1 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3 * gimple.h (enum gimple_debug_subcode): Add
4 GIMPLE_DEBUG_BEGIN_STMT.
5 (gimple_debug_begin_stmt_p): New.
6 (gimple_debug_nonbind_marker_p): New.
7 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
8 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
9 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
10 * insn-notes.def (BEGIN_STMT): New.
11 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
12 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
13 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
14 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
15 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
16 (INSN_DEBUG_MARKER_KIND): New.
17 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
18 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
19 (INSN_VAR_LOCATION_PTR): New.
20 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
22 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
23 deps only if debug bind insns are enabled.
24 (pass_expand::execute): Avoid deep TER and expand
25 debug locations for debug bind insns only.
26 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
27 debug stmts special handling down to debug bind stmts.
28 * combine.c (try_combine): Narrow debug insns special handling
29 down to debug bind insns.
30 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
31 Narrow debug insns preexisting special handling down to debug
33 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
34 handling down to debug bind insns.
35 * function.c (instantiate_virtual_regs): Skip debug markers,
36 adjust handling of debug binds.
37 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
38 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
39 * haifa-sched.c (schedule_insn): Narrow special handling of debug
40 insns to debug bind insns.
41 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
42 special handling of debug stmts to debug bind stmts.
43 * ipa-split.c (split_function): Likewise.
44 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
45 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
47 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
48 * regrename.c (build_def_use): Likewise.
49 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
50 (pass_cprop_hardreg): Narrow special casing of debug insns to
52 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
53 * reload1.c (reload): Likewise.
54 * sese.c (sese_insert_phis_for_liveouts): Narrow special
55 casing of debug stmts to debug bind stmts.
56 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
57 * ssa-iterators.h (num_imm_uses): Likewise.
58 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
59 debug stmts to debug bind stmts.
60 * tree-inline.c (tree_function_versioning): Narrow special casing
61 of debug stmts to debug bind stmts.
62 * tree-loop-distribution.c (generate_loops_for_partition):
63 Narrow special casing of debug stmts to debug bind stmts.
64 * tree-sra.c (analyze_access_subtree): Narrow special casing
65 of debug stmts to debug bind stmts.
66 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
67 stmts to debug bind stmts.
68 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
69 casing of debug stmts to debug bind stmts.
70 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
71 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
72 casing of debug stmts to debug bind stmts.
73 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
75 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
76 debug stmts to debug bind stmts.
77 (gimple_replace_ssa_lhs): Likewise.
78 (insert_debug_temp_for_var_def): Likewise.
79 (insert_debug_temps_for_defs): Likewise.
80 (reset_debug_uses): Likewise.
81 * tree-ssanames.c (release_ssa_name_fn): Likewise.
82 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
83 (adjust_debug_stmts): Likewise.
84 (adjust_phi_and_debug_stmts): Likewise.
85 (vect_do_peeling): Likewise.
86 * tree-vect-loop.c (vect_transform_loop): Likewise.
87 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
88 * var-tracking.c (adjust_mems): Narrow special casing of debug
89 insns to debug bind insns.
90 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
91 (compute_bb_dataflow, vt_find_locations): Likewise.
92 (vt_expand_loc, emit_notes_for_changes): Likewise.
93 (vt_init_cfa_base): Likewise.
94 (vt_emit_notes): Likewise.
95 (vt_initialize): Likewise.
96 (vt_finalize): Likewise.
98 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
99 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
100 (next_nonnote_nondebug_insn_bb): New.
101 (prev_nonnote_nondebug_insn_bb): New.
102 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
103 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
104 (prev_nonnote_nondebug_insn_bb): Declare.
105 (next_nonnote_nondebug_insn_bb): Declare.
106 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
107 * cfgrtl.c (get_last_bb_insn): Likewise.
108 * lra.c (push_insns): Likewise.
110 2017-12-11 David Malcolm <dmalcolm@redhat.com>
113 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
115 * selftest.c (selftest::test_runner::test_runner): New ctor.
116 (selftest::test_runner::~test_runner): New dtor.
117 * selftest.h (class selftest::test_runner): New class.
119 2017-12-11 Carl Love <cel@us.ibm.com>
121 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
122 vec_extract_fp32_from_shortl]): Add #defines.
123 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
124 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
125 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
126 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
127 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
128 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
129 * doc/extend.texi: Add documentation for the added builtins.
131 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
133 PR rtl-optimization/80693
134 PR rtl-optimization/81019
135 PR rtl-optimization/81020
136 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
137 are not mentioned in i3. Place the REG_UNUSED note on i2,
138 possibly modified to REG_DEAD, if it did not originate in i3.
140 2017-12-11 Jakub Jelinek <jakub@redhat.com>
142 * recog.c (store_data_bypass_p_1): New function.
143 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
144 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
146 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
148 PR rtl-optimization/83361
149 * ifcvt.c (if_convert): Call fixup_partitions.
151 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
153 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
154 early folding of splat_u{8,16,32}.
156 2017-12-11 Jakub Jelinek <jakub@redhat.com>
158 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
159 output_operand_lossage first argument with capital letter.
160 (aarch64_override_options): Don't start error and sorry first argument
163 2017-12-11 Andi Kleen <ak@linux.intel.com>
165 PR gcov-profile/83355
166 * auto-profile.c (string_table::get_index_by_decl): Don't
167 recurse when abstract origin points to itself.
169 2017-12-11 Bin Cheng <bin.cheng@arm.com>
171 PR tree-optimization/83320
172 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
173 (prune_datarefs_not_in_loop): Ditto.
175 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
177 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
179 2017-12-10 Jakub Jelinek <jakub@redhat.com>
181 PR tree-optimization/83337
182 * gimple-loop-interchange.cc (compute_access_stride): Handle
183 bitfield DRs properly.
185 2017-12-09 Jakub Jelinek <jakub@redhat.com>
187 PR tree-optimization/83338
188 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
189 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
190 vector of unsigned integers to vector of signed integers.
192 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
194 PR rtl-optimization/83317
195 * lra-constraints.c (process_address_1): Add insn code check.
197 2017-12-08 Michael Matz <matz@suse.de>
199 Fix PR tree-optimization/83323
200 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
201 head-controlled loops and loop BBs.
202 * common.opt (funroll-and-jam): Remove, instead ...
203 (floop-unroll-and-jam): ... reuse this option.
204 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
205 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
206 (-floop-unroll-and-jam): ... this option.
208 2017-12-08 Jakub Jelinek <jakub@redhat.com>
210 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
211 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
212 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
214 2017-12-08 Richard Biener <rguenther@suse.de>
217 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
218 handle accesses outside of zero-sized vars.
220 2017-12-08 Martin Jambor <mjambor@suse.cz>
222 PR tree-optimization/83141
223 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
224 test for MEM_REFs implicitely changing types with padding. Remove
226 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
228 2017-12-08 Jakub Jelinek <jakub@redhat.com>
230 * config/arc/arc.c (arc_attribute_table): Add exclusions to
232 * config/avr/avr.c (avr_attribute_table): Likewise.
233 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
234 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
235 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
236 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
237 of struct attribute_spec.
238 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
240 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
243 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
245 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
248 * profile-count.c (profile_count::from_gcov_type): Move from
249 profile-count.h; handle overflow.
250 * profile-count.h (profile_count::from_gcov_type): Move offline.
252 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
254 PR rtl-optimization/83304
255 * combine.c (move_deaths): If we do not know where a register died,
258 2017-12-08 Richard Biener <rguenther@suse.de>
260 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
261 Provide -fopt-info-loop feedback when we interchange in a nest.
263 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
265 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
266 for armv6 ARM CPU IDs.
268 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
270 * common/config/arm/arm-common.c: Include <algorithm>.
271 (INCLUDE_VECTOR): Define.
272 (compare_opt_names): New function.
273 (arm_rewrite_selected_arch): Only strip out extensions that can be
274 expressed through -mfpu. Sort the remaining extensions
277 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
279 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
280 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
281 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
282 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
284 2017-06-08 Tristan Gingold <gindold@adacore.com>
287 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
288 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
290 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
292 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
293 of struct attribute_spec.
295 2017-12-08 Julia Koval <julia.koval@intel.com>
297 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
298 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
299 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
300 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
301 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
302 _mm_maskz_dpwssds_epi32): Ditto.
304 2017-12-08 Richard Biener <rguenther@suse.de>
306 PR tree-optimization/81303
307 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
308 conditions try to create a comparison vector type matching
309 the data vector type.
310 (vectorizable_condition): Adjust.
311 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
312 Leave invariant conditions alone in case we can vectorize those.
314 2017-12-08 Julia Koval <julia.koval@intel.com>
316 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
317 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
318 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
319 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
320 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
322 2017-12-08 Julia Koval <julia.koval@intel.com>
324 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
325 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
326 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
327 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
328 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
329 _mm_maskz_dpbusds_epi32): New intrinsics.
331 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
333 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
335 (ld<bh>io_signed): Likewise.
336 (st<bhw>io): Likewise.
337 * config/nios2/predicates.md (ldstio_memory_operand): Allow
338 SMALL_INT12 constant integer operand.
340 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
342 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
344 (nios2_large_constant_allowed): ...this. Adjust uses.
345 (nios2_plus_symbolic_constant_p): Rename to...
346 (nios2_plus_large_constant_p): ...this. Adjust uses.
347 (nios2_legitimate_address_p): Correct CONST_INT handling.
348 (nios2_symbolic_memory_operand_p): Rename to...
349 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
350 (nios2_large_constant_p): Check for large constant integers too.
351 (nios2_split_large_constant): Handle constant integers.
352 (nios2_split_symbolic_memory_operand): Rename to...
353 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
354 (nios2_legitimize_constant_address): Handle constant integers.
355 (r0rel_constant_p): Handle small constant integers.
356 (nios2_print_operand_address): Handle r0-relative integer addresses.
357 * config/nios2/nios2-protos.h: Adjust for renamed functions.
358 * config/nios2/nios2.md: Adjust for renamed functions.
360 2017-12-07 Andrew Waterman <andrew@sifive.com>
362 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
363 (TARGET_HAVE_SRODATA_SECTION): New define.
364 (riscv_select_section): New function.
366 2017-12-08 Joseph Myers <joseph@codesourcery.com>
367 Alexander Monakov <amonakov@ispras.ru>
368 Jakub Jelinek <jakub@redhat.com>
371 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
373 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
375 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
376 a bool success value. Don't call output_operand_lossage here.
377 (aarch64_print_ldpstp_address): Return a bool success value.
378 (aarch64_print_operand_address): Call output_addr_const if
379 aarch64_print_address_internal fails.
380 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
381 'y'; call output_operand_lossage instead. Call output_operand_lossage
382 if aarch64_print_ldpstp_address fails.
384 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
386 * tree-vector-builder.h
387 (tree_vector_builder::binary_encoded_nelts): Declare.
388 * tree-vector-builder.c
389 (tree_vector_builder::binary_encoded_nelts): New function.
390 * fold-const.c (negate_expr_p): Likewise.
391 (operand_equal_p, fold_checksum_tree): Likewise.
392 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
393 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
394 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
395 (uniform_vector_p): Likewise.
396 * varasm.c (const_hash_1, compare_constant): Likewise.
397 * tree-ssa-ccp.c: Include tree-vector-builder.h.
398 (valid_lattice_transition): Operate directly on the VECTOR_CST
400 * ipa-icf.c: Include tree-vector-builder.h.
401 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
402 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
404 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
406 * tree.c (build_vector): Delete.
407 * tree.h (build_vector): Make static and move into the self-testing
410 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
412 * vector-builder.h (vector_builder::derived): New const overload.
413 (vector_builder::elt): New function.
414 * tree-vector-builder.h (tree_vector_builder::type): New function.
415 (tree_vector_builder::apply_step): Declare.
416 * tree-vector-builder.c (tree_vector_builder::apply_step): New
418 * gimple-fold.h (tree_vector_builder): Declare.
419 (gimple_build_vector): Take a tree_vector_builder instead of a
420 type and vector of elements.
421 * gimple-fold.c (gimple_build_vector): Likewise.
422 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
424 (get_initial_defs_for_reduction): Likewise.
425 (vectorizable_induction): Likewise.
427 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
429 * tree-vector-builder.h
430 (tree_vector_builder::new_binary_operation): Declare.
431 * tree-vector-builder.c
432 (tree_vector_builder::new_binary_operation): New function.
433 * fold-const.c (fold_relational_const): Use it.
434 (const_binop): Likewise. Check that both input vectors have
435 the same number of elements, thus excluding things like WIDEN_SUM.
436 Check whether it is possible to operate directly on the encodings
439 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
441 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
442 new_unary_operation, operating only on the encoded elements.
443 (const_unop): Likewise.
444 (exact_inverse): Likewise.
445 (distributes_over_addition_p): New function.
446 (const_binop): Use tree_vector_builder and new_unary_operation
447 for combinations of VECTOR_CST and INTEGER_CST. Operate only
448 on the encoded elements unless the encoding is strided and the
449 operation does not distribute over addition.
450 (fold_convert_const): Use tree_vector_builder and
451 new_unary_operation. Operate only on the encoded elements
452 for truncating integer conversions, or for non-stepped encodings.
454 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
456 * config/sparc/sparc.c: Include tree-vector-builder.h.
457 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
458 * expmed.c: Include tree-vector-builder.h.
459 (make_tree): Use tree_vector_builder instead of build_vector.
460 * fold-const.c: Include tree-vector-builder.h.
461 (const_binop): Use tree_vector_builder instead of build_vector.
462 (const_unop): Likewise.
463 (native_interpret_vector): Likewise.
464 (fold_vec_perm): Likewise.
465 (fold_ternary_loc): Likewise.
466 * gimple-fold.c: Include tree-vector-builder.h.
467 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
469 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
470 (simplify_vector_constructor): Use tree_vector_builder instead
472 * tree-vect-generic.c: Include tree-vector-builder.h.
473 (add_rshift): Use tree_vector_builder instead of build_vector.
474 (expand_vector_divmod): Likewise.
475 (optimize_vector_constructor): Likewise.
476 * tree-vect-loop.c: Include tree-vector-builder.h.
477 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
478 of build_vector. Explicitly use a stepped encoding for
480 * tree-vect-slp.c: Include tree-vector-builder.h.
481 (vect_get_constant_vectors): Use tree_vector_builder instead
483 (vect_transform_slp_perm_load): Likewise.
484 (vect_schedule_slp_instance): Likewise.
485 * tree-vect-stmts.c: Include tree-vector-builder.h.
486 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
487 (vect_gen_perm_mask_any): Likewise.
488 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
489 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
491 (build_vector_from_val): Likewise. Explicitly use a duplicate
494 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
496 * doc/generic.texi (VECTOR_CST): Describe new representation of
498 * vector-builder.h: New file.
499 * tree-vector-builder.h: Likewise.
500 * tree-vector-builder.c: Likewise.
501 * Makefile.in (OBJS): Add tree-vector-builder.o.
502 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
503 * tree-core.h (tree_base): Add a vector_cst field to the u union.
504 (tree_vector): Change the number of elements to
505 vector_cst_encoded_nelts.
506 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
507 (VECTOR_CST_ELTS): Delete.
508 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
509 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
510 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
511 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
512 (VECTOR_CST_ENCODED_ELT): Likewise.
513 (vector_cst_encoded_nelts): New function.
514 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
515 VECTOR_CST_NELTS_PER_PATTERN as arguments.
516 (vector_cst_int_elt, vector_cst_elt): Declare.
517 * tree.c: Include tree-vector-builder.h.
518 (tree_code_size): Abort if passed VECTOR_CST.
519 (tree_size): Update for new VECTOR_CST layout.
520 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
521 VECTOR_CST_NELTS_PER_PATTERN as arguments.
522 (build_vector): Use tree_vector_builder.
523 (vector_cst_int_elt, vector_cst_elt): New functions.
524 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
525 encoded elements and then create the vector in the canonical form.
526 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
527 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
528 (tree_c_tests): Call test_vector_cst_patterns.
529 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
531 (hash_tree): Likewise.
532 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
533 (streamer_write_tree_header): Likewise.
534 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
535 (streamer_alloc_tree): Likewise. Update call to make_vector.
536 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
538 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
540 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
541 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
542 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
544 2017-12-07 Bin Cheng <bin.cheng@arm.com>
545 Richard Biener <rguenther@suse.de>
547 PR tree-optimization/81303
548 * Makefile.in (gimple-loop-interchange.o): New object file.
549 * common.opt (floop-interchange): Reuse the option from graphite.
550 * doc/invoke.texi (-floop-interchange): Ditto. New document for
551 -floop-interchange and mention it for -O3.
552 * opts.c (default_options_table): Enable -floop-interchange at -O3.
553 * gimple-loop-interchange.cc: New file.
554 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
555 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
556 * passes.def (pass_linterchange): New pass.
557 * timevar.def (TV_LINTERCHANGE): New time var.
558 * tree-pass.h (make_pass_linterchange): New declaration.
559 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
560 interchange. Record IV before/after increment in new parameters.
561 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
562 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
564 (check_reduction_path): ...New function here.
565 * tree-vectorizer.h (check_reduction_path): New declaration.
567 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
570 PR rtl-optimization/80818
571 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
572 always early clobbered.
573 * lra-lives.c (process_bb_lives): Check input hard regs for early
574 clobbered non-operand hard reg.
576 2017-12-07 Jakub Jelinek <jakub@redhat.com>
579 * tree-cfg.c (verify_gimple_assign_binary): Don't require
580 types_compatible_p, just that TYPE_MODE is the same.
582 2017-12-07 Martin Sebor <msebor@redhat.com>
585 * attribs.c (empty_attribute_table): Initialize new member of
586 struct attribute_spec.
