): Remove references to Java.
36537 2017-01-13 Jeff Law <law@redhat.com>
36539 PR tree-optimization/33562
36540 PR tree-optimization/61912
36541 PR tree-optimization/77485
36542 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
36544 (delete_dead_assignment): Likewise.
36545 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
36546 statement to delete_dead_call and delete_dead_assignment.
36548 2017-01-13 David Malcolm <dmalcolm@redhat.com>
36551 * substring-locations.c (format_warning_va): Strengthen case 1 so
36552 that both endpoints of the substring must be within the format
36553 range for just the substring to be printed.
36555 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
36557 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
36558 * config/i386/i386.c (ix86_target_string): Add missing options
36559 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
36560 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
36561 flags_other and ix86_target_other to flags2_other. Display unknown
36563 (ix86_valid_target_attribute_inner_p): Add missing options and
36564 reorder options by implied ISAs, as in ix86_target_string.
36566 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36568 * hash-table.h (hash_table::too_empty_p): New function.
36569 (hash_table::expand): Use it.
36570 (hash_table::traverse): Likewise.
36571 (hash_table::empty_slot): Use sizeof (value_type) instead of
36572 sizeof (PTR) to convert bytes to elements. Shrink the table
36573 if the current size is excessive for the current number of
36576 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36578 * ira-costs.c (record_reg_classes): Break from the inner loop
36579 early once alt_fail is known to be true. Update outer loop
36580 handling accordingly.
36582 2017-01-13 Jeff Law <law@redhat.com>
36584 * tree-ssa-dse.c (decrement_count): New function.
36585 (increment_start_addr, maybe_trim_memstar_call): Likewise.
36586 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
36587 when we know the partially dead statement is a mem* function.
36589 PR tree-optimization/61912
36590 PR tree-optimization/77485
36591 * tree-ssa-dse.c: Include expr.h.
36592 (maybe_trim_constructor_store): New function.
36593 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
36595 PR tree-optimization/33562
36596 PR tree-optimization/61912
36597 PR tree-optimization/77485
36598 * doc/invoke.texi: Document new dse-max-object-size param.
36599 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
36600 * tree-ssa-dse.c: Include params.h.
36601 (dse_store_status): New enum.
36602 (initialize_ao_ref_for_dse): New, partially extracted from
36604 (valid_ao_ref_for_dse, normalize_ref): New.
36605 (setup_live_bytes_from_ref, compute_trims): Likewise.
36606 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
36607 (maybe_trim_partially_dead_store): Likewise.
36608 (maybe_trim_complex_store): Likewise.
36609 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
36610 Track what bytes live from the original store. Return tri-state
36611 for dead, partially dead or live.
36612 (dse_dom_walker): Add constructor, destructor and new private members.
36613 (delete_dead_call, delete_dead_assignment): New extracted from
36615 (dse_optimize_stmt): Make a member of dse_dom_walker.
36616 Use initialize_ao_ref_for_dse.
36618 PR tree-optimization/33562
36619 PR tree-optimization/61912
36620 PR tree-optimization/77485
36621 * sbitmap.h (bitmap_count_bits): Prototype.
36622 (bitmap_clear_range, bitmap_set_range): Likewise.
36623 * sbitmap.c (bitmap_clear_range): New function.
36624 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
36626 2017-01-13 Martin Liska <mliska@suse.cz>
36629 * function.c (set_cfun): Add new argument force.
36630 * function.h (set_cfun): Likewise.
36631 * ipa-inline-transform.c (inline_call): Use the function when
36632 strict alising from is dropped for function we inline to.
36634 2017-01-13 Richard Biener <rguenther@suse.de>
36636 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
36637 for dumping GIMPLE INTEGER_CSTs.
36639 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
36641 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
36642 to 201112L since C++17.
36644 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
36647 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
36648 if -fsanitize=kernel-address is present.
36650 2017-01-13 Richard Biener <rguenther@suse.de>
36652 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
36653 as _Literal ( type ) number in case usual suffixes do not
36654 preserve all information.
36656 2017-01-13 Richard Biener <rguenther@suse.de>
36658 PR tree-optimization/77283
36659 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
36660 and ssa-iterators.h.
36661 (is_feasible_trace): Implement a cost model based on joiner
36664 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
36667 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
36668 char or short to __float128/_Float128 directly.
36670 2017-01-12 Martin Sebor <msebor@redhat.com>
36672 to -Wformat-overflow.
36673 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
36674 (min_bytes_remaining): Same.
36675 (get_string_length): Same.
36676 (format_string): Same.
36677 (format_directive): Same.
36679 (pass_sprintf_length::handle_gimple_call): Same.
36681 2017-01-12 Jakub Jelinek <jakub@redhat.com>
36683 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
36684 info.nowrite calls with no lhs that can't throw. Return bool
36685 whether gsi_remove has been called or not.
36686 (pass_sprintf_length::handle_gimple_call): Return bool whether
36687 try_substitute_return_value called gsi_remove. Formatting fix.
36688 (pass_sprintf_length::execute): Don't use gsi_remove if
36689 handle_gimple_call returned true.
36692 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
36693 be removed due to side-effects, don't remove following barrier nor
36694 turn the successor edge into fallthru edge.
36696 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36699 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
36700 element-reversing loads and stores as not swappable.