587 (decl_attributes): Add argument. Handle mutually exclusive
588 combinations of attributes.
589 (selftests::test_attribute_exclusions): New function.
590 (selftests::attribute_c_tests): Ditto.
591 * attribs.h (decl_attributes): Add default argument.
592 * selftest.h (attribute_c_tests): Declare.
593 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
594 * tree-core.h (attribute_spec::exclusions, exclude): New type and
596 * doc/extend.texi (Common Function Attributes): Update const and pure.
597 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
598 of struct attribute_spec.
599 * config/arc/arc.c (arc_attribute_table): Same.
600 * config/arm/arm.c (arm_attribute_table): Same.
601 * config/avr/avr.c ( avr_attribute_table): Same.
602 * config/bfin/bfin.c (bfin_attribute_table): Same.
603 * config/cr16/cr16.c (cr16_attribute_table): Same.
604 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
605 * config/h8300/h8300.c (h8300_attribute_table): Same.
606 * config/i386/i386.c (ix86_attribute_table): Same.
607 * config/ia64/ia64.c (ia64_attribute_table): Same.
608 * config/m32c/m32c.c (m32c_attribute_table): Same.
609 * config/m32r/m32r.c (m32r_attribute_table): Same.
610 * config/m68k/m68k.c (m68k_attribute_table): Same.
611 * config/mcore/mcore.c (mcore_attribute_table): Same.
612 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
613 * config/mips/mips.c (mips_attribute_table): Same.
614 * config/msp430/msp430.c (msp430_attribute_table): Same.
615 * config/nds32/nds32.c (nds32_attribute_table): Same.
616 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
617 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
618 * config/rl78/rl78.c (rl78__attribute_table): Same.
619 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
620 * onfig/rx/rx.c (rx_attribute_table): Same.
621 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
622 * config/sh/sh.c (sh_attribute_table): Same.
623 * config/sparc/sparc.c (sparc_attribute_table): Same.
624 * config/spu/spu.c (spu_attribute_table): Same.
625 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
626 * config/v850/v850.c (v850_attribute_table): Same.
627 * config/visium/visium.c (visium_attribute_table): Same.
629 2017-12-07 Tamar Christina <tamar.christina@arm.com>
632 * config/arm/arm.c (INCLUDE_STRING): Define.
633 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
634 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
637 2017-12-07 Michael Matz <matz@suse.de>
639 Add unroll and jam pass
641 * gimple-loop-jam.c: New file.
642 * Makefile.in (OBJS): Add gimple-loop-jam.o.
643 * common.opt (funroll-and-jam): New option.
644 * opts.c (default_options_table): Add unroll-and-jam at -O3.
645 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
646 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
647 * passes.def: Add pass_loop_jam.
648 * timevar.def (TV_LOOP_JAM): Add.
649 * tree-pass.h (make_pass_loop_jam): Declare.
650 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
651 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
652 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
653 to flow_loop_tree_node_add.
654 (duplicate_subloops, copy_loops_to): Append to sibling list.
655 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
656 * doc/invoke.texi (-funroll-and-jam): Document new option.
657 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
659 2017-12-07 Richard Biener <rguenther@suse.de>
661 PR tree-optimization/83296
662 PR tree-optimization/67769
663 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
664 flow sensitive info in an unrelated BB.
665 (value_replacement): Use reset_flow_sensitive_info.
666 (minmax_replacement): Reset flow sensitive info on the def
667 we move. Do not reset flow sensitive info in the whole BB
669 (abs_replacement): Likewise.
671 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
674 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
675 rs6000_cpu to the given -mcpu=, or to the default processor.
677 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
679 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
680 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
681 instead of rs6000_cpu_attr.
682 (rs6000_adjust_cost): Ditto.
683 (is_microcoded_insn): Ditto.
684 (rs6000_adjust_priority): Ditto.
685 (rs6000_issue_rate): Ditto.
686 (rs6000_use_sched_lookahead): Ditto.
687 (rs6000_use_sched_lookahead_guard): Ditto.
688 (rs6000_sched_reorder): Ditto.
689 (force_new_group): Ditto.
690 * config/rs6000/rs6000.md (cpu attribute): Ditto.
691 (group_ending_nop): Ditto.
693 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
695 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
696 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
697 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
698 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
700 (rs6000_reassociation_width): Ditto.
701 (rs6000_emit_epilogue): Ditto.
702 (rs6000_adjust_cost): Ditto.
703 (is_microcoded_insn): Ditto.
704 (is_cracked_insn): Ditto.
705 (rs6000_adjust_priority): Ditto.
706 (rs6000_sched_reorder): Ditto.
707 (rs6000_sched_reorder2): Ditto.
708 (insn_must_be_first_in_group): Ditto.
709 (insn_must_be_last_in_group): Ditto.
710 (rs6000_register_move_cost): Ditto.
711 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
714 2017-12-07 Julia Koval <julia.koval@intel.com>
716 * config.gcc: Add vaesintrin.h.
717 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
719 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
720 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
722 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
723 * config/i386/immintrin.h: Include vaesintrin.h.
724 * config/i386/sse.md (vaesdec_<mode>): New pattern.
725 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
726 _mm_aesdec_epi128): New intrinsics.
728 2017-12-06 David Malcolm <dmalcolm@redhat.com>
730 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
731 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
732 c-family/c-spellcheck.cc.
733 (best_macro_match::best_macro_match): Likewise.
735 (struct edit_distance_traits<cpp_hashnode *>): Move to
736 c-family/c-spellcheck.h.
737 (class best_macro_match): Likewise.
739 2017-12-06 Jakub Jelinek <jakub@redhat.com>
741 PR tree-optimization/83293
742 * gimple-ssa-strength-reduction.c (insert_initializers): Use
743 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
744 might insert into empty bb.
747 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
748 simplify for plus with :c added, and pointer_plus without that.
749 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
750 with undefined overflow and the conversion is not widening,
751 perform negation in utype and only convert to type afterwards.
752 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
753 simplify for plus with :c added, and pointer_plus without that.
754 If type is integral with undefined overflow and the conversion is
755 not widening, perform minus in utype and only convert to type
756 afterwards. Move the last pointer_diff_expr simplify into the
759 2017-12-06 Martin Sebor <msebor@redhat.com>
761 PR tree-optimization/82646
762 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
765 2017-12-06 Martin Sebor <msebor@redhat.com>
767 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
770 PR tree-optimization/83075
771 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
772 strncat/strncpy don't change length of source string.
774 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
777 2017-11-29 Martin Aberg <maberg@gaisler.com>
779 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
780 to prevent b2bst errata sequence.
781 (sqrtdf2_fix): Likewise.
783 2017-12-06 Jakub Jelinek <jakub@redhat.com>
785 PR tree-optimization/81945
786 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
787 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
788 to dest_cfun has orig_loop_num set, either remap it to the new
789 loop number if the loop got moved too, or clear it.
791 2017-12-05 Steve Ellcey <sellcey@cavium.com>
793 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
795 (thunderx2t99_nothing): New insn reservation.
796 (thunderx2t99_mrs): New insn reservation.
797 (thunderx2t99_multiple): New insn reservation.
798 (thunderx2t99_alu_basi): Add bfx to reservation.
799 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
801 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
804 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
806 2017-12-05 Bin Cheng <bin.cheng@arm.com>
808 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
809 tree-ssa-pre.c::remove_dead_inserted_code.
810 * tree-ssa-dce.h: New file.
811 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
812 (remove_dead_inserted_code): Move and rename to function
813 tree-ssa-dce.c::simple_dce_from_worklist.
814 (pass_pre::execute): Update use.
816 2017-12-05 Richard Biener <rguenther@suse.de>
818 PR tree-optimization/83277
819 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
820 to code-gen liveout vars.
822 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
824 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
825 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
826 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
829 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
831 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
833 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
834 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
837 2017-12-05 Richard Biener <rguenther@suse.de>
839 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
840 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
841 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
842 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
844 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
846 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
848 2017-12-05 Martin Liska <mliska@suse.cz>
849 Jakub Jelinek <jakub@redhat.com>
851 * doc/invoke.texi: Document the options.
852 * flag-types.h (enum sanitize_code): Add
853 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
854 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
855 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
856 * opts.c: Define new sanitizer options.
857 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
858 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
860 2017-12-05 Julia Koval <julia.koval@intel.com>
862 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
863 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
864 (ix86_handle_option): Handle -mavx512vnni.
865 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
866 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
867 * config/i386/i386-c (__AVX512VNNI__): New.
868 * config/i386/i386.c (ix86_target_string): Handle new option.
869 (ix86_valid_target_attribute_inner_p): Handle new option.
870 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
871 * config/i386/i386.opt (mavx512vnni): New option.
873 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
876 * athlon.md: Disable for generic.
877 * haswell.md: Enable for generic.
878 * i386.c (ix86_sched_init_global): Add core hooks for generic.
879 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
881 (ix86_adjust_cost): Move generic to haswell path.
883 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
885 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
886 instead of MEM_P in a couple more places. Fix formatting issues.
888 2017-12-04 Jim Wilson <jimw@sifive.com>
890 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
891 instead of GP_REG_LAST-1.
892 (riscv_adjust_libcall_cfi_prologue): Likewise.
893 (riscv_adjust_libcall_cri_epilogue): Likewise.
894 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
897 2017-12-04 Luis Machado <luis.machado@linaro.org>
899 * ipa-pure-const.c (check_decl): Add missing newline.
900 (state_from_flags): Likewise.
902 2017-12-04 Jeff Law <law@redhat.com>
904 PR tree-optimizatin/78496
905 * gimple-ssa-evrp-analyze.h
906 (evrp_range_analyzer::get_vr_values): Simplify.
907 * gimple-ssa-evrp-analyze.c: Corresponding changes.
908 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
909 and gimple-ssa-evrp-analyze.h.
910 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
911 (simplify_stmt_for_jump_threading): Copy a blob of code from
912 tree-vrp.c to use ranges to simplify statements.
913 (dom_opt_dom_walker::before_dom_children): Call
914 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
915 (dom_opt_dom_walker::after_dom_children): Similarly for
916 evrp_range_analyzer::leave.
917 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
920 * gimple-ssa-evrp-analyze.c
921 (evrp_range_analyzer::extract_range_from_stmt): Always use
922 vr_values::update_value_range so preexisting range info is
923 medged with new range info, even if the new range is VR_VARYING.
925 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
927 * combine.c: Adjust comment.
928 (use_crosses_set_p): Delete.
929 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
930 (try_combine): Ditto.
932 2017-12-04 Richard Biener <rguenther@suse.de>
934 PR tree-optimization/83255
935 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
936 Re-add zero-iteration check.
938 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
940 PR rtl-optimization/83245
941 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
942 hard registers as earlyclobber, also if not in an asm.
944 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
948 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
951 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
952 rs6000_cpu based on cpu_index, not tune_index.
954 2017-12-04 Richard Biener <rguenther@suse.de>
956 PR tree-optimization/83238
957 * graphite-scop-detection.c (scop_detection::merge_sese): Make
958 code match comment, rejecting invalid SESE regions.
960 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
962 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
963 require base operand is a REG_POINTER prior to reload on targets
964 with non-equivalent space registers.
966 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
968 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
969 (update_specialized_profile): Fix updating of counts.
970 (perhaps_add_new_callers): Likewise.
972 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
975 * x86-tune.def: Remove obsolette FIXMEs.
976 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
977 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
978 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
980 (X86_TUNE_PAD_RETURNS): Disable for generic.
982 2017-12-02 Jakub Jelinek <jakub@redhat.com>
984 PR tree-optimization/83170
985 PR tree-optimization/83241
986 * gimple-ssa-store-merging.c
987 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
988 gimple_vuse (ins_stmt) in case it has changed.
989 (imm_store_chain_info::output_merged_store): Likewise.
991 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
995 * tree-cfg.c (pass_warn_function_return::execute): Handle
996 __builtin_ubsan_handle_missing_return like __builtin_unreachable
997 with BUILTINS_LOCATION.
1001 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1002 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1003 instead of DECL_MODE.
1005 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1006 last argument to rtx pointer.
1007 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1008 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1009 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1010 depending on the chosen ISAs.
1011 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1012 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1013 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1015 * config/i386/sse.md (mov<mode>_internal): Likewise.
1016 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1018 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1020 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1022 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1024 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1025 parameters from prototype.
1026 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1027 parameters. Don't print a comment.
1028 (emit_fusion_gpr_load): Adjust.
1029 (emit_fusion_load_store): Adjust.
1030 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1031 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1032 comment on the second line.
1034 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1037 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1038 rs6000_cpu based on cpu_index, not tune_index.
1040 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1042 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1043 which_alternative instead of which_alternative + 1.
1044 (output_asm_insn): Print an extra tab if the template is short.
1046 2017-12-01 Jim Wilson <jimw@sifive.com>
1048 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1050 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1051 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1053 * doc/tm.texi: Regenerate.
1055 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1058 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1059 whether we can allocate pseudos before trying to fix an address.
1060 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1061 memory address is indexed or indirect.
1062 (floatuns_<mode>si2_hw2): Likewise.
1064 2017-12-01 Jason Merrill <jason@redhat.com>
1066 * Makefile.in (TAGS): Add c-family/*.cc.
1068 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1070 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1071 (call_value_insn): Likewise.
1072 (sibcall_insn): Likewise.
1073 (sibcall_value_insn): Likewise.
1074 (movsi_aarch64): Likewise.
1075 (movdi_aarch64): Likewise.
1076 (add_losym_): Likewise.
1077 (ldr_got_small_): Likewise.
1078 (ldr_got_small_sidi): Likewise.
1079 (ldr_got_small_28k_): Likewise.
1080 (ldr_got_small_28k_sidi): Likewise.
1081 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1082 Move output_addr_const to symbolic case. Add error check.
1084 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1086 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1087 * config/s390/s390.md: Disable MVC merging peephole if it would
1088 disable operand forwarding.
1089 (new peephole2): Split MVCs if it would turn them into up to 2
1092 2017-12-01 Richard Biener <rguenther@suse.de>
1094 PR tree-optimization/83232
1095 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1096 detection of same access. Instead of breaking the group here
1097 do not consider the duplicate. Add comment explaining real fix.
1099 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1101 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1103 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1105 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1107 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1109 * function.h (struct function): Remove cilk_frame_decl,
1110 is_cilk_function and calls_cilk_spawn fields.
1111 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1113 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1114 cilk_elemental field.
1115 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1116 * target.def: Adjust comment.
1117 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1118 Don't test cilk_elemental.
1120 PR tree-optimization/83233
1121 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1122 bswap_stat name for the struct.
1125 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1126 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1127 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1128 SWITCH_BREAK_LABEL_P set on the label.
1129 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1130 added for default case if it was missing and not all cases covered.
1131 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1132 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1134 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1135 to avoid -Wimplicit-fallthrough warning.
1136 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1137 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1139 PR tree-optimization/83221
1140 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1142 (init_reassoc): Formatting fix.
1145 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1146 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1150 * tree-cfg.c: Include asan.h.
1151 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1153 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1155 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1157 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1159 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1160 vpcmpeqd instruction.
1162 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1164 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1165 registers type for 128bit mode.
1167 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1169 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1171 * gengtype-state.c (read_state_pair): Likewise.
1172 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1173 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1174 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1175 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1176 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1177 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1178 * builtins.c (fold_builtin_3): Likewise.
1179 * graphite-scop-detection.c
1180 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1181 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1184 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1187 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1188 parameter. Return true if ignore_decl_rtl_set_p is true and other
1189 conditions are satisfied.
1190 * asan.h (asan_protect_global): Add new parameter.
1191 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1192 to asan_protect_global calls.
1194 2017-11-30 Jim Wilson <jimw@sifive.com>
1196 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1197 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1198 -mstrict-align, add info on default value. Delete redundant lines for
1199 -mabi. Add missing -mexplicit-relocs docs.
1201 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1203 * config/arc/arc.md (trap): New pattern.
1205 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1207 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1208 instruction to end into a delay slot.
1209 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1210 can be placed into a delay slot against reg_note.
1212 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1214 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1215 labels number of usages.
1217 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1219 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1221 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1223 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1226 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1227 multiplication by power of 2 constant into two shifts + comparison.
1229 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1232 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1234 2017-11-30 Richard Biener <rguenther@suse.de>
1236 PR tree-optimization/83202
1237 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1238 allow_peel argument and guard peeling.
1239 (canonicalize_loop_induction_variables): Likewise.
1240 (canonicalize_induction_variables): Pass false.
1241 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1242 peeling from cunrolli.
1244 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1246 * combine.c (try_combine): Print a message to dump file whenever
1247 I0, I1, or I2 cannot be combined into I3.
1249 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1251 PR rtl-optimization/83156
1252 PR rtl-optimization/82621
1253 * combine.c (try_combine): Don't split an I2 if one of the dests is
1254 set again before I3. Allow unused dests.
1256 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1258 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1260 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1262 PR rtl-optimization/80818
1263 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1264 recursively. Use insn code for clobber.
1265 (lra_set_insn_recog_data): Pass the new arg to
1266 collect_non_operand_hard_regs.
1267 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1269 (lra_update_insn_regno_info): Pass insn to
1270 add_regs_to_insn_regno_info.
1272 2017-11-29 Jim Wilson <jimw@sifive.com>
1273 Andrew Waterman <andrew@sifive.com>
1275 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1276 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1277 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1278 use SINGLE_SHIFT_COST.