36702 2017-01-12 Nathan Sidwell <nathan@acm.org>
36703 Nicolai Stange <nicstange@gmail.com>
36705 * combine.c (try_combine): Don't ignore result of overlap checking
36706 loop. Combine overlap & asm check into single loop.
36708 2017-01-12 Richard Biener <rguenther@suse.de>
36710 * tree-pretty-print.c (dump_generic_node): Provide -gimple
36711 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
36713 2017-01-12 Richard Biener <rguenther@suse.de>
36715 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
36716 and TS_TARGET_OPTION directly derive from TS_BASE.
36717 * tree-core.h (tree_optimization_option): Derive from tree_base.
36718 (tree_target_option): Likewise.
36720 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
36722 * config/i386/i386.c (memory_address_length): Increase len
36723 only when rip_relative_addr_p returns false.
36725 2017-01-11 Julia Koval <julia.koval@intel.com>
36727 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
36728 (OPTION_MASK_ISA_SGX_SET): New.
36729 (ix86_handle_option): Handle OPT_msgx.
36730 * config.gcc: Added sgxintrin.h.
36731 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
36732 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
36733 * config/i386/i386.c (ix86_target_string): Add -msgx.
36735 (ix86_option_override_internal): Handle new options.
36736 (ix86_valid_target_attribute_inner_p): Add sgx.
36737 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
36738 * config/i386/i386.opt: Add msgx.
36739 * config/i386/sgxintrin.h: New file.
36740 * config/i386/x86intrin.h: Add sgxintrin.h.
36742 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36745 * fold-const.c (maybe_nonzero_address): Return 1 for function
36747 (tree_single_nonzero_warnv_p): Don't handle function local objects
36751 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
36754 2017-01-11 David Malcolm <dmalcolm@redhat.com>
36757 * opts.c: Include "spellcheck.h"
36758 (struct string_fragment): New struct.
36759 (struct edit_distance_traits<const string_fragment &>): New
36761 (get_closest_sanitizer_option): New function.
36762 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
36764 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36766 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
36768 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
36769 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
36770 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
36771 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
36772 for initial die_offset if dwarf_split_debug_info.
36773 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
36774 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
36776 (output_skeleton_debug_sections): Formatting fix. Use
36777 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
36778 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
36780 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
36782 * config/arm/cortex-a53.md: Add bypasses for
36783 cortex_a53_r2f_cvt.
36784 (cortex_a53_r2f): Only use for transfers.
36785 (cortex_a53_f2r): Likewise.
36786 (cortex_a53_r2f_cvt): Add reservation for conversions.
36787 (cortex_a53_f2r_cvt): Likewise.
36789 2017-01-11 Tamar Christina <tamar.christina@arm.com>
36791 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
36792 to all inlined functions, change static to extern.
36794 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
36797 * config/arm/arm.c (legitimize_pic_address): Handle reference to
36799 (arm_assemble_integer): Likewise.
36801 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
36803 * config.gcc: Use new awk script to check CPU, FPU and architecture
36804 parameters for --with-... options.
36805 * config/arm/parsecpu.awk: New file
36806 * config/arm/arm-cpus.in: New file.
36807 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
36809 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
36811 * config/arm/t-arm: Update dependency rules.
36812 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
36813 of processing .def files.
36814 * config/arm/genopt.sh: Deleted.
36815 * config/arm/gentune.sh: Deleted.
36816 * config/arm/arm-cores.def: Deleted.
36817 * config/arm/arm-arches.def: Deleted.
36818 * config/arm/arm-fpus.def: Deleted.
36819 * config/arm/arm-tune.md: Regenerated.
36820 * config/arm/arm-tables.opt: Regenerated.
36821 * config/arm/arm-cpu.h: New generated file.
36822 * config/arm/arm-cpu-data.h: New generated file.
36823 * config/arm/arm-cpu-cdata.h: New generated file.
36825 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
36828 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
36830 (input_varpool_node): Unpack dynamically_initialized bit.
36832 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
36834 PR rtl-optimization/79032
36835 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
36836 the alignment of the adjusted memory reference against that of MODE,
36837 instead of the alignment of the original memory reference.
36839 2017-01-11 Martin Jambor <mjambor@suse.cz>
36841 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
36843 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
36844 decorated functions.
36846 2017-01-11 Richard Biener <rguenther@suse.de>
36848 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
36849 set range/nonnull info for PHI results. Do not set it on
36850 stmts marked for removal.
36852 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
36854 * expr.c (store_field): In the bitfield case, fetch the return value
36855 from the registers before applying a single big-endian adjustment.
36856 Always do a final load for a BLKmode value not larger than a word.
36858 2017-01-10 David Malcolm <dmalcolm@redhat.com>
36861 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36862 that we correctly handle column numbers greater than
36863 LINE_MAP_MAX_COLUMN_NUMBER.
36865 2017-01-10 Martin Sebor <msebor@redhat.com>
36867 PR middle-end/78245
36868 * gimple-ssa-sprintf.c (get_destination_size): Call
36869 {init,fini}object_sizes.
36870 * tree-object-size.c (addr_object_size): Adjust.
36871 (pass_through_call): Adjust.
36872 (pass_object_sizes::execute): Adjust.
36873 * tree-object-size.h (fini_object_sizes): Declare.
36875 2017-01-10 Martin Sebor <msebor@redhat.com>
36877 PR tree-optimization/78775
36878 * builtins.c (get_size_range): Move...
36879 * calls.c: ...to here.