1279 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1280 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1282 2017-11-29 Julia Koval <julia.koval@intel.com>
1284 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1285 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1286 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1287 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1288 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1289 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1290 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1291 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1292 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1293 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1294 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1295 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1296 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1297 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1298 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1299 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1300 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1301 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1302 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1303 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1304 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1305 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1306 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1307 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1308 __builtin_ia32_vpshldv_v2di_mask,
1309 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1310 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1311 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1313 2017-11-29 Julia Koval <julia.koval@intel.com>
1315 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1316 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1317 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1318 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1319 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1320 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1321 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1322 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1323 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1324 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1325 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1326 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1327 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1328 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1329 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1330 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1331 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1332 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1333 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1334 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1335 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1337 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1339 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1340 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1343 2017-11-29 Martin Aberg <maberg@gaisler.com>
1345 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1346 to prevent b2bst errata sequence.
1347 (sqrtdf2_fix): Likewise.
1349 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1351 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1352 (div_sqrt_insn_p): New function.
1353 (sparc_do_work_around_errata): Insert NOP instructions to
1354 prevent sequences that could trigger the TN-0013 errata for
1355 certain LEON3 processors.
1356 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1357 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1358 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1359 (in_branch_delay): Prevent div and sqrt in delay slot if
1361 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1363 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1365 * config/sparc/sparc.c (atomic_insn_p): New function.
1366 (sparc_do_work_around_errata): Insert NOP instructions to
1367 prevent sequences that could trigger the TN-0010 errata for
1369 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1370 instruction referable in atomic_insns_p.
1372 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1374 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1375 (atomic_compare_and_swap_leon3_1): Likewise.
1378 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1380 * config/sparc/sparc.c (fpop_insn_p): New function.
1381 (sparc_do_work_around_errata): Insert NOP instructions to
1382 prevent sequences that could trigger the TN-0012 errata for
1384 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1385 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1386 (in_branch_annul_delay): Prevent floating-point instructions
1387 in delay slot of annulled integer branch.
1389 2017-11-29 Richard Biener <rguenther@suse.de>
1391 PR tree-optimization/83202
1392 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1394 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1395 nodes vectorized to the same stmts multiple times.
1396 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1397 (vect_analyze_slp_instance): Adjust.
1398 (scalar_stmts_to_slp_tree_map_t): New typedef.
1399 (vect_schedule_slp_instance): Add a map recording the SLP node
1400 representing the vectorized stmts for a set of scalar stmts.
1401 Avoid code-generating redundancies.
1402 (vect_schedule_slp): Allocate map and pass it down.
1404 2017-11-29 Nathan Sidwell <nathan@acm.org>
1407 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1410 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1412 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1413 cases for printing LDP/STP memory addresses.
1414 (aarch64_print_address_internal): Renamed from
1415 aarch64_print_operand_address, added parameter, add Pmode check.
1416 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1417 (aarch64_print_operand_address): Indirect to
1418 aarch64_print_address_internal.
1419 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1420 'y' operand output specifier.
1422 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1425 * tree.c (build_simple_mem_ref_loc): Handle
1426 get_addr_base_and_unit_offset returning a MEM_REF.
1429 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1432 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1435 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1437 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1439 (layout::move_to_column): Likewise.
1441 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1444 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1445 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1447 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1448 Martin Jambor <mjambor@suse.cz>
1451 * tree.h (expr_type_first_operand_type_p): Declare
1452 * tree.c (expr_type_first_operand_type_p): New function.
1453 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1454 (ipa_value_from_jfunc): Adjust declaration.
1455 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1456 Use it as result type for arithmetics, unless it is NULL in which case
1457 be more conservative.
1458 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1459 ipa_get_jf_pass_through_result.
1460 (propagate_vals_across_pass_through): Likewise.
1461 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1462 is to propagate_vals_across_pass_through.
1463 (propagate_constants_across_call): Pass PARM_TYPE to
1464 propagate_scalar_across_jump_function.
1465 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1466 ipa_value_from_jfunc.
1467 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1468 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1469 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1470 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1471 target_type, pass it to ipa_value_from_jfunc.
1472 (update_indirect_edges_after_inlining): Pass parameter type to
1473 try_make_edge_direct_simple_call.
1475 2017-11-28 Jeff Law <law@redhat.com>
1477 * gimple-ssa-evrp-analyze.c
1478 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1479 refine ranges if scev_initialized_p returns true.
1480 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1482 2017-11-28 Julia Koval <julia.koval@intel.com>
1484 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1485 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1486 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1487 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1488 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1489 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1490 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1491 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1492 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1493 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1494 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1495 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1496 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1497 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1498 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1499 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1500 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1501 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1502 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1503 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1504 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1505 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1507 2017-11-28 Julia Koval <julia.koval@intel.com>
1509 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1510 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1511 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1512 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1513 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1514 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1515 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1516 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1517 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1518 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1519 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1520 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1521 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1522 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1523 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1524 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1525 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1526 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1527 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1528 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1529 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1530 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1531 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1532 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1533 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1534 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1535 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1536 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1537 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1538 __builtin_ia32_vpshld_v2di_mask): New builtins.
1539 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1540 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1542 2017-11-28 Richard Biener <rguenther@suse.de>
1544 PR tree-optimization/80776
1545 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1547 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1549 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1550 If the incoming edge is an effective fallthru because the other
1551 edge only reaches a __builtin_unreachable () then record ranges
1552 derived from the controlling condition in SSA info.
1553 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1554 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1556 2017-11-28 Olivier Hainque <hainque@adacore.com>
1558 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1559 on mingw build hosts.
1561 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1564 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1566 2017-11-28 Richard Biener <rguenther@suse.de>
1569 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1570 copies generated from memcpy use a character array as reference
1573 2017-11-28 Julia Koval <julia.koval@intel.com>
1574 Sebastian Peryt <sebastian.peryt@intel.com>
1576 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1577 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1578 cilk-common.o, cilk.h, cilk-common.c): Remove.
1580 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1581 * builtins.c (is_builtin_name): Remove cilkplus condition.
1582 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1583 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1584 cilk-builtins.def, cilkplus.def): Remove.
1585 * cif-code.def (CILK_SPAWN): Remove.
1586 * cilk-builtins.def: Delete.
1587 * cilk-common.c: Ditto.
1589 * cilkplus.def: Ditto.
1590 * config/darwin.h (fcilkplus): Delete.
1591 * cppbuiltin.c: Ditto.
1592 * doc/extend.texi: Remove cilkplus doc.
1593 * doc/generic.texi: Ditto.
1594 * doc/invoke.texi: Ditto.
1595 * doc/passes.texi: Ditto.
1596 * gcc.c (fcilkplus): Remove.
1597 * gengtype.c (cilk.h): Remove.
1598 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
1600 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
1602 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
1604 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
1605 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
1606 cilkplus conditions.
1607 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
1608 inline_read_section): Ditto.
1609 * ipa-inline-analysis.c (cilk.h): Remove.
1610 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
1611 * lto-wrapper.c (merge_and_complain, append_compiler_options,
1612 append_linker_options): Remove condition for fcilkplus.
1613 * lto/lto-lang.c (cilk.h): Remove.
1614 (lto_init): Remove condition for fcilkplus.
1615 * omp-expand.c (expand_cilk_for_call): Delete.
1616 (expand_omp_taskreg, expand_omp_for_static_chunk,
1617 expand_omp_for): Remove cilkplus
1619 (expand_cilk_for): Delete.
1620 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
1621 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
1622 execute_lower_omp, diagnose_sb_0): Ditto.
1623 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
1624 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1625 * tree-nested.c: Ditto.
1626 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
1627 (dump_generic_node): Ditto.
1628 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1629 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
1631 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
1633 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1635 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
1636 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
1637 for vector mode and !TARGET_SIMD.
1639 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1641 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
1643 * tree.h (SWITCH_LABELS): Remove.
1644 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
1645 assert SWITCH_BODY is non-NULL.
1646 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
1648 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
1650 PR tree-optimization/80788
1651 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
1652 has TREE_OVERFLOW set, call drop_tree_overflow.
1654 2017-11-28 Richard Biener <rguenther@suse.de>
1656 PR tree-optimization/83158
1657 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
1659 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
1662 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
1663 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
1665 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
1667 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
1668 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
1670 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1673 * varasm.c (compare_constant): Take the mode of the constants into
1674 account when comparing floating point constants.
1676 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
1678 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
1679 from explicit instantiation of debug_helper.
1680 * vec.h (DEFINE_DEBUG_VEC): Ditto.
1682 2017-11-27 Richard Biener <rguenther@suse.de>
1684 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
1687 2017-11-27 Richard Biener <rguenther@suse.de>
1689 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
1690 (type_hash_canon): Also clear int_cst_hash_table entry for
1692 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
1694 2017-11-27 Tamar Christina <tamar.christina@arm.com>
1696 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1697 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1698 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1700 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1702 * hash-map.h (gt_cleare_cache): Avoid UB.
1704 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
1706 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
1707 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
1708 (decide_unroll_runtime_iterations): Likewise.
1709 (decide_unroll_stupid): Likewise.
1711 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1714 * config/i386/i386.md: Add a loop with incssp.
1716 2017-11-27 Martin Jambor <mjambor@suse.cz>
1718 PR tree-optimization/81248
1719 * tree-sra.c (splice_param_accesses): Remove size check.
1720 (decide_one_param_reduction): Fix size check.
1721 * gimple-pretty-print.c (dump_profile): Silence warning.
1722 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
1724 2017-11-27 Jakub Jelinek <jakub@redhat.com>
1727 * dbxout.c (lastlineno): New variable.
1728 (dbx_debug_hooks): Use dbxout_switch_text_section as
1729 switch_text_section debug hook.
1730 (dbxout_function_end): Switch to current_function_section
1731 rather than function_section. If crtl->has_bb_partition,
1732 output just one N_FUN, depending on in_cold_section_p.
1733 (dbxout_source_line): Remember last lineno in lastlineno.
1734 (dbxout_switch_text_section): New function.
1735 (dbxout_function_decl): Adjust dbxout_block caller.
1736 (dbx_block_with_cold_children): New function.
1737 (dbxout_block): Return true if any LBRAC/RBRAC have been
1738 emitted. Use dbx_block_with_cold_children at depth == 0
1739 in second partition. Add PARENT_BLOCKNUM argument, pass
1740 it optionally adjusted to children. Output LBRAC/RBRAC
1741 around recursive call only if the block is in the current
1742 partition, if not and anything was output, emit empty
1744 * final.c (final_scan_insn): Compute cold_function_name
1745 before calling switch_text_section debug hook. Call
1746 that hook even if dwarf2out_do_frame if not emitting
1750 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
1751 TREE_READONLY decls.
1753 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1755 PR rtl-optimization/82488
1756 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
1758 2017-11-26 Julia Koval <julia.koval@intel.com>
1760 * config/i386/i386.c (processor_target_table): Add skylake_cost for
1762 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
1765 2017-11-26 Julia Koval <julia.koval@intel.com>
1767 * config/i386/driver-i386.c (host_detect_local_cpu):
1768 Detect skylake-avx512.
1770 2017-11-26 Julia Koval <julia.koval@intel.com>
1772 * config.gcc: Add -march=cannonlake.
1773 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
1774 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
1775 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
1776 (PTA_CANNONLAKE): New.
1777 (processor_target_table): Add cannonlake.
1778 (ix86_option_override_internal): Ditto.
1779 (fold_builtin_cpu): Ditto.
1780 (get_builtin_code_for_version): Handle cannonlake.
1781 (M_INTEL_COREI7_CANNONLAKE): New.
1782 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
1783 * doc/invoke.texi: Add -march=cannonlake.
1785 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
1787 * plugin.c (add_new_plugin): Use platform-specific library extensions.
1788 (try_init_one_plugin): Alternative implementation for MinGW.
1789 * Makefile.in (plugin_implib): New.
1790 (gengtype-lex.c): Fix broken AIX workaround.
1791 * configure: Regenerate.
1792 * doc/plugins.texi: Document support for MinGW.
1794 2017-11-25 Jakub Jelinek <jakub@redhat.com>
1796 PR rtl-optimization/81553
1797 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
1798 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
1799 is a shift where C1 has different mode than the whole shift, use C1's
1800 mode for MULT rather than the shift's mode.
1803 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
1804 builtins not enabled in the currently selected ISA.
1806 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
1808 PR tree-optimization/71026
1809 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
1810 (insert_reciprocals): Change to insert reciprocals before a division
1811 by a square and to insert the square of a reciprocal.
1812 (execute_cse_reciprocals_1): Change to consider division by a square.
1813 (register_division_in): Add importance parameter.
1815 2017-11-24 Richard Biener <rguenther@suse.de>
1817 PR tree-optimization/82402
1818 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
1819 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
1821 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
1823 * match.pd (0-ptr): New transformation.
1825 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
1828 * ipa-inline.c (inline_small_functions): Set current badnes correctly
1829 when skipping checking.
1831 2017-11-24 Richard Biener <rguenther@suse.de>
1833 PR tree-optimization/83128
1834 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
1835 (vn_reference_lookup_3): Likewise.
1837 2017-11-24 Jakub Jelinek <jakub@redhat.com>
1840 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
1841 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
1844 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
1845 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
1846 check for the latter with gimple_call_builtin_p. Do not handle
1847 BUILT_IN_STPNCPY_CHK which is not a pass through call.
1849 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
1851 * config/arm/arm_neon.h: Fix pragma GCC push_options before
1854 2017-11-23 Julia Koval <julia.koval@intel.com>
1856 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
1857 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
1858 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
1859 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
1860 _mm512_maskz_expandloadu_epi16): New intrinsics.
1861 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
1862 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
1863 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
1864 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
1865 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
1866 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
1867 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
1868 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
1869 _mm256_maskz_expandloadu_epi8): New intrinsics.
1870 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
1871 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
1872 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
1873 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
1874 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
1875 * config/i386/sse.md (VI248_VLBW): New iterator.
1876 (expand<mode>_mask, expand<mode>_maskz): New patterns.
1878 2017-11-23 Julia Koval <julia.koval@intel.com>
1880 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
1881 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
1882 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
1883 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
1884 _mm512_mask_compressstoreu_epi16): New.
1885 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
1886 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
1887 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
1888 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
1889 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
1890 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
1891 _mm256_mask_compressstoreu_epi8): New.
1892 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
1893 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
1894 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
1895 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
1896 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
1897 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
1898 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
1899 __builtin_ia32_compresshi128_mask,
1900 __builtin_ia32_compressstoreuqi512_mask,
1901 __builtin_ia32_compressstoreuhi512_mask,
1902 __builtin_ia32_compressstoreuqi256_mask,
1903 __builtin_ia32_compressstoreuqi128_mask,
1904 __builtin_ia32_compressstoreuhi256_mask,
1905 __builtin_ia32_compressstoreuhi128_mask): New builtins.
1906 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
1908 (ix86_expand_special_args_builtin): Handle new types.
1909 (s4fma_expand): Handle new builtin array.
1910 * config/i386/immintrin.h: Include new headers.
1911 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
1912 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
1914 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1916 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
1917 fixes. Declare temp and g variables at the top in order to avoid
1918 {} in most of the cases.
1920 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
1922 * match.pd (ptr-0): New transformation.
1924 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1926 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
1927 (struct addr_mode_cost_table): New.
1928 (struct tune_params): Add field addr_mode_costs.
1929 * config/arm/arm.c (generic_addr_mode_costs): New.
1930 (arm_slowmul_tune): Initialise addr_mode_costs field.
1931 (arm_fastmul_tune): Likewise.
1932 (arm_strongarm_tune): Likewise.
1933 (arm_xscale_tune): Likewise.
1934 (arm_9e_tune): Likewise.
1935 (arm_marvell_pj4_tune): Likewise.
1936 (arm_v6t2_tune): Likewise.
1937 (arm_cortex_tune): Likewise.
1938 (arm_cortex_a8_tune): Likewise.
1939 (arm_cortex_a7_tune): Likewise.
1940 (arm_cortex_a15_tune): Likewise.
1941 (arm_cortex_a35_tune): Likewise.
1942 (arm_cortex_a53_tune): Likewise.
1943 (arm_cortex_a57_tune): Likewise.
1944 (arm_exynosm1_tune): Likewise.
1945 (arm_xgene1_tune): Likewise.
1946 (arm_cortex_a5_tune): Likewise.
1947 (arm_cortex_a9_tune): Likewise.
1948 (arm_cortex_a12_tune): Likewise.
1949 (arm_cortex_a73_tune): Likewise.
1950 (arm_v7m_tune): Likewise.
1951 (arm_cortex_m7_tune): Likewise.
1952 (arm_v6m_tune): Likewise.
1953 (arm_fa726te_tune): Likewise.
1954 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
1956 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1958 * config/arm/arm.c (arm_mem_costs): New function.
1959 (arm_rtx_costs_internal): Use arm_mem_costs.
1961 2017-11-23 Mark Wielaard <mark@klomp.org>
1963 * dwarf2out.c (init_sections_and_labels): Use generation to create
1964 unique ranges_section_label and ranges_base_label. Return generation.
1965 (output_rnglists): Add generation argument. Use generation to create
1966 unique ranges labels.
1967 (dwarf2out_finish): Get generation from init_sections_and_labels
1968 and pass generation to output_rnglists.
1970 2017-11-23 Mike Stump <mikestump@comcast.net>
1971 Eric Botcazou <ebotcazou@adacore.com>
1973 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
1974 * cfgloop.h (struct loop): Add unroll field.
1975 * function.h (struct function): Add has_unroll bitfield.
1976 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
1977 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
1978 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
1979 (pass_rtl_unroll_loops::gate): Likewise.
1980 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
1981 for which loop->unroll==1.
1982 (decide_unroll_constant_iterations): Use note for consistency and
1983 take loop->unroll into account. Return early if loop->unroll is set.
1984 Fix thinko in existing test.
1985 (decide_unroll_runtime_iterations): Use note for consistency and
1986 take loop->unroll into account.
1987 (decide_unroll_stupid): Likewise.
1988 * lto-streamer-in.c (input_cfg): Read loop->unroll.
1989 * lto-streamer-out.c (output_cfg): Write loop->unroll.
1990 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
1992 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
1993 (print_loop): Print loop->unroll if set.
1994 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
1995 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
1996 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
1998 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
1999 loop->unroll is set and smaller than the trip count. Otherwise bypass
2000 entirely the heuristics if loop->unroll is set. Remove dead note.
2001 Fix off-by-one bug in other note.
2002 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2003 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2005 (tree_unroll_loops_completely): Make static.
2006 (pass_complete_unroll::execute): Use correct type for variable.
2007 (pass_complete_unrolli::execute): Fix formatting.
2008 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2010 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2012 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2013 enable when TARGET_PREFER_AVX128 is set.
2015 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2017 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2019 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2020 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2022 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2024 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2025 frequencies. Fix estimation of aggregate parameters.
2027 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2029 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2032 2017-11-23 Tom de Vries <tom@codesourcery.com>
2034 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2036 2017-11-23 Tom de Vries <tom@codesourcery.com>
2038 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2041 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2044 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2045 sibcall_value_pcrel_fdpic): Use local variable instead of
2047 (calli_tbr_rel): Add missing operand 2.
2048 (call_valuei_tbr_rel): Add missing operand 3.
2050 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2053 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2054 bitpos/bitsize covering the whole destination, use store_expr only if
2055 the complex mode is the same. Otherwise, use expand_normal and if
2056 it returns CONCAT, subreg each part separately instead of trying to
2057 subreg the whole result.
2059 2017-11-23 Richard Biener <rguenther@suse.de>
2061 PR tree-optimization/23094
2062 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2063 come here from walking over backedges in the first iteration.
2064 (vn_reference_lookup_3): Skip clobbers that store the same value.
2066 2017-11-23 Richard Biener <rguenther@suse.de>
2068 PR tree-optimization/81403
2069 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2070 a block we need a leader relative to.
2071 (phi_translate_1): For nary processing require a leader from
2072 get_representative_for given we run expression simplification
2073 using match-and-simplify. Remove previous fix.
2075 2017-11-22 Jeff Law <law@redhat.com>
2077 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2078 Use new method allocate_value_range rather than accessing the
2079 vrp_value_range_pool data member directly.
2080 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2081 to use extract_range_from_stmt method to avoid need for
2082 extract_range_from_assignment method.
2083 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2084 method rather than setting values_propgated data member directly.
2085 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2086 and values propagated data members and extract_range_from_assignment
2087 method. Reorder private data members to conform to standards.
2088 Add new methods set_lattice_propagation_complete and
2089 allocate_value_range.
2091 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2093 PR rtl-optimization/83030
2094 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2095 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2096 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2097 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2098 CROSSING_JUMP_P flag is already set before setting it.
2099 * cfgrtl.c (fixup_partition_crossing): Likewise.
2100 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2103 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2105 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2106 Handle the case where both arguments are using gen_const_vec_series.
2108 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2111 * pretty-print.c (pp_format): Move quoting implementation to
2112 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2113 to pass address of "quote" local.
2114 (pp_begin_quote): New function.
2115 (pp_end_quote): New function.
2116 * pretty-print.h (printer_fn): Convert penultimate param from bool
2118 (pp_begin_quote): New decl.
2119 (pp_end_quote): New decl.
2120 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2121 param from bool to bool *.
2122 * tree-diagnostic.h (default_tree_printer): Likewise.
2124 2017-11-22 Jeff Law <law@redhat.com>
2126 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2127 Initialize vr_values.
2128 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2129 vr_values via vr_values class instance rather than delegators.
2130 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2131 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2132 (evrp_range_analyzer::push_value_range): Likewise.
2133 (evrp_range_analyzer::pop_value_range): Likewise.
2134 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2135 most delegators. Those remaining are exposed as public interfaces.
2136 Make vr_values a pointer and private.
2137 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2139 (evrp_range_analyzer::get_vr_value): New method.
2140 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2141 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2142 (class evrp_dom_walker): Attach evrp_folder class, initialize
2143 it in the ctor. Remove temporary delegators.
2144 (evrp_dom_walker::before_dom_children): Call methods in attached
2145 evrp_range_analyzer class via class instance pointer. Use
2146 free value_range_constant_singleton to remove need for
2147 op_with_constant_singleton_value delegator method. Do not
2148 create a vrp_prop class instance for every call! Narrow
2149 scope of a couple variables.
2150 (evrp_dom_walker::cleanup): Call methods in attached
2151 evrp_range_analyzer class via class instance pointer.
2152 * vr-values.h (class vr_values): Privatize many methods and
2155 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2157 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2158 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2159 * expr.c (expand_expr_real_2): Likewise.
2160 * fold-const.c (const_unop): Likewise.
2161 * optabs-tree.c (optab_for_tree_code): Likewise.
2162 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2163 * tree-inline.c (estimate_operator_cost): Likewise.
2164 * tree-pretty-print.c (dump_generic_node): Likewise.
2165 (op_code_prio): Likewise.
2166 (op_symbol_code): Likewise.
2167 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2168 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2169 * internal-fn.c (direct_internal_fn_optab): New function.
2170 (direct_internal_fn_array, direct_internal_fn_supported_p
2171 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2172 * fold-const-call.c (fold_const_reduction): New function.
2173 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2175 * tree-vect-loop.c: Include internal-fn.h.
2176 (reduction_code_for_scalar_code): Rename to...
2177 (reduction_fn_for_scalar_code): ...this and return an internal
2179 (vect_model_reduction_cost): Take an internal_fn rather than
2181 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2183 (vectorizable_reduction): Use internal functions rather than tree
2184 codes for the reduction operation. Update calls to the functions
2186 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2187 Use calls to internal functions rather than REDUC tree codes.
2188 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2190 2017-11-22 Olivier Hainque <hainque@adacore.com>
2192 * config/vxworks.c (vxworks_override_options): Pick default
2193 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2194 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2195 DWARF_GNAT_ENCODINGS_DEFAULT.
2196 * config/vxworksae.h: Likewise.
2198 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2200 PR tree-optimization/83104
2201 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2204 2017-11-22 Marek Polacek <polacek@redhat.com>
2205 H.J. Lu <hongjiu.lu@intel.com>
2206 Jason Merrill <jason@redhat.com>
2211 * calls.c (initialize_argument_information): Call
2212 warn_parameter_passing_abi target hook.
2213 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2214 argument onto stack.
2215 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2216 * common.opt: Update -fabi-version description.
2217 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2218 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2220 (ix86_is_empty_record): New function.
2221 (ix86_warn_parameter_passing_abi): New function.
2222 (TARGET_EMPTY_RECORD_P): Redefine.
2223 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2224 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2225 * doc/tm.texi: Regenerated.
2226 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2227 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2228 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2229 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2230 instead of int_size_in_bytes.
2231 * expr.c (copy_blkmode_to_reg): Likewise.
2232 * function.c (aggregate_value_p): Return 0 for empty types.
2233 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2235 (locate_and_pad_parm): Call arg size_in_bytes instead
2237 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2238 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2239 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2241 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2242 (std_gimplify_va_arg_expr): Skip empty records. Call
2243 arg_size_in_bytes instead size_in_bytes.
2244 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2245 * tree-core.h (tree_type_common): Add empty_flag.
2246 (tree_decl_common): Update comments.
2247 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2249 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2250 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2252 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2253 * tree.c (default_is_empty_type): New function.
2254 (default_is_empty_record): New function.
2255 (arg_int_size_in_bytes): New function.
2256 (arg_size_in_bytes): New function.
2257 (get_ultimate_context): New function.
2258 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2259 TRANSLATION_UNIT_WARN_EMPTY_P.
2260 (default_is_empty_record, arg_int_size_in_bytes,
2261 arg_size_in_bytes, get_ultimate_context): Declare.
2263 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2265 * config/arm/arm.c (cmse_clear_registers): New function.
2266 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2267 code by call to cmse_clear_registers.
2268 (cmse_nonsecure_entry_clear_before_return): Likewise.
2270 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2272 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2273 (vdot_s32, vdotq_s32): New.
2274 (vdot_lane_u32, vdotq_lane_u32): New.
2275 (vdot_lane_s32, vdotq_lane_s32): New.
2278 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2281 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2282 for unsigned values with fewer HWIs than the precision.
2283 (test_overflow): New function.
2284 (wide_int_cc_tests): Call it.
2286 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2288 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2289 has the same mode class as Pmode.
2291 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2293 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2294 padding_bits_to_clear_ptr.
2295 (cmse_nonsecure_entry_clear_before_return): Likewise.
2297 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2299 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2300 auto_sbitap instead of integer bitfield to control register needing
2303 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2305 PR tree-optimization/83044
2306 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2307 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2308 that need the upper bound. Subtract offset from
2309 get_addr_base_and_unit_offset only if positive and subtract it
2310 before division by eltsize rather than after it.
2313 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2314 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2315 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2319 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2320 expand_binop with *mul_widen_optab, make sure at least one of the
2321 operands doesn't have VOIDmode.
2324 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2326 PR rtl-optimization/82044
2327 PR tree-optimization/82042
2328 * dse.c (record_store): Check for overflow.
2329 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2330 add_wild_read instead of clear_rhs_from_active_local_stores on
2331 overflow and log it into dump_file.
2333 2017-11-22 Richard Biener <rguenther@suse.de>
2335 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2336 fake edges to exit when looking for a place to insert.
2337 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2339 (insert_into_preds_of_block): Commit edge insertion immediately,
2340 assert that doesn't require new BBs.
2341 (fini_pre): Release expressions.
2342 (pass_pre::execute): Shuffle things around a bit, if the fn
2343 is too large do not compute AVAIL either as this is really the
2346 2017-11-22 Richard Biener <rguenther@suse.de>
2348 PR tree-optimization/83089
2349 * tree-if-conv.c (pass_if_conversion::execute): If anything
2350 changed reset SCEV and free the number of iteration estimates.
2352 2017-11-21 Martin Sebor <msebor@redhat.com>
2354 PR tree-optimization/82945
2355 * calls.h (warn_nonstring_bound): Remove unused function.
2357 2017-11-21 Martin Sebor <msebor@redhat.com>
2359 PR tree-optimization/82945
2360 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2361 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2362 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2364 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2365 * calls.h (get_attr_nonstring_decl): Declare new function.
2366 * doc/extend.texi (attribute nonstring): Update.
2367 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2368 get_attr_nonstring_decl and handle it.
2369 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2370 detection of nul-termination.
2371 (strlen_to_stridx): Change to a pointer.
2372 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2373 (pass_strlen::execute): Same.
2375 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2377 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2378 for the new option -mprefer-vector-width=[none|128|256|512].
2379 * config/i386/i386.c (ix86_target_string): remove old style options
2380 -mprefer-avx256 and make -mprefer-avx128 as alias.
2381 (ix86_option_override_internal): Apply defaults for the
2382 -mprefer-vector-width=[128|256] option.
2383 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2384 Implement macros to work with -mprefer-vector-width=.
2385 * config/i386/i386.opt: Implemented option
2386 -mprefer-vector-width=[none|128|256|512].
2387 * doc/invoke.texi: Documentation for
2388 -mprefer-vector-width=[none|128|256|512].
2390 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2392 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2394 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2396 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2398 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2400 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2402 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2403 * expr.c (expand_expr_real_2): Likewise.
2404 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2405 fold_binary_loc): Likewise.
2406 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2407 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2408 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2409 MINUS_EXPR transformations.
2410 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2411 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2412 * tree-inline.c (estimate_operator_cost): Likewise.
2413 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2414 op_symbol_code): Likewise.
2415 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2416 * vr-values.c (extract_range_from_binary_expr): Likewise.
2417 * varasm.c (initializer_constant_valid_p_1): Likewise.
2418 * tree.def: New tree code POINTER_DIFF_EXPR.
2420 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2422 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2423 integer suffix to movbe mnemonic.
2424 (*bswaphi2_movbe): Ditto.
2425 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2427 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2430 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2431 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2432 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2434 2017-11-21 Martin Liska <mliska@suse.cz>
2436 * tree-inline.c (expand_call_inline): Remove not needed
2439 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2440 Jakub Jelinek <jakub@redhat.com>
2443 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2444 Remove static keyword from f variable.
2446 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2448 PR tree-optimization/83086
2449 * gimple-ssa-store-merging.c
2450 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2451 rather than n.base_addr.
2453 2017-11-21 Martin Liska <mliska@suse.cz>
2455 PR rtl-optimization/82044
2456 PR tree-optimization/82042
2457 * dse.c (check_mem_read_rtx): Check for overflow.
2459 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2461 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2464 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2466 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2467 and its corresponding call arg location note.
2469 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2470 Andrew Burgess <andrew.burgess@embecosm.com>
2472 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2474 (arc_return_slot_offset): Likewise.
2475 (arc_eh_return_address_location): New declaration.
2476 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2477 (MUST_SAVE_REGISTER): Add exception handler case.
2478 (MUST_SAVE_RETURN_ADDR): Likewise.
2479 (arc_frame_pointer_required): Likewise.
2480 (arc_frame_pointer_needed): New function.
2481 (arc_compute_frame_size): Changed.
2482 (arc_expand_prologue): Likewise.
2483 (arc_expand_epilogue): Likewise.
2484 (arc_initial_elimination_offset): Likewise.
2485 (arc_return_slot_offset): Delete.
2486 (arc_eh_return_address_location): New function.
2487 (arc_builtin_setjmp_frame_value): Likewise.
2488 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2489 (EH_RETURN_STACKADJ_RTX): Define.
2490 (EH_RETURN_HANDLER_RTX): Likewise.
2491 * config/arc/arc.md (eh_return): Delete.
2493 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2495 * print-tree.h (debug_vec_tree): Remove prototype.
2496 * gdbinit.in (pvt): Remove macro.
2498 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2500 PR tree-optimization/83047
2501 * gimple-ssa-store-merging.c
2502 (imm_store_chain_info::output_merged_store): If the loads with the
2503 same vuse are in different basic blocks, for load_gsi pick a load
2504 location that is dominated by the other loads.
2507 * config/i386/i386.c (ix86_memmodel_check): Start
2508 -Winvalid-memory-model diagnostics with lowercase letter.
2511 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2512 set in_cold_section_p to the partition containing loc_list->first.
2513 When seeing loc_list->last_before_switch node, update secname and
2514 perform range_across_switch second partition handling only after that.
2517 * run-rtl-passes.c: Include debug.h.
2518 (run_rtl_passes): Call debug_hooks->assembly_start.
2519 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2523 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2524 OPTAB_DIRECT in calls to expand_simple_binop.
2526 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2529 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2531 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2534 * spellcheck.h (best_match::blithely_get_best_candidate): New
2537 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2539 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2540 with lower case letter.
2542 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2544 * config/i386/i386.md (bswaphi2): New expander.
2545 (*bswaphi2_movbe): New insn pattern.
2546 (bswaphi -> rorhi pepehole2): New peephole pattern.
2548 2017-11-20 Jeff Law <law@redhat.com>
2550 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2551 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2552 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2553 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2554 gimple-ssa-evrp-analyze.h.
2556 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2557 set BB_VISITED here.
2558 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2560 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2562 * tree-predcom.c: Add general comment on Store-Store chains.
2563 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2565 (get_chain_last_ref_at): Rename into...
2566 (get_chain_last_write_at): ...this.
2567 (get_chain_last_write_before_load): New function.
2568 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2569 CT_STORE_STORE when write reference is added.
2570 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2571 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2572 (initialize_root_vars_store_elim_1): Ditto.
2573 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2574 definition is created.
2575 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2576 chain by replacing it with dominant stored value.
2578 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2580 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2582 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2584 * vr-values.c (extract_range_from_binary_expr): Use a full range
2587 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2589 * config/arm/arm.md (R4_REGNUM): Define constant.
2590 (nonsecure_call_internal): Remove r4 clobber.
2591 (nonsecure_call_value_internal): Likewise.
2592 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2593 clobber and resequence match_operands.
2594 (nonsecure_call_value_reg_thumb1_v5): Likewise.
2595 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
2596 (nonsecure_call_value_reg_thumb2): Likewise.
2598 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2600 PR tree-optimization/78821
2601 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
2602 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
2603 to the address of the base rather than the base itself.
2604 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
2605 (find_bswap_or_nop_finalize): New function.
2606 (find_bswap_or_nop): Use it.
2607 (bswap_replace): Return a tree rather than bool, change first
2608 argument from gimple * to gimple_stmt_iterator, allow inserting
2609 into an empty sequence, allow ins_stmt to be NULL - then emit
2610 all stmts into gsi. Fix up MEM_REF address gimplification.
2611 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
2612 (struct store_immediate_info): Add N and INS_STMT non-static
2614 (store_immediate_info::store_immediate_info): Initialize them
2615 from newly added ctor args.
2616 (merged_store_group::apply_stores): Formatting fixes. Sort by
2618 (stmts_may_clobber_ref_p): For stores call also
2619 refs_anti_dependent_p.
2620 (gather_bswap_load_refs): New function.
2621 (imm_store_chain_info::try_coalesce_bswap): New method.
2622 (imm_store_chain_info::coalesce_immediate_stores): Use it.
2623 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
2624 (imm_store_chain_info::output_merged_store): Fail if number of
2625 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
2626 and NOP_EXPR rhs_code.
2627 (pass_store_merging::process_store): Compute n and ins_stmt, if
2628 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
2629 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
2631 (pass_store_merging::execute): Calculate dominators.
2633 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
2634 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
2635 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
2636 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2637 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
2638 class pass_optimize_bswap, bswap_replace,
2639 pass_optimize_bswap::execute): Moved to ...
2640 * gimple-ssa-store-merging.c: ... this file.