36880 (alloc_max_size): Accept zero argument.
36881 (operand_signed_p): Remove.
36882 (maybe_warn_alloc_args_overflow): Call get_size_range.
36883 * calls.h (get_size_range): Declare.
36885 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
36887 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
36888 from TI's devices.csv file as of September 2016.
36889 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
36891 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
36893 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
36894 * doc/invoke.texi: Likewise.
36895 * doc/md.texi: Likewise.
36896 * doc/objc.texi: Likewise.
36898 2017-01-10 Joshua Conner <joshconner@google.com>
36900 * config/arm/fuchsia-elf.h: New file.
36901 * config/fuchsia.h: New file.
36902 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
36903 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
36905 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
36907 2016-01-10 Richard Biener <rguenther@suse.de>
36909 PR tree-optimization/79034
36910 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
36911 Propagate out degenerate PHIs in the joiner.
36913 2017-01-10 Martin Liska <mliska@suse.cz>
36915 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
36916 (sort_congruence_classes_by_decl_uid): Likewise.
36917 (sort_congruence_class_groups_by_decl_uid): Likewise.
36918 (sem_item_optimizer::merge_classes): Sort class, groups in these
36919 classes and members in the groups by DECL_UID of declarations.
36920 This would make merge operations stable.
36922 2017-01-10 Martin Liska <mliska@suse.cz>
36924 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
36925 usage of m_classes_vec.
36926 (sem_item_optimizer::~sem_item_optimizer): Likewise.
36927 (sem_item_optimizer::get_group_by_hash): Likewise.
36928 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
36929 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
36930 (sem_item_optimizer::verify_classes): Likewise.
36931 (sem_item_optimizer::process_cong_reduction): Likewise.
36932 (sem_item_optimizer::dump_cong_classes): Likewise.
36933 (sem_item_optimizer::merge_classes): Likewise.
36934 * ipa-icf.h (congruence_class_hash): Rename from
36935 congruence_class_group_hash. Remove declaration of m_classes_vec.
36937 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
36939 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
36940 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
36941 * config.gcc: Add avx512vpopcntdqintrin.h.
36942 * config/i386/avx512vpopcntdqintrin.h: New.
36943 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
36944 * config/i386/i386-builtin-types.def: Add new types.
36945 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
36946 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
36947 __builtin_ia32_vpopcountq_v8di_mask): New.
36948 * config/i386/i386-c.c (ix86_target_macros_internal): Define
36949 __AVX512VPOPCNTDQ__.
36950 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
36951 (PTA_AVX512VPOPCNTDQ): Define.
36952 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
36953 TARGET_AVX512VPOPCNTDQ_P): Define.
36954 * config/i386/i386.opt: Add mavx512vpopcntdq.
36955 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
36956 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
36958 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36960 PR middle-end/77484
36961 * predict.def (PRED_CALL): Set to 67.
36963 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
36965 * expr.c (store_field): In the bitfield case, if the value comes from
36966 a function call and is of an aggregate type returned in registers, do
36967 not modify the field mode; extract the value in all cases if the mode
36968 is BLKmode and the size is not larger than a word.
36970 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
36973 * config/i386/cpuid.h: Fix undefined behavior.
36975 2017-01-04 Jeff Law <law@redhat.com>
36977 PR tree-optimization/79007
36978 PR tree-optimization/67955
36979 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
36980 conservative for pt.null when flag_non_call_exceptions is on.
36982 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36984 PR translation/79019
36985 PR translation/79020
36986 * params.def (PARAM_INLINE_MIN_SPEEDUP,
36987 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
36988 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
36990 * config/avr/avr.opt (maccumulate-args): Likewise.
36991 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
36992 * common.opt (freport-bug): Likewise.
36993 * cif-code.def (CIF_FINAL_ERROR): Likewise.
36994 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
36995 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
36996 translatable string.
36997 * config/i386/i386.c (function_value_32): Likewise.
36998 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
36999 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37001 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37002 * common/config/msp430/msp430-common.c (msp430_handle_option):
37004 * symtab.c (symtab_node::verify_base): Likewise.
37005 * opts.c (set_debug_level): Likewise.
37006 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37007 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37008 missing whitespace to translatable strings.
37009 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37010 * config/sh/superh.h: Likewise.
37011 * config/i386/xopintrin.h: Likewise.
37012 * config/i386/znver1.md: Likewise.
37013 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37014 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37015 * double-int.h (struct double_int): Likewise.
37016 * double-int.c (div_and_round_double): Likewise.
37017 * wide-int.cc: Likewise.
37018 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
37019 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
37020 * cfgcleanup.c (crossjumps_occured): Renamed to ...
37021 (crossjumps_occurred): ... this.
37022 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
37025 PR tree-optimization/78899
37026 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
37027 returning bool return struct loop *, NULL for failure and the new
37029 (versionable_outer_loop_p): Don't version outer loop if it has
37030 dont_vectorized bit set.
37031 (tree_if_conversion): When versioning outer loop, ensure
37032 tree_if_conversion is performed also on the inner loop of the
37033 non-vectorizable outer loop copy.
37034 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
37035 LOOP_VECTORIZED in inner loop of the scalar outer loop and
37036 prevent vectorization of it.
37037 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
37038 the outer loop vectorization of the non-scalar version is attempted
37039 before vectorization of the inner loop in scalar version. If
37040 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
37041 vectorization of its inner loop.