2641 Include optabs-tree.h.
2642 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
2643 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2644 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
2645 anonymous namespace, remove static keywords.
2646 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
2647 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
2649 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
2652 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
2654 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
2656 * vec.h (debug_helper): New function.
2657 (DEFINE_DEBUG_VEC): New macro.
2658 * hash-set.h (debug_helper): New function.
2659 (DEFINE_DEBUG_HASH_SET): New macro.
2660 * cfg.c (debug_slim (edge)): New function.
2661 Call DEFINE_DEBUG_VEC for edges.
2662 Call DEFINE_DEBUG_HASH_SET for edges.
2663 * cfghooks.c (debug_slim (basic_block)): New function.
2664 Call DEFINE_DEBUG_VEC for basic blocks.
2665 Call DEFINE_DEBUG_HASH_SET for basic blocks.
2666 * print-tree.c (debug_slim): New function to handle trees.
2667 Call DEFINE_DEBUG_VEC for trees.
2668 Call DEFINE_DEBUG_HASH_SET for trees.
2669 (debug (vec<tree, va_gc>) &): Remove.
2670 (debug (<vec<tree, va_gc>) *): Remove.
2671 * print-rtl.c (debug_slim): New function to handle const_rtx.
2672 Call DEFINE_DEBUG_VEC for rtx_def.
2673 Call DEFINE_DEBUG_VEC for rtx_insn.
2674 Call DEFINE_DEBUG_HASH_SET for rtx_def.
2675 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
2676 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
2677 (debug (vec<rtx_insn *> *ptr): Remove.
2678 (debug_insn_vector): Remove.
2679 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
2681 2017-11-20 Tom de Vries <tom@codesourcery.com>
2683 PR rtl-optimization/82020
2684 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
2685 IF_THEN_ELSE condition.
2687 2017-11-19 Jeff Law <law@redhat.com>
2689 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2690 of degenerates resulting from ignoring an edge.
2692 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2695 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
2697 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2700 * profile-count.c (profile_count::to_sreal_scale): Fix return value
2701 for uninitialied counts.
2703 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2706 * tree-inline.c (estimate_num_insns): Set to 1 at least.
2708 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2711 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
2713 2017-11-19 Tom de Vries <tom@codesourcery.com>
2715 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
2718 2017-11-19 Tom de Vries <tom@codesourcery.com>
2720 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
2723 2017-11-19 Tom de Vries <tom@codesourcery.com>
2725 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2728 2017-11-19 Tom de Vries <tom@codesourcery.com>
2730 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
2732 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
2734 2017-11-19 Tom de Vries <tom@codesourcery.com>
2736 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
2738 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
2739 semicolon after MCORE_EXPORT_NAME call.
2741 2017-11-19 Tom de Vries <tom@codesourcery.com>
2744 * vmsdbgout.c (vmsdbgout_early_finish): New function.
2745 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
2747 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
2749 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
2750 (cgraph_edge::clone): Cleanup updating of profile.
2751 * ipa-cp.c (update_profiling_info): Likewise.
2752 * ipa-inline-transform.c (inline_transform): Likewise.
2753 * ipa-inline.c (inline_small_functions): Add missing space to dump.
2754 * ipa-split.c (execute_split_functions): Do not split when function
2756 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
2757 * profile-count.c (profile_count::dump): Add global0.
2758 (profile_count::to_cgraph_frequency): Do not ICE when entry is
2760 (profile_count::to_sreal_scale): Likewise.
2761 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
2762 (profile_count::combine_with_ipa_count): New function.
2763 * profile-count.h (profile_guessed_global0adjusted): New.
2764 (profile_count::adjusted_zero): New.
2765 (profile_count::global0adjusted): New.
2766 (profile_count::combine_with_ipa_count): New.
2767 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
2768 correct profile of return block of split functions.
2769 (copy_cfg_body): Remove unused profile_count.
2770 (copy_body): Likewise.
2771 (expand_call_inline): Update.
2772 (tree_function_versioning): Update.
2774 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
2776 * hash-set.h (hash_set::empty): New.
2777 * tree-ssa-threadbackward.h: Delete.
2778 * tree-ssa-threadbackward.c (class thread_jumps): New.
2779 Move max_threaded_paths into class.
2780 (fsm_find_thread_path): Remove arguments that are now in class.
2781 (profitable_jump_thread_path): Rename to...
2782 (thread_jumps::profitable_jump_thread_path): ...this.
2783 (convert_and_register_jump_thread_path): Rename to...
2784 (thread_jumps::convert_and_register_current_path): ...this.
2785 (check_subpath_and_update_thread_path): Rename to...
2786 (thread_jumps::check_subpath_and_update_thread_path): ...this.
2787 (register_jump_thread_path_if_profitable): Rename to...
2788 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
2789 (handle_phi): Rename to...
2790 (thread_jumps::handle_phi): ...this.
2791 (handle_assignment): Rename to...
2792 (thread_jumps::handle_assignment): ...this.
2793 (fsm_find_control_statement_thread_paths): Rename to...
2794 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
2795 (find_jump_threads_backwards): Rename to...
2796 (thread_jumps::find_jump_threads_backwards): ...this.
2797 Initialize path local data.
2798 (pass_thread_jumps::execute): Call find_jump_threads_backwards
2799 from within thread_jumps class.
2800 (pass_early_thread_jumps::execute): Same.
2802 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2804 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
2806 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2808 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
2809 to canonical location.
2811 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2814 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
2816 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
2818 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
2820 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
2821 rcrt1.o%s/grcrt1.o%s for -static-pie.
2823 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2825 * i386.c (ix86_multiplication_cost, ix86_division_cost,
2826 ix86_shift_rotate_cost): Break out from ...
2827 (ix86_rtx_costs): ... here.
2828 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
2831 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2833 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
2834 when entry block was promoted unlikely.
2835 (estimate_bb_frequencies): Increase frequency scale.
2836 * profile-count.h (profile_count): Export precision info.
2838 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2840 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
2841 disturbin profile of entry block.
2843 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2845 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
2848 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2850 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
2851 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
2853 2017-11-17 Jeff Law <law@redhat.com>
2855 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
2856 from evrp_dom_walker class. Various methods moved into new class.
2857 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
2858 (evrp_range_analyzer::enter): New method.
2859 (evrp_range_analyzer::leave): New method.
2860 (evrp_dom_walker): Remove delegators no longer needed by this class.
2861 Replace vr_values data member with evrp_range_analyzer
2863 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
2864 method extracted from evrp_dom_walker::before_dom_children.
2865 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
2866 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
2868 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
2869 Add private copy constructor and move assignment operators.
2870 Privatize methods and class data where trivially possible.
2871 (evrp_dom_walker::cleanup): New function, extracted from
2872 execute_early_vrp. Simplify access to class data.
2874 * vr-values.h (get_output_for_vrp): Prototype.
2875 * vr-values.c (get_output_for_vrp): New function extracted from
2876 vrp_visit_assignment_or_call and extract_range_from_stmt.
2877 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
2879 2017-11-17 Luis Machado <luis.machado@linaro.org>
2881 * config/aarch64/aarch64.c
2882 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
2883 (qdf24xx_tunings) <autoprefetcher_model>: Set to
2884 tune_params::AUTOPREFETCHER_WEAK.
2886 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2889 * config/arm/arm.c (arm_valid_target_attribute_rec):
2890 Parse "arch=" and "+<ext>".
2891 (arm_valid_target_attribute_tree): Re-init global options.
2892 (arm_option_override): Make non-static.
2893 (arm_options_perform_arch_sanity_checks): Make errors fatal.
2894 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
2895 (__ARM_FEATURE_CRC32): Support undef.
2896 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
2897 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
2899 2017-11-17 David Malcolm <dmalcolm@redhat.com>
2901 * gdbinit.in (break-on-diagnostic): New command.
2903 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2905 * config/i386/i386.c (ix86_expand_epilogue): Change simple
2906 return to indirect jump for EH return if control-flow
2907 protection is enabled. Change explicit 'false' argument in
2908 pro_epilogue_adjust_stack with a value of
2910 * config/i386/i386.md (simple_return_indirect_internal):
2911 Remove SImode restriction to support 64-bit.
2913 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2915 * combine.c (added_notes_insn): New.
2916 (try_combine): Handle added_notes_insn like added_links_insn.
2917 Rewrite return value code.
2918 (distribute_notes): Set added_notes_insn to the earliest insn we added
2921 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2923 PR rtl-optimization/82621
2924 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
2925 dest of one of those SETs is unused.
2927 2017-11-17 Richard Biener <rguenther@suse.de>
2930 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
2931 * tree-pretty-print.c (dump_generic_node): Handle
2932 annot_expr_parallel_kind.
2933 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
2934 * gimplify.c (gimple_boolify): Likewise.
2936 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2938 * config.gcc (extra_headers): Add cet.h for x86 targets.
2939 * config/i386/cet.h: New file.
2940 * doc/install.texi: Add --enable-cet/--disable-cet.
2942 2017-11-17 Richard Biener <rguenther@suse.de>
2944 PR tree-optimization/83017
2945 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
2946 (gen_parallel_loop): Properly count iterations.
2947 (parallelize_loops): Handle loop->can_be_parallel independent
2948 of flag_loop_parallelize_all. Make static profitability test match
2950 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
2951 * invoke.texi (parloops-min-per-thread): Document.
2953 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
2955 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
2956 upstreaming review comments.
2958 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2960 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
2961 with fast unaligned access.
2962 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
2964 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
2966 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
2967 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
2969 2017-11-17 Richard Biener <rguenther@suse.de>
2971 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
2972 folding of references.
2974 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
2977 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
2978 of replacing call to strncmp with corresponding call to strcmp when
2981 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
2983 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
2984 option prefer-avx256 for skylake-avx512 configuration.
2985 * config/i386/i386.c (ix86_option_override_internal): Ditto.
2986 (get_builtin_code_for_version): Ditto.
2988 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2989 Monk Chiang <sh.chiang04@gmail.com>
2991 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
2992 (FIXED_REGISTERS): Reserve more register numbers.
2993 (CALL_USED_REGISTERS): Likewise.
2994 (REG_ALLOC_ORDER): Likewise.
2995 (REG_CLASS_CONTENTS): Likewise.
2996 (REGISTER_NAMES): Likewise.
2998 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2999 Kito Cheng <kito.cheng@gmail.com>
3001 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3003 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3005 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3007 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3008 parameter list for vec_splats.
3010 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3012 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3014 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3016 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3019 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3020 with the required version of automake 1.11.6.
3022 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3024 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3026 (DU_C2_3_power9): Correct reservation combinations.
3027 (FP_DIV_power9, VEC_DIV_power9): New.
3028 (power9-alu): Split out rotate/shift...
3029 (power9-rot): ...to here, correct dispatch resource.
3030 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3032 (power9-fp): Correct latency.
3033 (power9-sdiv): Add div/sqrt resource.
3034 (power9-ddiv): Correct latency, add div/sqrt resource.
3035 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3036 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3038 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3040 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3042 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3043 switch statement mapping KF built-ins to TF built-ins if we don't
3044 have the proper ISA 3.0 assembler support.
3046 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3048 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3049 (lower_emutls_function_body): Do not compute it.
3051 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3053 * ipa-split.c (split_bb_info): Turn time to sreal.
3054 (split_point): Likewise.
3055 (dump_split_point): Likewise.
3056 (fine_split_points): Likewise.
3057 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3059 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3061 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3064 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3066 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3068 (expensive_function_p): Remove useless assert.
3069 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3071 2017-11-16 Martin Sebor <msebor@redhat.com>
3073 PR tree-optimization/82588
3074 PR tree-optimization/82583
3075 * tree-vrp.c (check_array_ref): Handle flexible array members,
3076 string literals, and inner indices.
3077 (search_for_addr_array): Add detail to diagnostics.
3079 2017-11-16 Nathan Sidwell <nathan@acm.org>
3083 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3084 (SET_DECL_ASSEMBLER_NAME): Forward to
3085 overwrite_decl_assembler_name.
3086 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3087 (overwrite_decl_assembler_name): Declare.
3088 * tree.c (overwrite_decl_assembler_name): New.
3089 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3090 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3091 (LANG_HOOKS_INITIALIZER): Add it.
3092 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3093 * langhooks.c (lhd_set_decl_assembler_name): Use
3094 SET_DECL_ASSEMBLER_NAME.
3095 (lhd_overwrite_decl_assembler_name): Default implementation.
3097 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3098 Jackson Woodruff <jackson.woodruff@arm.com>
3100 PR tree-optimization/71026
3101 * match.pd: Canonicalize constant multiplies in division.
3103 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3105 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3106 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3107 Remove OPT_fomit_frame_pointer entry.
3108 * common/config/arc/arc-common.c: Likewise.
3109 * common/config/arm/arm-common.c: Likewise.
3110 * common/config/avr/avr-common.c: Likewise.
3111 * common/config/c6x/c6x-common.c: Likewise.
3112 * common/config/cr16/cr16-common.c: Likewise.
3113 * common/config/cris/cris-common.c: Likewise.
3114 * common/config/epiphany/epiphany-common.c: Likewise.
3115 * common/config/fr30/fr30-common.c: Likewise.
3116 * common/config/frv/frv-common.c: Likewise.
3117 * common/config/ia64/ia64-common.c: Likewise.
3118 * common/config/iq2000/iq2000-common.c: Likewise.
3119 * common/config/lm32/lm32-common.c: Likewise.
3120 * common/config/m32r/m32r-common.c: Likewise.
3121 * common/config/mcore/mcore-common.c: Likewise.
3122 * common/config/microblaze/microblaze-common.c: Likewise.
3123 * common/config/mips/mips-common.c: Likewise.
3124 * common/config/mmix/mmix-common.c: Likewise.
3125 * common/config/mn10300/mn10300-common.c: Likewise.
3126 * common/config/nios2/nios2-common.c: Likewise.
3127 * common/config/pa/pa-common.c: Likewise.
3128 * common/config/pdp11/pdp11-common.c: Likewise.
3129 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3130 * common/config/riscv/riscv-common.c: Likewise.
3131 * common/config/rs6000/rs6000-common.c: Likewise.
3132 * common/config/rx/rx-common.c: Likewise.
3133 * common/config/s390/s390-common.c: Likewise.
3134 * common/config/sh/sh-common.c: Likewise.
3135 * common/config/sparc/sparc-common.c: Likewise.
3136 * common/config/tilegx/tilegx-common.c: Likewise.
3137 * common/config/tilepro/tilepro-common.c: Likewise.
3138 * common/config/v850/v850-common.c: Likewise.
3139 * common/config/visium/visium-common.c: Likewise.
3140 * common/config/xstormy16/xstormy16-common.c: Likewise.
3141 * common/config/xtensa/xtensa-common.c: Likewise.
3142 * invoke.texi (-fomit-frame-pointer): Update documentation.
3144 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3146 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3148 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3150 * cfg.c (scale_bbs_frequencies_int,
3151 cale_bbs_frequencies_gcov_type): Remove.
3152 * cfg.h (scale_bbs_frequencies_int,
3153 cale_bbs_frequencies_gcov_type): Remove.
3155 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3157 * tree-ssa-loop-manip.c
3158 (scale_dominated_blocks_in_loop): Update to profile counts.
3159 (tree_transform_and_unroll_loop): Likewise.
3161 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3163 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3164 scale_bbs_frequencies_int.
3166 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3168 * final.c (compute_alignments): Use counts rather than frequencies.
3170 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3172 * cfgloopanal.c: Include sreal.h
3173 (average_num_loop_insns): Use counts and sreal for accounting.
3175 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3177 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3180 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3181 Kito Cheng <kito.cheng@gmail.com>
3183 * config/nds32/constraints.md: Provide more constraints.
3184 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3185 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3186 support constraints usage.
3188 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3190 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3192 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3193 Kito Cheng <kito.cheng@gmail.com>
3195 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3196 * config/nds32/nds32.opt: Refine the layout.
3197 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3198 TARGET_EXT_STRING): Support new options.
3199 * config/nds32/nds32.h: Likewise.
3200 * config/nds32/nds32.md: Likewise.
3201 * config/nds32/nds32-predicates.c: Likewise.
3202 * config/nds32/constraints.md: Likewise.
3203 * common/config/nds32/nds32-common.c: Likewise.
3205 2017-11-16 Julia Koval <julia.koval@intel.com>
3208 * config/i386/gfniintrin.h: Add sse check.
3209 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3211 2017-11-16 Julia Koval <julia.koval@intel.com>
3213 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3214 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3215 (ix86_handle_option): Handle -mavx512vbmi2.
3216 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3217 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3218 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3219 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3220 (ix86_valid_target_attribute_inner_p): Ditto.
3221 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3222 * config/i386/i386.opt (mavx512vbmi2): New option.
3223 * doc/invoke.texi: Add new option.
3225 2017-11-16 Julia Koval <julia.koval@intel.com>
3227 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3228 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3229 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3230 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3231 _mm512_gf2p8mul_epi8): New intrinsics.
3232 * config/i386/i386-builtin-types.def
3233 (V64QI_FTYPE_V64QI_V64QI): New type.
3234 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3235 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3236 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3237 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3238 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3239 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3241 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3243 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3244 explicitly as a stream of bytes.
3246 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3248 * config/rs6000/altivec.h (vec_xst_be): New #define.
3249 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3250 and externalize from *altivec_vperm_<mode>_internal.
3251 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3253 (XL_BE_V8HI): Likewise.
3254 (XL_BE_V4SI): Likewise.
3255 (XL_BE_V4SI): Likewise.
3256 (XL_BE_V2DI): Likewise.
3257 (XL_BE_V4SF): Likewise.
3258 (XL_BE_V2DF): Likewise.