37042 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
37043 has 2 inner loops, rename also on edges from bb whose single pred
37044 is outer_loop->header. Fix typo in function comment.
37046 2017-01-09 Martin Sebor <msebor@redhat.com>
37049 * asan.c (asan_emit_stack_protection): Increase local buffer size
37050 to avoid snprintf truncation warning.
37052 2017-01-09 Andrew Pinski <apinski@cavium.com>
37054 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
37055 to reference thunderx2t99 for the tuning structure
37056 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
37058 (thunderx2t99_extra_costs): This.
37059 * config/aarch64/aarch64-tune.md: Regenerate.
37060 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
37061 (vulcan_addrcost_table): This.
37062 (vulcan_regmove_cost): Rename to ...
37063 (thunderx2t99_regmove_cost): This.
37064 (vulcan_vector_cost): Rename to ...
37065 (thunderx2t99_vector_cost): this.
37066 (vulcan_branch_cost): Rename to ...
37067 (thunderx2t99_branch_cost): This.
37068 (vulcan_tunings): Rename to ...
37069 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
37070 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
37072 2017-01-09 Martin Jambor <mjambor@suse.cz>
37076 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
37077 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
37078 (propagate_vr_accross_jump_function): Use the above function for all
37079 value range computations for pass-through jump functions and type
37080 converasion from explicit value range values.
37081 (ipcp_propagate_stage): Do not attempt to deduce types of formal
37082 parameters from TYPE_ARG_TYPES.
37083 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
37084 (ipa_write_node_info): Stream type of the actual argument.
37085 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
37087 2017-01-09 Martin Liska <mliska@suse.cz>
37090 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
37091 (lookup_compiler): Do not show error message with have_E.
37093 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37095 PR tree-optimization/78938
37096 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
37097 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
37098 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
37099 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
37102 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37104 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
37107 2017-01-09 Richard Biener <rguenther@suse.de>
37109 PR tree-optimization/78997
37110 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
37111 name condition properly.
37113 2017-01-09 Richard Biener <rguenther@suse.de>
37116 * dwarf2out.c (is_cxx): New overload with context.
37117 (is_naming_typedef_decl): Use it.
37119 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37121 * invoke.texi (Option Summary): Correct spacing in option lists
37122 and add line breaks to fix over-long lines.
37124 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37126 PR middle-end/17660
37128 * extend.texi (Common Variable Attributes): Add xref to GCC
37129 Internals manual to explain mode attribute keywords.
37131 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37134 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
37135 and Preprocessor Options.
37136 (Options for Linking): Document -pthread here....
37137 (RS/6000 and PowerPC Options): ...not here.
37138 (Solaris 2 Options): ...or here.
37139 * doc/cppopts.texi: Document -pthread.
37141 2017-01-08 Martin Sebor <msebor@redhat.com>
37143 PR middle-end/77708
37144 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
37145 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
37146 New member functions.
37147 (format_directive): Used them.
37149 (pass_sprintf_length::handle_gimple_call): Same.
37150 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
37151 to avoid truncation for any argument.
37152 (extract_affine_mul): Same.
37153 * tree.c (get_file_function_name): Same.
37155 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37157 PR middle-end/77484
37158 * predict.def (PRED_INDIR_CALL): Set to 86.
37160 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37162 PR preprocessor/54124
37163 * doc/cppopts.texi: Reformat -d subtable to list the full name
37164 of the options. Add cross-reference to the docs for the general
37165 compiler -d options.
37166 * doc/invoke.texi (Developer Options): Add cross-reference to the
37167 preprocessor-specific -d option documentation.
37169 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37171 PR preprocessor/13498
37172 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
37173 redudant material, and reflect new command-line options.
37174 (System Headers): Likewise.
37176 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37178 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
37179 -isystem, and -idirafter. Copy-edit.
37180 * doc/cppopts.texi: Copy-edit. Remove contradiction about
37181 default for -ftrack-macro-expansion. Delete obsolete and
37182 badly-formatted implementation details about -fdebug-cpp output.
37183 * doc/cppwarnopts.texi: Copy-edit.
37185 2017-01-07 David Malcolm <dmalcolm@redhat.com>
37188 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37189 that the transition from a max line width >= 1<<10 to narrower
37190 lines works correctly.
37192 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
37194 * doc/options.texi (PerFunction): New.
37195 * opt-functions.awk (switch_flags): Map both Optimization and
37196 PerFunction to CL_OPTIMIZATION.
37197 * opth-gen.awk: Test for PerFunction flag along with
37199 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
37200 it only when the latter is present. Skip those that don't in
37201 the hash function generator.
37202 * common.opt (fvar-tracking): Mark as PerFunction instead of
37204 (fvar-tracking-assignments): Likewise.
37205 (fvar-tracking-assignments-toggle): Likewise.
37206 (fvar-tracking-uninit): Likewise.
37208 2017-01-07 Jakub Jelinek <jakub@redhat.com>
37210 PR translation/79018
37211 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
37214 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
37217 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
37218 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
37219 TARGET_LONG_JUMP_TABLE_OFFSETS.
37220 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
37221 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
37222 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
37223 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
37224 * config/m68k/m68k.md (tablejump expander): Likewise.
37225 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
37226 TARGET_LONG_JUMP_TABLE_OFFSETS.
37227 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
37228 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
37230 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
37231 David Holsgrove <david.holsgrove@xilinx.com>
37233 * common/config/microblaze/microblaze-common.c
37234 (TARGET_EXCEPT_UNWIND_INFO): Remove.
37235 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
37237 * config/microblaze/microblaze.c (microblaze_must_save_register)
37238 (microblaze_expand_epilogue, microblaze_return_addr): Handle
37240 (microblaze_eh_return): New function.
37241 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
37242 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
37243 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
37244 * config/microblaze/microblaze.md (eh_return): New pattern.
37246 2017-01-06 Jakub Jelinek <jakub@redhat.com>
37248 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
37249 GCC_DIAGNOSTIC_STRINGIFY): Define.
37251 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
37253 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37255 * config/arm/arm.md (<mcrr>): New.
37257 * config/arm/arm.c (arm_arch5te): New.
37258 (arm_option_override): Set arm_arch5te.
37259 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
37261 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
37262 (arm_mcrr_qualifiers): ... this. New.
37263 (MRRC_QUALIFIERS): Define to...
37264 (arm_mrrc_qualifiers): ... this. New.
37265 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
37267 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
37268 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
37269 (MRRCI, mrrc, MRRC): New.
37270 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
37271 VUNSPEC_MRRC2): New.
37273 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37275 * config/arm/arm.md (<mcr>): New.
37277 * config/arm/arm.c (arm_coproc_builtin_available): Add
37278 support for mcr, mrc, mcr2 and mrc2.
37279 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
37280 (arm_mcr_qualifiers): ... this. New.
37281 (MRC_QUALIFIERS): Define to ...
37282 (arm_mrc_qualifiers): ... this. New.
37283 (MCR_QUALIFIERS): Define to ...
37284 (arm_mcr_qualifiers): ... this. New.
37285 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
37287 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
37288 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
37289 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
37290 VUNSPEC_MRC2): New.
37292 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37294 * config/arm/arm.md (*ldc): New.
37298 * config/arm/arm.c (arm_coproc_builtin_available): Add
37299 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
37300 (arm_coproc_ldc_stc_legitimate_address): New.
37301 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
37302 'qualifier_const_pointer'.
37303 (LDC_QUALIFIERS): Define to...
37304 (arm_ldc_qualifiers): ... this. New.
37305 (STC_QUALIFIERS): Define to...
37306 (arm_stc_qualifiers): ... this. New.
37307 * config/arm/arm-protos.h
37308 (arm_coproc_ldc_stc_legitimate_address): New.
37309 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
37310 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
37311 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
37312 stc2, stcl, stc2l): New.
37313 * config/arm/constraints.md (Uz): New.
37314 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
37315 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
37316 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
37317 VUNSPEC_STC2L): New.
37319 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37321 * config/arm/arm.md (<cdp>): New.
37322 * config/arm/arm.c (neon_const_bounds): Rename this ...
37323 (arm_const_bounds): ... this.
37324 (arm_coproc_builtin_available): New.
37325 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
37326 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
37327 (CDP_QUALIFIERS): Define to...
37328 (arm_cdp_qualifiers): ... this. New.
37330 (arm_expand_builtin_args): Add case for 6 arguments.
37331 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
37332 (arm_const_bounds): ... this.
37333 (arm_coproc_builtin_available): New.
37334 * config/arm/arm_acle.h (__arm_cdp): New.
37336 * config/arm/arm_acle_builtins.def (cdp): New.
37338 * config/arm/iterators.md (CDPI,CDP,cdp): New.
37339 * config/arm/neon.md: Rename all 'neon_const_bounds' to
37340 'arm_const_bounds'.
37341 * config/arm/types.md (coproc): New.
37342 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
37343 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
37344 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
37345 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
37347 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37349 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
37350 (UBINOP_QUALIFIERS): New.
37352 (acle_builtin_data): New. Change comment.
37353 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
37354 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
37355 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
37356 arm_acle_builtins.def.
37357 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
37358 (arm_init_acle_builtins): New.
37359 (CRC32_BUILTIN): Remove.
37360 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
37361 crc32cb, crc32ch and crc32cw.
37362 (arm_init_crc32_builtins): Remove.
37363 (arm_init_builtins): Use arm_init_acle_builtins rather
37364 than arm_init_crc32_builtins.
37365 (arm_expand_acle_builtin): New.
37366 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
37367 * config/arm/arm_acle_builtins.def: New.
37369 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37371 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
37372 (arm_builtin_datum): ... this.
37373 (arm_init_neon_builtin): Rename to ...
37374 (arm_init_builtin): ... this. Add a new parameters PREFIX
37375 and USE_SIG_IN_NAME.
37376 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
37377 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
37378 'arm_builtin_datum'.
37379 (arm_init_vfp_builtins): Likewise.
37380 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
37381 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
37382 (arm_expand_neon_args): Rename to ...
37383 (arm_expand_builtin_args): ... this. Rename builtin_arg
37384 enum values and differentiate between ARG_BUILTIN_MEMORY
37385 and ARG_BUILTIN_NEON_MEMORY.
37386 (arm_expand_neon_builtin_1): Rename to ...
37387 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
37388 values, arm_expand_builtin_args and add bool parameter NEON.
37389 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
37390 (arm_expand_vfp_builtin): Likewise.
37391 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
37393 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37395 PR middle-end/77484
37396 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
37397 * predict.c (tree_estimate_probability_bb): Reverse direction of
37398 polymorphic call predictor.