3259 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3260 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3261 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3262 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3263 VSX_BUILTIN_VEC_XST_BE.
3264 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3265 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3267 (altivec_init_builtins): Replace conditional calls to def_builtin
3268 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3269 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3270 with unconditional calls. Remove calls to def_builtin for
3271 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3272 __builtin_vec_xst_be.
3273 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3274 to define_expand, and add alternate RTL generation for P8.
3275 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3276 vsx_ld_elemrev_v8hi.
3277 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3278 add alternate RTL generation for P8.
3279 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3280 vsx_ld_elemrev_v16qi.
3281 (vsx_st_elemrev_v8hi): Convert define_insn
3282 to define_expand, and add alternate RTL generation for P8.
3283 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3284 vsx_st_elemrev_v8hi.
3285 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3286 add alternate RTL generation for P8.
3287 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3288 vsx_st_elemrev_v16qi.
3290 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3293 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3294 TARGET_AVX512ER check.
3295 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3296 neither -mzeroupper nor -mno-zeroupper is used and
3297 TARGET_EMIT_VZEROUPPER is set.
3298 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3299 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3301 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3303 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3304 folding of vector compares.
3305 (fold_build_vec_cmp): New helper function.
3306 (fold_compare_helper): New helper function.
3307 (builtin_function_type): Add compare builtins to the list of functions
3308 having unsigned arguments. Cosmetic updates to comment indentation.
3309 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3310 the not+eq combination.
3312 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3314 PR tree-optimization/82726
3315 PR tree-optimization/70754
3316 * tree-predcom.c (order_drefs_by_pos): New function.
3317 (combine_chains): Move code setting has_max_use_after to...
3318 (try_combine_chains): ...here. New parameter. Sort combined chains
3319 according to position information.
3320 (tree_predictive_commoning_loop): Update call to above function.
3321 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3323 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3325 PR tree-optimization/82726
3327 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3329 PR tree-optimization/70754
3330 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3331 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3332 combined stmt before it if not NULL.
3333 (combine_chains): Process refs reversely and compute dominance point
3337 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3339 PR tree-optimization/79663
3340 * tree-predcom.c (combine_chains): Process refs in reverse order
3341 only for ZERO length chains, and add explaining comment.
3343 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3345 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3346 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3347 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3349 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3351 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3353 2017-11-15 Martin Liska <mliska@suse.cz>
3355 * tree-cfg.c (pass_warn_function_return::execute):
3356 Compare warn_return_type for greater than zero.
3358 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3362 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3363 to return true on Xeon and not on Xeon Phi.
3364 (ix86_check_avx256_register): Changed to ...
3365 (ix86_check_avx_upper_register): ... this. Add extra check for
3366 VALID_AVX512F_REG_OR_XI_MODE.
3367 (ix86_avx_u128_mode_needed): Changed
3368 ix86_check_avx256_register to ix86_check_avx_upper_register.
3369 (ix86_check_avx256_stores): Changed to ...
3370 (ix86_check_avx_upper_stores): ... this. Changed
3371 ix86_check_avx256_register to ix86_check_avx_upper_register.
3372 (ix86_avx_u128_mode_after): Changed
3373 avx_reg256_found to avx_upper_reg_found. Changed
3374 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3375 (ix86_avx_u128_mode_entry): Changed
3376 ix86_check_avx256_register to ix86_check_avx_upper_register.
3377 (ix86_avx_u128_mode_exit): Ditto.
3378 * config/i386/i386.h: (host_detect_local_cpu): New define.
3380 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3382 * config/arm/xgene1.md (xgene1): Split into automatons
3383 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3384 (xgene1_f_load): Adjust reservations and/or types.
3385 (xgene1_f_store): Likewise.
3386 (xgene1_load_pair): Likewise.
3387 (xgene1_store_pair): Likewise.
3388 (xgene1_fp_load1): Likewise.
3389 (xgene1_load1): Likewise.
3390 (xgene1_store1): Likewise.
3391 (xgene1_move): Likewise.
3392 (xgene1_alu): Likewise.
3393 (xgene1_simd): Likewise.
3394 (xgene1_bfm): Likewise.
3395 (xgene1_neon_load1): Likewise.
3396 (xgene1_neon_store1): Likewise.
3397 (xgene1_neon_logic): Likewise.
3398 (xgene1_neon_st1): Likewise.
3399 (xgene1_neon_ld1r): Likewise.
3400 (xgene1_alu_cond): Added.
3401 (xgene1_shift_reg): Likwise.
3402 (xgene1_bfx): Likewise.
3403 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3405 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3408 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3410 (can_widen_mult_without_libcall): New function.
3411 (expand_mul_overflow): If only checking unsigned mul overflow,
3412 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3413 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3414 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3415 (expand_DIVMOD): Formatting fix.
3416 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3417 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3418 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3420 PR tree-optimization/82977
3421 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3422 constructed temporary to strlen_to_stridx.put.
3424 2017-11-15 Martin Liska <mliska@suse.cz>
3426 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3427 * configure: Regenerate.
3429 2017-11-15 Martin Liska <mliska@suse.cz>
3432 * config/sh/sh-mem.cc: Use proper probability for
3435 2017-11-14 Jeff Law <law@redhat.com>
3437 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3438 the red zone for stack_clash_protection_final_dynamic_probe targets
3439 when the total dynamic stack size is zero bytes.
3441 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3442 blocks is post order.
3444 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3446 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3447 * final.c (rest_of_clean_state): Set it for the
3448 -fcompare-debug dump.
3449 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3450 class when TDF_COMPARE_DEBUG is set.
3452 * dwarf2out.c (gen_producer_string): Discard
3455 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3458 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3459 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3460 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3461 (__TGMATH_REAL_2_3): Remove macros.
3462 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3463 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3465 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3467 (remquo): Define using __TGMATH_REAL_3.
3469 2017-11-14 Jeff Law <law@redhat.com>
3471 * vr-values.c: New file with contents extracted from tree-vrp.c.
3472 * Makefile.in (OBJS): Add vr-values.o
3473 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3474 (set_value_range, set_and_canonicalize_value_range): Likewise.
3475 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3476 (value_range_constant_singleton, symbolic_range_p): Likewise.
3477 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3478 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3479 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3480 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3481 (find_case_label_range, find_case_label_index): Likewise.
3482 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3483 (range_int_cst_singleton_p, value_inside_range): Likewise.
3484 (get_single_symbol): Likewise.
3485 (switch_update): Move structure definition here.
3486 (to_remove_edges, to_update_switch_stmts): Provide externs.
3487 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3488 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3489 (vrp_val_is_min, set_value_range): Likewise.
3490 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3491 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3492 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3493 (range_is_nonnull, range_int_cst_p): Likewwise.
3494 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3495 (get_single_symbol, operand_less_p): Likewise
3496 (compare_values_warnv, compare_values): Likewise.
3497 (value_inside_range, value_range_constant_singleton): Likewise.
3498 (zero_nonzero_bitgs_from_vr): Likewise.
3499 (extract_range_from_binary_expr_1): Likewise.
3500 (overflow_comparison_p): Likewise.
3501 (to_remove_edges, to_update_switch_stmts): Likewise.
3502 (find_case_label-index, find_case_label_range): Likewise.
3503 (switch_update, set_value_range_to_nonnegative): Remove.
3504 (set_value_range_to_truthvalue): Likewise.
3505 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3506 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3507 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3508 (find_case_label_ranges, test_for_singularity): Likewise.
3509 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3510 (x_vr_values): Move to its remaining use site.
3512 2017-11-10 Jeff Law <law@redhat.com>
3514 * vr-values.h (VR_INITIALIZER): Move #define here.
3515 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3516 * Makefile.in (OBJS): Add tree-evrp.o
3517 * tree-vrp.h (assert_info): Move structure definition here.
3518 (set_value_range_to_varying): Prototype.
3519 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3520 (infer_value_range, register_edge_assert_for): Likewise.
3521 (stmt_interesting_for_vrp): Likewise.
3522 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3523 (set_value_range_to_varying): No longer static.
3524 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3525 (infer_value_range, register_edge_assert_for): Likewise.
3527 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3529 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3530 generate the XXBRD instruction.
3532 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3533 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3534 as being compatible if -mabi=ieeelongdouble.
3535 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3536 to setup float128 built-ins with hardware support.
3537 (BU_FLOAT128_HW_2): Likewise.
3538 (BU_FLOAT128_HW_3): Likewise.
3539 (BU_FLOAT128_HW_VSX_1): Likewise.
3540 (BU_FLOAT128_HW_VSX_2): Likewise.
3541 (scalar_extract_expq): Change float128 built-in functions to
3542 accommodate having both KFmode and TFmode functions. Use the
3543 KFmode variant as the default.
3544 (scalar_extract_sigq): Likewise.
3545 (scalar_test_neg_qp): Likewise.
3546 (scalar_insert_exp_q): Likewise.
3547 (scalar_insert_exp_qp): Likewise.
3548 (scalar_test_data_class_qp): Likewise.
3549 (sqrtf128_round_to_odd): Delete processing the round to odd
3550 built-in functions as special built-in functions, and define them
3551 as float128 built-ins. Use the KFmode variant as the default.
3552 (truncf128_round_to_odd): Likewise.
3553 (addf128_round_to_odd): Likewise.
3554 (subf128_round_to_odd): Likewise.
3555 (mulf128_round_to_odd): Likewise.
3556 (divf128_round_to_odd): Likewise.
3557 (fmaf128_round_to_odd): Likewise.
3558 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3559 support for KFmode and TFmode xststdcqp calls.
3560 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3561 point, switch the built-in handlers for the get/set float128
3562 exponent, get float128 mantissa, float128 test built-ins, and the
3563 float128 round to odd built-in functions. Eliminate creating the
3564 float128 round to odd built-in functions as special built-ins.
3565 (rs6000_init_builtins): Eliminate special creation of the float128
3566 round to odd built-in functions.
3567 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3568 function insns to support both TFmode and KFmode variants.
3569 (xsxsigqp_<mode>): Likewise.
3570 (xsiexpqpf_<mode>): Likewise.
3571 (xsiexpqp_<mode>): Likewise.
3572 (xststdcqp_<mode>): Likewise.
3573 (xststdcnegqp_<mode>): Likewise.
3574 (xststdcqp_<mode>): Likewise.
3576 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3578 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3579 unused path_in_freq_ptr parameter.
3580 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3582 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3584 * ipa-inline.c (edge_badness): Dump sreal frequency.
3585 (compute_inlined_call_time): Match natural implementaiton ...
3586 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3587 forgotten division by CGRAPH_FREQ_BASE.
3589 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3591 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3592 Solaris 11. Update comment.
3593 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
3595 * config/sol2.h (STARTFILE_SPEC): Likewise.
3596 * configure: Regenerate.
3598 2017-11-14 Carl Love <cel@us.ibm.com>
3600 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
3601 le_ and be_ prefixes to swap* variables. Remove
3602 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
3605 2017-11-14 Jason Merrill <jason@redhat.com>
3607 Support GTY((cache)) on hash_map.
3608 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
3609 (ggc_cache_remove): Override it instead of ggc_mx.
3610 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
3611 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
3612 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
3613 (simple_cache_map_traits): Override maybe_mx.
3614 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
3615 (hash_map): Friend gt_cleare_cache.
3616 (gt_cleare_cache): New.
3617 * tree.h (tree_cache_traits): New hash_map traits class.
3618 (tree_cache_map): New typedef.
3620 2017-11-14 Richard Biener <rguenther@suse.de>
3622 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
3623 paramter and handling.
3624 (cleanup_control_flow_bb): Likewise.
3625 (cleanup_control_flow_pre): New helper performing a DFS walk
3626 to call cleanup_control_flow_bb in PRE order.
3627 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
3628 via cleanup_control_flow_pre.
3630 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
3632 * config/aarch64/aarch64-simd.md
3633 (aarch64_simd_bsl<mode>_internal): Remove DImode.
3634 (*aarch64_simd_bsl<mode>_alt): Likewise.
3635 (aarch64_simd_bsldi_internal): New.
3636 (aarch64_simd_bsldi_alt): Likewise.
3638 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3640 * tracer.c (better_p): Do not compare frequencies.
3641 * reg-stack.c (better_edge): Likewise.
3642 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
3645 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3647 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
3648 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
3649 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
3650 * lto-streamer-in.c (input_function): Use update_max_bb_count.
3651 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
3652 * predict.c (maybe_hot_frequency_p): Inline to ...
3653 (maybe_hot_count_p): ... here; rewrite to counts.
3654 (counts_to_freqs): Rename to ...
3655 (update_max_bb_count): ... this one.
3656 (expensive_function_p): Use counts.
3657 (estimate_bb_frequencies): Update.
3658 (rebuild_frequencies): Update.
3659 * predict.h (counts_to_freqs): Rename to ...
3660 (update_max_bb_count): ... this one.
3661 * profile.c (compute_branch_probabilities): Add debug info
3662 * tree-inline.c (expand_call_inline): Update debug info.
3663 (optimize_inline_calls): Use update_max_bb_count..
3664 (tree_function_versioning): Use update_max_bb_count..
3665 * value-prof.c (gimple_value_profile_transformations):
3666 Do not use update_max_bb_count.
3668 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3670 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
3671 always use frequencies.
3673 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3675 * bb-reorder.c: Remove frequencies from comments.
3676 (better_edge_p): Use profile counts.
3677 (find_traces): Dump profile counts.
3678 (rotate_loop): Use profile counts.
3679 (find_traces_1_round): Likewise.
3680 (connect_better_edge_p): Use counts instead of probabilities for
3682 (copy_bb_p): Drop early check for non-0 frequency.
3683 (sanitize_hot_paths): Update comments.
3685 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3687 * ipa-split.c (struct split_point): Add count.
3688 (consider_split): Do not compute incoming frequency; compute incoming
3689 count and store it to split_point.
3690 (split_function): Set count of the call to split part correctly.
3692 2017-11-13 Carl Love <cel@us.ibm.com>
3694 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
3696 2017-11-13 Tom Tromey <tom@tromey.com>
3698 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
3700 2017-11-13 Carl Love <cel@us.ibm.com>
3702 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3703 Add support for builtins:
3704 unsigned int vec_first_{,miss}_match_{,or_eos}index,
3705 vector {un,}signed {char,int,short},
3706 vector {un,}signed {char,int,short}) arguments.
3707 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
3708 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
3709 Add BU_P9V_AV_2 expansions for the builtins.
3710 * config/rs6000/altivec.h (vec_first_match_index,
3711 vec_first_mismatch_index, vec_first_match_or_eos_index,
3712 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
3713 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
3714 new extern declaration.
3715 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
3716 * config/rs6000/vsx.md (first_match_index_<mode>,
3717 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
3718 first_mismatch_or_eos_index_<mode>): Add define expand.
3719 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
3720 * doc/extend.texi: Update the built-in documenation file for the new
3723 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
3725 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
3726 into the min/max operations for _Float<N> and _Float<N>X types.
3728 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3731 * dwarf2out.c (do_eh_frame): New static variable.
3732 (dwarf2out_begin_prologue): Set it.
3733 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
3735 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3737 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
3739 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
3741 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3744 * debug.h (dwarf2out_do_eh_frame): Declare.
3745 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
3746 (dwarf2out_do_frame): Use it.
3747 (dwarf2out_do_cfi_asm): Likewise.
3748 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
3749 (dwarf2out_assembly_start): Likewise.
3750 (dwarf2out_begin_prologue): Fix comment.
3751 * toplev.c (compile_file): Always call dwarf2out_frame_finish
3752 if the target needs either debug or unwind DWARF2 info.
3753 * lto-opts.c (lto_write_options): Do not save -fexceptions,
3754 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
3755 -ftrapping-math, -ftrapv and -fwrapv.
3757 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3759 * cgraph.c (cgraph_edge::sreal_frequency): New function.
3760 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
3761 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
3762 (estimate_edge_size_and_time): Likewise.
3763 (ipa_merge_fn_summary_after_inlining): Likewise.
3764 * ipa-inline.c (cgraph_freq_base_rec): Remove.
3765 (compute_uninlined_call_time): Use sreal_frequency.
3766 (compute_inlined_call_time): Likewise.
3767 (ipa_inline): Do not initialize cgraph_freq_base_rec.
3768 * profile-count.c: Include sreal.h.
3769 (profile_count::to_sreal_scale): New.
3770 * profile-count.h: Forward declare sreal.
3771 (profile_count::to_sreal_scale): Declare.
3773 2017-11-13 Nathan Sidwell <nathan@acm.org>
3775 * diagnostic.c (maybe_line_and_column): New.
3776 (diagnostic_get_location_text): Use it.
3777 (diagnostic_report_current_module): Likewise.
3778 (test_diagnostic_get_location_text): Add tests.
3780 2017-11-13 Luis Machado <luis.machado@linaro.org>
3782 * doc/md.texi (Specifying processor pipeline description): Fix
3783 incorrect latency for the div instruction example.
3785 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3787 PR tree-optimization/78821
3788 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
3789 that bit_not_p is the same.
3790 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
3791 (split_group): Count precisely bit_not_p bits in each statement.
3792 (invert_op): New function.
3793 (imm_store_chain_info::output_merged_store): Use invert_op to
3794 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
3795 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
3797 2017-11-13 Martin Liska <mliska@suse.cz>
3799 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
3800 (struct source_info): Likewise.
3801 (add_branch_counts): Likewise.
3802 (add_line_counts): Likewise.
3803 (function_summary): Likewise.
3804 (output_intermediate_line): Likewise.
3805 (generate_results): Likewise.
3807 2017-11-13 Martin Liska <mliska@suse.cz>
3809 * gcov.c (struct block_info): Remove typedef for block_t.
3810 (struct line_info): Likewise.
3811 (line_info::has_block): Likewise.
3812 (EXIT_BLOCK): Likewise.
3813 (unblock): Likewise.
3814 (circuit): Likewise.
3815 (get_cycles_count): Likewise.
3816 (process_file): Likewise.
3817 (read_graph_file): Likewise.
3818 (solve_flow_graph): Likewise.
3819 (find_exception_blocks): Likewise.
3820 (add_line_counts): Likewise.
3821 (accumulate_line_info): Likewise.
3822 (output_line_details): Likewise.
3824 2017-11-13 Martin Liska <mliska@suse.cz>
3826 * gcov.c (struct arc_info): Remove typedef for arc_t.
3827 (struct line_info): Likewise.
3828 (add_branch_counts): Likewise.
3829 (output_branch_count): Likewise.
3830 (function_info::~function_info): Likewise.
3831 (circuit): Likewise.
3832 (output_intermediate_line): Likewise.
3833 (read_graph_file): Likewise.
3834 (solve_flow_graph): Likewise.
3835 (find_exception_blocks): Likewise.
3836 (add_line_counts): Likewise.
3837 (accumulate_line_info): Likewise.
3838 (output_line_details): Likewise.
3839 (output_function_details): Likewise.
3841 2017-11-13 Martin Liska <mliska@suse.cz>
3843 * gcov.c (struct function_info): Remove typedef for function_t.
3844 (struct source_info): Likewise.
3845 (source_info::get_functions_at_location): Likewise.
3846 (solve_flow_graph): Likewise.
3847 (find_exception_blocks): Likewise.
3848 (add_line_counts): Likewise.
3849 (output_intermediate_file): Likewise.
3850 (process_file): Likewise.
3851 (generate_results): Likewise.
3852 (release_structures): Likewise.
3853 (read_graph_file): Likewise.
3854 (read_count_file): Likewise.
3855 (accumulate_line_counts): Likewise.
3856 (output_lines): Likewise.
3858 2017-11-13 Martin Liska <mliska@suse.cz>
3860 * gcov.c (function_info::function_info): Remove num_counts
3861 and add vector<gcov_type>.
3862 (function_info::~function_info): Use the vector.
3863 (process_file): Likewise.
3864 (read_graph_file): Likewise.
3865 (read_count_file): Likewise.
3866 (solve_flow_graph): Likewise.
3868 2017-11-13 Martin Liska <mliska@suse.cz>
3870 * gcov.c (function_info::is_artificial): New function.
3871 (process_file): Erase all artificial early.
3872 (generate_results): Skip as all artificial are already
3875 2017-11-13 Martin Liska <mliska@suse.cz>
3877 * gcov.c (read_graph_file): Store to global vector of functions.
3878 (read_count_file): Iterate the vector.
3879 (process_file): Likewise.
3880 (generate_results): Likewise.
3881 (release_structures): Likewise.
3883 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3885 PR tree-optimization/82954
3886 * gimple-ssa-store-merging.c
3887 (imm_store_chain_info::coalesce_immediate_stores): If
3888 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
3890 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
3892 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
3893 Upddate call to ENDIAN_LANE_N.
3894 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
3895 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
3896 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
3897 and use aarch64_endian_lane_rtx.
3898 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
3900 2017-11-12 Tom de Vries <tom@codesourcery.com>
3902 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
3904 2017-11-12 Tom de Vries <tom@codesourcery.com>
3906 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
3908 2017-11-12 Tom de Vries <tom@codesourcery.com>
3910 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
3912 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3913 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3914 * defaults.h (ASM_OUTPUT_LABELREF): Same.
3916 2017-11-11 Martin Sebor <msebor@redhat.com>
3919 * doc/extend.texi (attribute nonstring): Remove spurious argument.
3922 * prefic.c (translate_name): Replace strncpy with memcpy to
3923 avoid -Wstringop-truncation.
3925 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3927 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
3929 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3931 * predict.c (maybe_hot_frequency_p): Do not use cfun.
3933 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3935 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
3938 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
3940 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
3941 enable generating XXBRH if the value is in a vector register.
3942 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
3943 value is in a vector register.
3944 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
3945 register to register bswap64's instead of doing the GPR sequence
3946 used on previous machines.
3947 (bswapdi2_xxbrd): New insn.
3948 (bswapdi2_reg): Disallow on ISA 3.0.
3949 (register to register bswap64 splitter): Do not split the insn on
3950 ISA 3.0 systems that use XXBRD.
3952 2017-11-10 Martin Sebor <msebor@redhat.com>
3955 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
3957 (find_subframework_file): Same.
3959 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3961 * auto-profile.c (afdo_indirect_call): Drop frequency.
3962 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
3963 (cgraph_node::create_edge): Drop frequency argument.
3964 (cgraph_node::create_indirect_edge): Drop frequency argument.
3965 (cgraph_edge::make_speculative): Drop frequency arguments.
3966 (cgraph_edge::resolve_speculation): Do not update frequencies
3967 (cgraph_edge::dump_edge_flags): Do not dump frequency.
3968 (cgraph_node::dump): Check consistency in IPA mode.
3969 (cgraph_edge::maybe_hot_p): Use IPA counter.
3970 (cgraph_edge::verify_count_and_frequency): Rename to ...
3971 (cgraph_edge::verify_count): ... this one; drop frequency checking.
3972 (cgraph_node::verify_node): Update.
3973 * cgraph.h (struct cgraph_edge): Drop frequency.
3974 (cgraph_edge::frequency): New function.
3975 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
3977 (cgraph_edge::rebuild_edges): Likewise.
3978 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
3979 (duplicate_thunk_for_node): Do not pass frequency.
3980 (cgraph_node::create_clone): Scale only counts.
3981 (cgraph_node::create_virtual_clone): Do not pass frequency.
3982 (cgraph_node::create_edge_including_clones): Do not pass frequency.
3983 (cgraph_node::create_version_clone): Do not pass frequency.
3984 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
3985 (cgraph_node::expand_thunk): Do not pass frequency.
3986 (cgraph_node::create_wrapper): Do not pass frequency.
3987 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
3989 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
3990 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
3991 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
3992 (gather_caller_stats): Use frequency function.
3993 (ipcp_cloning_candidate_p): Use frequency function.
3994 (ipcp_propagate_stage): Use frequency function.
3995 (get_info_about_necessary_edges): Use frequency function.
3996 (update_profiling_info): Update only IPA profile.
3997 (update_specialized_profile): Use frequency functoin.
3998 (perhaps_add_new_callers): Update only IPA profile.
3999 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4000 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4001 (dump_ipa_call_summary): Use frequency function.
4002 (estimate_edge_size_and_time): Use frequency function.
4003 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4004 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4005 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4006 (update_noncloned_counts): ... ths one; scale counts only.
4007 (clone_inlined_nodes): Do not scale frequency.
4008 (inline_call): Do not pass frequency.
4009 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4010 (compute_inlined_call_time): Use IPA profile.
4011 (want_inline_small_function_p): Use IPA profile.
4012 (want_inline_self_recursive_call_p): Use IPA profile.
4013 (edge_badness): Use IPA profile.
4014 (lookup_recursive_calls): Use IPA profile.
4015 (recursive_inlining): Do not pass frequency.
4016 (resolve_noninline_speculation): Do not update frequency.
4017 (inline_small_functions): Collect max of IPA profile.
4018 (dump_overall_stats): Dump IPA porfile.
4019 (dump_inline_stats): Dump IPA porfile.
4020 (ipa_inline): Collect IPA stats.
4021 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4022 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4023 (ipa_propagate_frequency): Use frequency function.
4024 (ipa_profile): Cleanup.
4025 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4026 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4027 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4028 (input_node): Do not stream frequency.
4029 (input_edge): Do not stream frequency.
4030 (merge_profile_summaries): Scale only IPA profiles.
4031 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4032 * predict.c (drop_profile): Do not recompute frequency.
4033 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4034 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4035 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4036 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4037 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4038 * tree-inline.c (copy_bb): Do not scale frequency.
4039 (expand_call_inline): Do not scale frequency.
4040 (tree_function_versioning): Do not scale frequency.
4041 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4043 2017-11-10 Julia Koval <julia.koval@intel.com>
4045 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4046 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4047 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4048 (_mm256_maskz_gf2p8affine_epi64_epi8)
4049 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4050 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4051 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4052 (__builtin_ia32_vgf2p8affineqb_v32qi)
4053 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4054 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4056 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4060 (arm_option_override): Refactor.
4061 (arm_option_reconfigure_globals): New.
4062 (arm_options_perform_arch_sanity_checks): New.
4063 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4065 (arm_options_perform_arch_sanity_checks): Likewise
4067 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4069 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4070 (power9-qpmul): New.
4071 * rs6000/rs6000.md ("type" attr): Add qmul.
4072 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4073 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4074 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4076 2017-11-10 Martin Sebor <msebor@redhat.com>
4079 * builtins.c (compute_objsize): Handle arrays that
4080 compute_builtin_object_size likes to fail for. Make extern.
4081 * builtins.h (compute_objsize): Declare.
4082 (check_strncpy_sizes): New function.
4083 (expand_builtin_strncpy): Call check_strncpy_sizes.
4084 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4085 -Wstringop-truncation.
4086 (gimple_fold_builtin_strncat): Same.
4087 * gimple.c (gimple_build_call_from_tree): Set call location.
4088 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4089 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4090 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4091 (handle_builtin_strlen): Use strlen_to_stridx.
4092 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4094 Use strlen_to_stridx.
4095 (pass_strlen::execute): Release strlen_to_stridx.
4096 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4097 (-Wstringop-truncation): Document new option.
4099 2017-11-10 Martin Liska <mliska@suse.cz>
4101 PR gcov-profile/82702
4102 * gcov.c (main): Handle intermediate files in a different
4104 (get_gcov_intermediate_filename): New function.
4105 (output_gcov_file): Remove support of intermediate files.
4106 (generate_results): Allocate intermediate file.
4107 (release_structures): Clean-up properly fn_end.
4108 (output_intermediate_file): Start iterating with line 1.
4110 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4112 PR tree-optimization/82929
4113 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4114 ops_swapped_p non-static data member.
4115 (store_immediate_info::store_immediate_info): Clear it.
4116 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4117 ops set ops_swapped_p.
4118 (count_multiple_uses): Handle ops_swapped_p.
4120 2017-11-10 Martin Liska <mliska@suse.cz>
4122 * coverage.c (coverage_init): Stream information about
4123 support of has_unexecuted_blocks.
4124 * doc/gcov.texi: Document that.
4125 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4126 * gcov.c (read_graph_file): Likewise.
4127 (output_line_beginning): Fix a small issue with
4130 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4132 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4133 reference of trivial component.
4135 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4138 * gimple-ssa-store-merging.c
4139 (pass_store_merging::terminate_all_aliasing_chains): For
4140 gimple_store_p stmts also call refs_output_dependent_p.
4142 PR rtl-optimization/82913
4143 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4146 2017-11-09 Jeff Law <law@redhat.com>
4148 * vr-values.h: New file with vr_values class.
4149 * tree-vrp.c: Include vr-values.h
4150 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4151 data objects into the vr_values class.
4152 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4153 (get_value_range): Make it a member function within vr_values class.
4154 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4155 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4156 (op_with_constant_singleton_value_range): Likewise.
4157 (extract_range_for_var_from_comparison_expr): Likewise.
4158 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4159 (extract_range_from_binary_expr): Likewise.
4160 (extract_range_from_unary_expr): Likewise.
4161 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4162 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4163 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4164 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4165 (compare_name_with_value, compare_names): Likewise.
4166 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4167 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4168 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4169 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4170 (extract_range_from_phi_node): Likewise.
4171 (simplify_truth_ops_using_ranges): Likewise.
4172 (simplify_div_or_mod_using_ranges): Likewise.
4173 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4174 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4175 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4176 (simplify_float_conversion_using_ranges): Likewise.
4177 (simplify_internal_call_using_ranges): Likewise.
4178 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4179 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4180 poitner via x_vr_values for calls into gimple folder.
4181 (vrp_initialize_lattice): Make this the vr_values ctor.
4182 (vrp_free_lattice): Make this the vr_values dtor.
4183 (set_vr_value): New function.
4184 (class vrp_prop): Add vr_values data member. Add various member
4185 functions as well as member functions that delegate to vr_values.
4186 (check_array_ref): Make a member function within vrp_prop class.
4187 (search_for_addr_array, vrp_initialize): Likewise.
4188 (vrp_finalize): Likewise. Revamp to avoid direct access to
4189 vr_value, values_propagated, etc.
4190 (check_array_bounds): Extract vrp_prop class instance pointer from
4191 walk info structure. Use it to call member functions.
4192 (check_all_array_refs): Make a member function within vrp_prop class.
4193 Smuggle class instance pointer via walk info structure.
4194 (x_vr_values): New local static.
4195 (vrp_valueize): Use x_vr_values to get class instance.
4196 (vr_valueize_1): Likewise.
4197 (class vrp_folder): Add vr_values data member. Add various member
4198 functions as well as member functions that delegate to vr_values.
4199 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4200 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4201 class instance from vr_values. Use it to call member functions.
4202 (vrp_dom_walker): Add vr_values data member.
4203 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4204 instance via x_vr_values.
4205 (identify_jump_threads): Accept vr_values as argument. Store
4206 it into the walker structure.
4207 (evrp_dom_walker): Add vr_values class data member. Add various
4209 (evrp_dom_walker::try_find_new_range): Use vr_values data
4210 member to access the memory allocator.
4211 (evrp_dom_walker::before_dom_children): Store vr_values class
4212 instance into the vrp_folder class.
4213 (evrp_dom_walker::push_value_range): Rework to avoid direct
4214 access to num_vr_values and vr_value.
4215 (evrp_dom_walker::pop_value_range): Likewise.
4216 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4217 Use vr_values to get to dump_all_value_ranges member function.
4218 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4219 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4220 Pass vr_values class instance down to identify_jump_threads.
4221 Remove call to vrp_free_lattice.
4222 (debug_all_value_ranges): Remove.
4224 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4225 (vrp_folder): Likewise.
4227 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4228 Get it from the existing bitmap instead.
4229 (vrp_intersect_ranges_1): Likewise.
4231 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4233 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4235 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4236 set bit_not_p to it.
4237 (imm_store_chain_info::coalesce_immediate_stores): Break group
4238 if bit_not_p is different.
4239 (count_multiple_uses, split_group,
4240 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4241 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4242 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4243 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4244 to store_immediate_info ctor.
4246 2017-11-09 Jim Wilson <jimw@sifive.com>
4248 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4249 (scan_prog_file): Likewise.
4251 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4253 * bb-reorder.c (max_entry_frequency): Remove.
4254 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4255 connect_traces, push_to_next_round_p): Remove prototypes.
4256 (find_traces_1_round): Use counts only.
4257 (push_to_next_round_p): Likewise.
4258 (find_traces): Likewise.
4259 (rotate_loop): Likewise.
4260 (find_traces_1_round): Likewise.
4261 (connect_traces): Likewise.
4262 (edge_order): Likewise.
4264 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4266 * config/arm/arm.c (output_return_instruction): Add comments to
4267 indicate requirement for cmse_nonsecure_entry return to account
4268 for the size of clearing instruction output here.
4269 (thumb_exit): Likewise.
4270 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4271 return in hardfloat mode.
4273 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4275 * config/rs6000/rs6000.c (machine_function): Add a bool,
4276 "toc_is_wrapped_separately".
4277 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4278 if it wasn't explicitly set or unset, we are optimizing for speed, and
4279 doing separate shrink-wrapping.
4280 (rs6000_get_separate_components): Enable the TOC component if
4281 saving the TOC register in the prologue.
4282 (rs6000_components_for_bb): Handle the TOC component.
4283 (rs6000_emit_prologue_components): Store the TOC register where needed.
4284 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4285 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4287 2017-11-09 Martin Jambor <mjambor@suse.cz>
4289 * ipa-param-manipulation.c: New file.
4290 * ipa-param-manipulation.h: Likewise.
4291 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4292 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4293 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4294 (ipa_parm_adjustment): Likewise.
4295 (ipa_parm_adjustment_vec): Likewise.
4296 (ipa_get_vector_of_formal_parms): Moved declaration to
4297 ipa-param-manipulation.h.
4298 (ipa_get_vector_of_formal_parm_types): Likewise.
4299 (ipa_modify_formal_parameters): Likewise.
4300 (ipa_modify_call_arguments): Likewise.
4301 (ipa_combine_adjustments): Likewise.
4302 (ipa_dump_param_adjustments): Likewise.
4303 (ipa_modify_expr): Likewise.
4304 (ipa_get_adjustment_candidate): Likewise.
4305 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4306 ipa-param-manipulation.c.
4307 (ipa_get_vector_of_formal_parm_types): Likewise.
4308 (ipa_modify_formal_parameters): Likewise.
4309 (ipa_modify_call_arguments): Likewise.
4310 (ipa_modify_expr): Likewise.
4311 (get_ssa_base_param): Likewise.
4312 (ipa_get_adjustment_candidate): Likewise.
4313 (index_in_adjustments_multiple_times_p): Likewise.
4314 (ipa_combine_adjustments): Likewise.
4315 (ipa_dump_param_adjustments): Likewise.
4316 * tree-sra.c: Also include ipa-param-manipulation.h
4317 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4320 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4321 Alan Hayward <alan.hayward@arm.com>
4322 David Sherwood <david.sherwood@arm.com>
4324 * doc/sourcebuild.texi (vect_masked_store): Document.
4326 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4327 Alan Hayward <alan.hayward@arm.com>
4328 David Sherwood <david.sherwood@arm.com>
4330 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4332 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4333 Alan Hayward <alan.hayward@arm.com>
4334 David Sherwood <david.sherwood@arm.com>
4336 * doc/sourcebuild.texi (vect_variable_length): Document.