37400 2017-01-06 David Malcolm <dmalcolm@redhat.com>
37402 * passes.c (execute_one_pass): Split out pass-skipping logic into...
37403 (determine_pass_name_match): ...this new function and...
37404 (should_skip_pass_p): ...this new function.
37406 2017-01-06 Nathan Sidwell <nathan@acm.org>
37408 * ipa-visibility.c (function_and_variable_visibility): Reformat
37409 comments and long lines. Remove extrneous if.
37410 * symtab.c (symtab_node::make_decl_local): Fix code format.
37411 (symtab_node::set_section_for_node): Fix comment typo.
37413 2017-01-06 Martin Liska <mliska@suse.cz>
37416 * lra-constraints.c: Rename invariant to lra_invariant.
37417 * predict.c (set_even_probabilities): Initialize e to NULL.
37419 2017-01-05 Martin Sebor <msebor@redhat.com>
37421 PR tree-optimization/78910
37422 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
37423 (format_integer): Correct off-by-one error in the handling
37424 of precision with negative numbers in signed conversions..
37426 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
37428 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
37430 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37432 PR tree-optimization/71016
37433 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
37434 factor_out_conditional_conversion. Formatting fix.
37435 (factor_out_conditional_conversion): Add cond_stmt argument.
37436 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
37437 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
37440 2017-01-05 David Malcolm <dmalcolm@redhat.com>
37442 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
37443 read-rtl-function.o, and selftest-rtl.o.
37444 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
37445 (selftest::aarch64_test_loading_full_dump): New function.
37446 (selftest::aarch64_run_selftests): New function.
37447 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
37448 selftest::aarch64_run_selftests.
37449 * config/i386/i386.c
37450 (selftest::ix86_test_loading_dump_fragment_1): New function.
37451 (selftest::ix86_test_loading_call_insn): New function.
37452 (selftest::ix86_test_loading_full_dump): New function.
37453 (selftest::ix86_test_loading_unspec): New function.
37454 (selftest::ix86_run_selftests): Call the new functions.
37455 * emit-rtl.c (maybe_set_max_label_num): New function.
37456 * emit-rtl.h (maybe_set_max_label_num): New decl.
37457 * function.c (instantiate_decls): Guard call to
37458 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
37459 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
37461 * gensupport.c (gen_reader::gen_reader): Pass "false"
37462 for new "compact" param of rtx_reader.
37463 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
37464 rather than an empty string for NULL strings.
37465 * read-md.c: Potentially include config.h rather than bconfig.h.
37466 Wrap include of errors.h with #ifdef GENERATOR_FILE.
37467 (have_error): New global, copied from errors.c.
37468 (md_reader::read_name): Rename to...
37469 (md_reader::read_name_1): ...this, adding "out_loc" param,
37470 and converting "missing name or number" to returning false, rather
37472 (md_reader::read_name): Reimplement in terms of read_name_1.
37473 (md_reader::read_name_or_nil): New function.
37474 (md_reader::read_string): Handle "(nil)" by returning NULL.
37475 (md_reader::md_reader): Add new param "compact".
37476 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
37477 (md_reader::read_file): New method.
37478 * read-md.h (md_reader::md_reader): Add new param "compact".
37479 (md_reader::read_file): New method.
37480 (md_reader::is_compact): New accessor.
37481 (md_reader::read_name): Convert return type from void to file_location.
37482 (md_reader::read_name_or_nil): New decl.
37483 (md_reader::read_name_1): New decl.
37484 (md_reader::m_compact): New field.
37485 (noop_reader::noop_reader): Pass "false" for new "compact" param
37487 (rtx_reader::rtx_reader): Add new "compact" param.
37488 (rtx_reader::read_rtx_operand): Make virtual and convert return
37489 type from void to rtx.
37490 (rtx_reader::read_until): New decl.
37491 (rtx_reader::handle_any_trailing_information): New virtual function.
37492 (rtx_reader::postprocess): New virtual function.
37493 (rtx_reader::finalize_string): New virtual function.
37494 (rtx_reader::m_in_call_function_usage): New field.
37495 (rtx_reader::m_reuse_rtx_by_id): New field.
37496 * read-rtl-function.c: New file.
37497 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
37498 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
37499 (selftest::verify_three_block_rtl_cfg): New decl.
37500 * read-rtl-function.h: New file.
37501 * read-rtl.c: Potentially include config.h rather than bconfig.h.
37502 For host, include function.h, memmodel.h, and emit-rtl.h.
37503 (one_time_initialization): New function.
37504 (struct compact_insn_name): New struct.
37505 (compact_insn_names): New array.
37506 (find_code): Handle insn codes in compact dumps.
37507 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
37508 (bind_subst_iter_and_attr): Likewise.
37509 (add_condition_to_string): Likewise.
37510 (add_condition_to_rtx): Likewise.
37511 (apply_attribute_uses): Likewise.
37512 (add_current_iterators): Likewise.
37513 (apply_iterators): Likewise.
37514 (initialize_iterators): Guard usage of apply_subst_iterator with
37515 #ifdef GENERATOR_FILE.
37516 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
37517 (md_reader::read_mapping): Likewise.
37518 (add_define_attr_for_define_subst): Likewise.
37519 (add_define_subst_attr): Likewise.
37520 (read_subst_mapping): Likewise.
37521 (check_code_iterator): Likewise.
37522 (rtx_reader::read_rtx): Likewise. Move one-time initialization
37524 (one_time_initialization): New function.
37525 (rtx_reader::read_until): New method.
37526 (read_flags): New function.
37527 (parse_reg_note_name): New function.
37528 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
37529 Handle reuse_rtx ids.
37530 Wrap iterator lookup within #ifdef GENERATOR_FILE.
37531 Add parsing support for RTL dumps, mirroring the special-cases in
37532 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
37533 values, and calling handle_any_trailing_information.
37534 (rtx_reader::read_rtx_operand): Convert return type from void
37535 to rtx, returning return_rtx. Handle case 'e'. Call
37536 finalize_string on XSTR and XTMPL fields.
37537 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
37538 "(nil)" values were omitted. Call the postprocess vfunc on the
37540 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
37541 class ctor. Initialize m_in_call_function_usage. Call
37542 one_time_initialization.
37543 * rtl-tests.c (selftest::test_uncond_jump): Call
37544 set_new_first_and_last_insn.
37545 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
37546 * selftest-rtl.c: New file.
37547 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
37548 (selftest::get_insn_by_uid): New decl.
37549 * selftest-run-tests.c (selftest::run_tests): Call
37550 read_rtl_function_c_tests.
37551 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
37552 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
37555 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
37557 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
37558 operands in a special way. Assert that pos+len <= mode precision.
37560 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37562 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
37563 3 argument Alias with unlimited for the negative form.
37564 (fno-vect-cost-model): Removed.
37566 2017-01-05 Martin Liska <mliska@suse.cz>
37568 * hsa-gen.c (gen_hsa_divmod): New function.
37569 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
37571 2017-01-05 Martin Liska <mliska@suse.cz>
37574 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
37577 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37579 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
37580 small constant length operands.
37582 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37584 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
37585 between loop iterations.
37587 2017-01-05 Martin Liska <mliska@suse.cz>
37590 * gimplify.c (gimplify_decl_expr): Compare to
37591 asan_poisoned_variables instread of checking flags.
37592 (gimplify_target_expr): Likewise.
37593 (gimplify_expr): Likewise.
37594 (gimplify_function_tree): Conditionally initialize
37595 asan_poisoned_variables.
37597 2017-01-04 Jeff Law <law@redhat.com>
37599 PR tree-optimizatin/78812
37600 * rtl.h (contains_mem_rtx_p): Prototype.
37601 * ifcvt.c (containts_mem_rtx_p): Move from here to...
37602 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
37603 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
37604 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
37605 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
37607 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37609 * input.c (assert_char_at_range): Default-initialize actual_range.
37611 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37613 * df-scan.c (df_ref_create_structure): Make regno unsigned,
37614 to match the caller.
37616 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37618 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
37619 insns after final jump in test to emit dummy move.
37621 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37623 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
37624 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
37626 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37628 * multiple_target.c (create_dispatcher_calls): Init e_next.
37629 * tree-ssa-loop-split.c (split_loop): Init border.
37630 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
37633 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
37638 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
37639 (altivec_register_operand): Do not return true if the operand
37640 contains a SUBREG mixing SImode and SFmode.
37641 (vsx_register_operand): Likewise.
37642 (vsx_reg_sfsubreg_ok): New predicate.
37643 (vfloat_operand): Do not return true if the operand contains a
37644 SUBREG mixing SImode and SFmode.
37645 (vint_operand): Likewise.
37646 (vlogical_operand): Likewise.
37647 (gpc_reg_operand): Likewise.
37648 (int_reg_operand): Likewise.
37649 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
37650 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
37651 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
37653 (rs6000_emit_move_si_sf_subreg): New helper function.
37654 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
37655 fixup SUBREGs involving SImode and SFmode.
37656 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
37657 numbers for the new peephole2 optimization.
37658 (peephole2 for SFmode unions): New peephole2 to optimize cases in
37659 the GLIBC math library that do AND/IOR/XOR operations on single
37660 precision floating point.
37661 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
37662 target macros to say whether we need to avoid SUBREGs mixing
37664 (TARGET_ALLOW_SF_SUBREG): Likewise.
37665 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
37666 (UNSPEC_SI_FROM_SF): Likewise.
37667 (iorxor): Change spacing.
37668 (and_ior_xor): New iterator for AND, IOR, and XOR.
37669 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
37670 (movdi_from_sf_zero_ext): Likewise.
37671 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
37672 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
37673 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
37674 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
37675 (fms<mode>4): Likewise.
37676 (fnma<mode>4): Likewise.
37677 (fnms<mode>4): Likewise.
37678 (nfma<mode>4): Likewise.
37679 (nfms<mode>4): Likewise.
37681 2017-01-04 Marek Polacek <polacek@redhat.com>
37684 * doc/invoke.texi: Document -Wpointer-compare.
37686 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37688 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
37691 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
37692 descriptions for -gdwarf-5 and emit them as uleb128 instead of
37695 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
37698 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
37699 documentation of the powerpc_popcntb_ok attribute.
37700 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
37701 code to issue warning messages if a requested CPU configuration is
37702 not supported by the binary (assembler and loader) toolchain.
37703 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
37704 made to define a built-in function that has been disabled.
37705 (paired_init_builtins): Add assertion to prevent ICE if attempt is
37706 made to define a built-in function that has been disabled.
37707 (altivec_init_builtins): Add comment explaining why definition
37708 of the DST built-in functions is not preceded by an assertion
37709 check. Add assertions to prevent ICE if attempts are made to
37710 define an altivec predicate or an abs* built-in function that has
37712 (htm_init_builtins): Add comment explaining why definition of the
37713 htm built-in functions is not preceded by an assertion check.
37715 2017-01-04 Jeff Law <law@redhat.com>
37717 PR tree-optimizatin/67955
37718 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
37719 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
37720 the points-to solution does not include pt_null. Use DECL_PT_UID
37723 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
37725 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
37726 Use gen_int_mode instead of gen_lopwart for const_int operands.
37728 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37730 PR tree-optimization/71563
37731 * match.pd: Simplify X << Y into X if Y is known to be 0 or
37732 out of range value - has low bits known to be zero.
37734 2017-01-04 Alan Modra <amodra@gmail.com>
37736 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
37737 * configure: Regenerate.
37738 * config.in: Regenerate.
37740 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37743 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
37744 a substring of the message, but strcmp with the whole message. Ifdef
37745 ENABLE_NLS, translate the message first using dgettext.
37747 2017-01-03 Jeff Law <law@redhat.com>
37749 PR tree-optimizatin/78856
37750 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
37751 (mark_threaded_blocks): Remove code to truncate thread paths that
37752 cross multiple loop headers. Instead invalidate the cached loop
37753 iteration information and handle case of a thread path walking
37754 into an irreducible region.
37756 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
37759 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
37760 assertions. Add support for doing the signbit if the IEEE 128-bit
37761 floating point value is in a GPR.
37762 * config/rs6000/rs6000.md (Fsignbit): Delete.
37763 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
37764 Update the length attribute if the value is in a GPR.
37765 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
37766 the sign or zero extension instruction, since the value is always 0/1.
37767 (signbit<mode>2_dm2): Delete using <Fsignbit>.
37770 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
37771 extracting SImode to a GPR register so that we can generate a
37772 store, limit the vector to be in a traditional Altivec register
37773 for the vextuwrx instruction.
37775 2017-01-03 Ian Lance Taylor <iant@google.com>
37777 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
37779 2017-01-03 Martin Sebor <msebor@redhat.com>
37781 PR tree-optimization/78696
37782 * gimple-ssa-sprintf.c (format_floating): Correct handling of
37783 precision. Use MPFR for %f for greater fidelity. Correct handling
37785 (pass_sprintf_length::compute_format_length): Set width and precision
37786 specified by asrerisk to void_node for vararg functions.
37787 (try_substitute_return_value): Adjust dump output.
37789 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
37791 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
37793 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
37795 * doc/invoke.texi (SPARC options): Document -mlra as the default.
37796 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
37797 -mlra/-mno-lra was passed to the compiler.
37799 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
37801 PR rtl-optimization/65618
37802 * emit-rtl.c (try_split): Move initialization of "before" and
37803 "after" to just before the call to emit_insn_after_setloc.
37805 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
37807 * doc/md.texi (Standard Names): Remove reference to Java frontend.
37809 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
37811 * dwarf2out.c (gen_enumeration_type_die): When
37812 -gno-strict-dwarf, add a DW_AT_encoding attribute.
37814 2017-01-03 Jakub Jelinek <jakub@redhat.com>
37816 PR tree-optimization/78965
37817 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
37818 Change first argument from const call_info & to call_info &. For %n
37819 set info.nowrite to false.
37821 PR middle-end/78901
37822 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
37823 possibly throwing calls.
37825 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
37826 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
37827 and fns handling, rather than in a separate case SSA_NAME.
37829 2017-01-02 Jeff Law <law@redhat.com>
37831 * config/darwin-driver.c (darwin_driver_init): Const-correctness
37832 fixes for first_period and second_period variables.
37834 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
37837 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
37838 (*insvqi_1): New insn pattern.
37839 (*insvqi_1_mem_rex64): Ditto.
37840 (*insvqi_2): Ditto.
37841 (*insvqi_3): Rename from *insvqi.
37843 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
37845 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
37847 * doc/cfg.texi (Edges): Remove reference to Java.
37848 (Maintaining the CFG): Ditto.
37850 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37852 PR middle-end/77674
37853 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
37854 transparent aliases.
37856 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37858 PR middle-end/77484
37859 * predict.def (PRED_CALL): Update hitrate.
37860 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
37861 * predict.c (tree_estimate_probability_bb): Split CALL predictor
37862 into direct/indirect/polymorphic variants.
37864 2017-01-01 Jakub Jelinek <jakub@redhat.com>
37866 Update copyright years.
37868 * gcc.c (process_command): Update copyright notice dates.
37869 * gcov-dump.c (print_version): Ditto.
37870 * gcov.c (print_version): Ditto.
37871 * gcov-tool.c (print_version): Ditto.
37872 * gengtype.c (create_file): Ditto.
37873 * doc/cpp.texi: Bump @copying's copyright year.
37874 * doc/cppinternals.texi: Ditto.
37875 * doc/gcc.texi: Ditto.
37876 * doc/gccint.texi: Ditto.
37877 * doc/gcov.texi: Ditto.
37878 * doc/install.texi: Ditto.
37879 * doc/invoke.texi: Ditto.
37881 Copyright (C) 2017 Free Software Foundation, Inc.
37883 Copying and distribution of this file, with or without modification,
37884 are permitted in any medium without royalty provided the copyright
37885 notice and this notice are preserved.