4338 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4339 Alan Hayward <alan.hayward@arm.com>
4340 David Sherwood <david.sherwood@arm.com>
4342 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4344 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4345 Alan Hayward <alan.hayward@arm.com>
4346 David Sherwood <david.sherwood@arm.com>
4348 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4350 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4351 Alan Hayward <alan.hayward@arm.com>
4352 David Sherwood <david.sherwood@arm.com>
4354 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4355 previously undocumented selectors.
4356 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4358 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4360 * doc/rtl.texi (const_vector): Say that elements can be
4361 const_wide_ints too.
4362 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4363 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4364 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4365 * optabs.c (expand_vector_broadcast): Likewise.
4367 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4368 Alan Hayward <alan.hayward@arm.com>
4369 David Sherwood <david.sherwood@arm.com>
4371 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4372 scaled index even if the unscaled address was invalid.
4373 Don't increase the complexity of using a scale in that case.
4375 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4376 Alan Hayward <alan.hayward@arm.com>
4377 David Sherwood <david.sherwood@arm.com>
4379 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4380 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4381 UNITS_PER_WORD bytes.
4382 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4383 based on REGMODE_NATURAL_SIZE of the inner mode.
4384 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4385 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4386 * expmed.c (lowpart_bit_field_p): Divide the value up into
4387 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4388 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4389 whether something is likely to occupy more than one register.
4391 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4394 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4395 profile_count::adjust_for_ipa_scaling.
4396 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4397 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4399 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4401 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4403 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4404 (split_group): Add total_orig and total_new arguments, estimate the
4405 number of statements related to the store group without store merging
4406 and with store merging.
4407 (imm_store_chain_info::output_merged_store): Adjust split_group
4408 callers, punt if estimated number of statements with store merging
4409 is not smaller than estimated number of statements without it.
4411 (handled_load): Remove has_single_use checks.
4412 (pass_store_merging::process_store): Likewise.
4414 2017-11-09 Richard Biener <rguenther@suse.de>
4416 PR tree-optimization/82902
4417 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4419 2017-11-09 Martin Liska <mliska@suse.cz>
4422 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4425 2017-11-09 Martin Liska <mliska@suse.cz>
4427 PR tree-optimization/82669
4428 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4430 2017-11-09 Martin Liska <mliska@suse.cz>
4432 PR gcov-profile/48463
4433 * coverage.c (coverage_begin_function): Output also end locus
4434 of a function and information whether the function is
4436 * gcov-dump.c (tag_function): Parse and print the information.
4437 * gcov.c (INCLUDE_MAP): Add include.
4438 (INCLUDE_SET): Likewise.
4439 (struct line_info): Move earlier in the source file because
4440 of vector<line_info> in function_info structure.
4441 (line_info::line_info): Likewise.
4442 (line_info::has_block): Likewise.
4443 (struct source_info): Add new member index.
4444 (source_info::get_functions_at_location): New function.
4445 (function_info::group_line_p): New function.
4446 (output_intermediate_line): New function.
4447 (output_intermediate_file): Use the mentioned function.
4448 (struct function_start): New.
4449 (struct function_start_pair_hash): Likewise.
4450 (process_file): Add code that identifies group functions.
4451 Assign lines either to global or function scope.
4452 (generate_results): Skip artificial functions.
4453 (find_source): Assign index for each source file.
4454 (read_graph_file): Read new flag artificial and end_line.
4455 (add_line_counts): Assign it either to global of function scope.
4456 (accumulate_line_counts): Isolate core of the function to
4457 accumulate_line_info and call it for both function and global
4459 (accumulate_line_info): New function.
4460 (output_line_beginning): Fix GNU coding style.
4461 (print_source_line): New function.
4462 (output_line_details): Likewise.
4463 (output_function_details): Likewise.
4464 (output_lines): Iterate both source (global) scope and function
4466 (struct function_line_start_cmp): New class.
4467 * doc/gcov.texi: Reflect changes in documentation.
4469 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4472 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4473 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4474 and similarly for not instead of neg.
4476 2017-11-08 Andi Kleen <ak@linux.intel.com>
4478 * config/i386/i386.opt: Add -mforce-indirect-call.
4479 * config/i386/predicates.md: Check for flag_force_indirect_call.
4480 * doc/invoke.texi: Document -mforce-indirect-call
4482 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4484 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4486 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4487 config/riscv/riscv.c (predict.h): New include.
4488 (riscv_slow_unaligned_access_p): No longer static.
4489 (riscv_block_move_straight): Add require.
4490 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4492 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4495 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4496 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4497 nonimmediate_operand predicate for operand 1 instead of
4500 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4502 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4504 * config/aarch64/constraints.md (Uml): New constraint.
4505 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4508 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4510 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4511 of two vec_duplicates into a vec_concat.
4513 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4515 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4516 Simplify vec_merge of vec_duplicate and vec_concat.
4517 * config/aarch64/constraints.md (Utq): New constraint.
4518 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4521 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4523 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4524 Simplify vec_merge of vec_duplicate and const_vector.
4525 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4527 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4528 mode iterator. Update predicate on operand 1 to
4529 handle non-const_vec constants. Delete constraints.
4530 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4532 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4534 PR tree-optimization/78821
4535 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4537 (store_operand_info::store_operand_info): Initialize it to false.
4538 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4539 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4540 store in the group, and if chain_info is non-NULL, to ignore altogether
4542 (compatible_load_p): Fail if bit_not_p does not match.
4543 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4544 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4545 (pass_store_merging::process_store): Adjust
4546 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4547 call terminate_all_aliasing_chains newly when adding a store into
4548 a chain with non-NULL chain_info.
4550 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4552 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4554 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4556 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4558 (aarch64_layout_frame): Initialise emit_frame_chain.
4559 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4560 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4562 2017-11-08 Martin Liska <mliska@suse.cz>
4564 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4565 of gimple_call_internal_p.
4567 2017-11-07 Tom de Vries <tom@codesourcery.com>
4569 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4571 2017-11-07 Tom de Vries <tom@codesourcery.com>
4573 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4576 2017-11-08 Martin Liska <mliska@suse.cz>
4579 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4581 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4583 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4585 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4588 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4589 (*cmp<mode>_ccz_1): New insn with $k alternative.
4592 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4593 RTX_COMM_COMPARE as commutative as well.
4594 (ix86_binary_operator_ok): Formatting fix.
4595 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
4596 *<code><mode>3<mask_name><round_saeonly_name>,
4597 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
4598 *<s>mul<mode>3_highpart<mask_name>,
4599 *vec_widen_umult_even_v16si<mask_name>,
4600 *vec_widen_umult_even_v8si<mask_name>,
4601 *vec_widen_umult_even_v4si<mask_name>,
4602 *vec_widen_smult_even_v16si<mask_name>,
4603 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
4604 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
4605 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
4606 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
4607 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
4608 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
4609 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
4610 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
4611 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
4612 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
4613 ix86_binary_operator_ok. Formatting fixes.
4614 (*<plusminus_insn><mode>3<mask_name><round_name>,
4615 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
4618 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4620 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
4621 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
4622 eq and ne if TARGET_ISEL.
4623 (cmp): New code_iterator.
4624 (UNS, UNSU_, UNSIK): New code_attrs.
4625 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
4626 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
4628 ("eq<mode>3"): ... this.
4629 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
4631 ("ne<mode>3"): ... this.
4633 2017-11-07 Julia Koval <julia.koval@intel.com>
4636 * common/config/i386/i386-common.c
4637 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
4638 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
4639 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
4640 * config/i386/i386.opt: Ditto.
4641 * config/i386/i386.c (ix86_target_string): Ditto.
4642 (ix86_option_override_internal): Ditto.
4643 (ix86_init_mpx_builtins): Move MPX to args2.
4644 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
4645 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4646 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4647 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4648 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4649 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4650 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
4652 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
4655 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
4656 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
4657 (zero-extendsidi peephole2): Remove peephole.
4659 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4662 * stor-layout.c: Include attribs.h.
4663 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
4664 explicit lookup of "aligned" attribute.
4666 2017-11-07 Andrew Waterman <andrew@sifive.com>
4668 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
4669 (riscv_expand_block_move): Likewise.
4670 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
4672 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
4673 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
4674 * config/riscv/riscv.c (riscv_block_move_straight): New function.
4675 (riscv_adjust_block_mem): Likewise.
4676 (riscv_block_move_loop): Likewise.
4677 (riscv_expand_block_move): Likewise.
4678 * config/riscv/riscv.md (movmemsi): New pattern.
4680 2017-11-07 Michael Clark <michaeljclark@mac.com>
4682 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
4683 (MUSL_DYNAMIC_LINKER): Likewise.
4685 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
4687 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
4688 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
4691 2017-11-07 Richard Biener <rguenther@suse.de>
4693 * match.pd: Fix build.
4695 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
4696 Jackson Woodruff <jackson.woodruff@arm.com>
4698 PR tree-optimization/71026
4699 * match.pd: Canonicalize negate in division.
4701 2017-11-07 Sudakshina Das <sudi.das@arm.com>
4704 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
4706 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4708 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
4709 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
4711 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4713 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
4714 non-scalar integral types.
4715 * match.pd (negate_expr_p): Handle MINUS_EXPR.
4716 (-(A-B), -(~A)): New transformations.
4718 2017-11-07 Tom de Vries <tom@codesourcery.com>
4720 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
4721 semicolon after "do {} while (0)".
4722 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4723 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4724 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4725 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4726 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4728 2017-11-07 Tom de Vries <tom@codesourcery.com>
4730 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
4731 after "do {} while (0)".
4732 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4733 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4734 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4735 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4736 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4738 2017-11-07 Tom de Vries <tom@codesourcery.com>
4741 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
4743 (arm_rtx_costs_internal): Add missing semicolon after
4744 HANDLE_NARROW_SHIFT_ARITH call.
4746 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4748 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4749 disable isel if it was not set explicitly.
4751 2017-11-06 James Bowman <james.bowman@ftdichip.com>
4753 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
4754 (add_type_attribute) likewise.
4756 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
4758 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
4759 of struct ix86_frame.
4760 (ix86_initial_elimination_offset): Likewise.
4761 (ix86_expand_split_stack_prologue): Likewise.
4763 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
4765 * tree-vrp.h (enum value_range_type): Update stale comment.
4767 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4768 Alan Hayward <alan.hayward@arm.com>
4769 David Sherwood <david.sherwood@arm.com>
4771 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
4772 (aarch64_expand_vec_perm_const): Take the number of units too.
4773 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
4774 (aarch64_expand_vec_perm_const): Likewise.
4775 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
4776 (vec_perm<mode>): Update accordingly.
4778 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4779 Alan Hayward <alan.hayward@arm.com>
4780 David Sherwood <david.sherwood@arm.com>
4782 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
4783 Take the number of units too.
4784 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
4785 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
4786 but check for a vector mode before rather than after the call.
4787 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
4788 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
4789 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
4790 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
4791 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
4792 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
4793 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
4794 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
4795 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
4796 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
4797 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
4798 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
4799 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
4800 (aarch64_sqdmull2_n<mode>): Update accordingly.
4802 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4803 Alan Hayward <alan.hayward@arm.com>
4804 David Sherwood <david.sherwood@arm.com>
4806 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
4807 the number of units too.
4808 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
4809 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
4810 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
4811 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
4812 (vec_store_lanesxi<mode>): Update accordingly.
4814 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4815 Alan Hayward <alan.hayward@arm.com>
4816 David Sherwood <david.sherwood@arm.com>
4818 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
4819 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
4820 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
4821 of units rather than the mode.
4822 * config/aarch64/iterators.md (nunits): New mode attribute.
4823 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4824 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
4825 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
4826 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
4827 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
4828 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
4829 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
4830 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
4831 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
4832 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
4833 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4834 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
4835 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
4836 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
4837 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
4838 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
4839 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
4840 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
4841 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
4842 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
4843 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
4844 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
4845 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
4846 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4847 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
4848 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
4849 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4850 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
4851 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
4852 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
4853 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
4854 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
4855 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
4856 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
4857 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
4858 (aarch64_simd_vec_setv2di): Likewise.
4860 2017-11-06 Carl Love <cel@us.ibm.com>
4862 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
4864 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
4865 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
4867 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
4869 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
4870 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
4871 Add power 8 macro expansions.
4872 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
4873 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
4874 power 8 instructions. (VSX_XXBR): Add iterator.
4876 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
4878 * config/arm/arm.md (predicable_short_it): Change default to "no",
4879 improve documentation, remove uses that are identical to the default.
4880 (enabled_for_depr_it): Rename to enabled_for_short_it.
4881 * gcc/config/arm/arm-fixed.md (predicable_short_it):
4882 Remove default uses.
4883 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
4884 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
4885 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
4886 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
4888 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4891 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
4892 float128 helper macros, which are no longer used after deleting
4893 the old 'q' built-in functions, and moving the round to odd
4894 built-in functions to being special built-in functions.
4895 (BU_FLOAT128_2): Likewise.
4896 (BU_FLOAT128_1_HW): Likewise.
4897 (BU_FLOAT128_2_HW): Likewise.
4898 (BU_FLOAT128_3_HW): Likewise.
4899 (FABSQ): Delete old 'q' built-in functions.
4900 (COPYSIGNQ): Likewise.
4901 (SQRTF128_ODD): Move round to odd built-in functions to be
4902 special built-in functions, so that we can handle
4903 -mabi=ieeelongdouble.
4904 (TRUNCF128_ODD): Likewise.
4905 (ADDF128_ODD): Likewise.
4906 (SUBF128_ODD): Likewise.
4907 (MULF128_ODD): Likewise.
4908 (DIVF128_ODD): Likewise.
4909 (FMAF128_ODD): Likewise.
4910 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
4911 built-in names to 'f128'.
4912 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
4913 old 'q' built-in functions, as the machine independent code for
4914 'f128' built-in functions handles this.
4915 (rs6000_expand_builtin): Add expansion for float128 round to odd
4916 functions, keying off on -mabi=ieeelongdouble of whether to use
4917 the KFmode or TFmode variant.
4918 (rs6000_init_builtins): Initialize the _Float128 round to odd
4920 * doc/extend.texi (PowerPC Built-in Functions): Document the old
4921 _Float128 'q' built-in functions are now mapped into the new
4922 'f128' built-in functions.
4924 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
4926 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
4927 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
4929 2017-11-06 David Malcolm <dmalcolm@redhat.com>
4932 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
4933 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
4934 * toplev.c: Include "ipa-fnsummary.h".
4935 (toplev::finalize): Call ipa_fnsummary_c_finalize.
4937 2017-11-06 Jakub Jelinek <jakub@redhat.com>
4939 PR tree-optimization/82838
4940 * gimple-ssa-store-merging.c
4941 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
4942 on a separate gimple_seq which is then appended to seq.
4944 2017-11-06 Jeff Law <law@redhat.com>
4947 * config/i386/i386.c (PROBE_INTERVAL): Remove.
4948 (get_probe_interval): New functions.
4949 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
4950 (ix86_adjust_stack_and_probe): Likewise.
4951 (output_adjust_stack_and_probe): Likewise.
4952 (ix86_emit_probe_stack_range): Likewise.
4953 (ix86_expand_prologue): Likewise.
4955 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4957 PR tree-optimization/82816
4958 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
4959 if the modes of the two types are the same.
4960 (convert_plusminus_to_widen): Likewise.
4962 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4964 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
4966 (usadv16qi): New define_expand.
4967 (usadv8hi): New define_expand.
4969 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
4972 * ipa-inline-transform.c (update_noncloned_frequencies): Always
4974 (inline_transform): Likewise.
4975 * predict.c (counts_to_freqs): Remove useless conditional.
4976 * profile-count.h (profile_count::apply_scale): Move sanity check.
4977 * tree-inline.c (copy_bb): Always scale.
4978 (copy_cfg_body): Likewise.
4980 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
4983 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
4985 (*cmp_ite0): Add enabled_for_depr_it attribute.
4986 (*cmp_ite1): Likewise.
4988 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4990 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
4993 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4995 * tree-vrp.c (vrp_int_const_binop): Return true on success and
4996 return the value by pointer.
4997 (extract_range_from_multiplicative_op_1): Update accordingly.
4998 Return as soon as an operation fails.
5000 2017-11-05 Tom de Vries <tom@codesourcery.com>
5003 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5004 (DEF_SANITIZER_BUILTIN): ... here.
5005 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5006 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5008 2017-11-05 Tom de Vries <tom@codesourcery.com>
5011 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5013 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5014 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5015 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5017 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5018 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5019 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5021 2017-11-05 Tom de Vries <tom@codesourcery.com>
5024 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5027 2017-11-04 Michael Clark <michaeljclark@mac.com>
5029 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5030 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5032 (*addsi3_extended): Likewise.
5033 (*addsi3_extended2): Likewise.
5034 (<optab>si3): Likewise.
5035 (<optab>di3): Likewise.
5036 (<optab><mode>3): Likewise.
5037 (<*optabe>si3_internal): Likewise.
5038 (zero_extendqi<SUPERQI:mode>2): Likewise.
5039 (*add<mode>hi3): Likewise.
5040 (*xor<mode>hi3): Likewise.
5041 (<optab>di3): Likewise.
5042 (*<optab>si3_extend): Likewise.
5043 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5044 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5045 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5047 2017-11-04 Andrew Waterman <andrew@sifive.com>
5049 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5050 TARGET_STRICT_ALIGN based upon -mtune argument.
5052 2017-11-04 Andrew Waterman <andrew@sifive.com>
5054 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5056 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5058 * config/i386/i386.c (choose_basereg): Use optional scratch
5059 register and add assertion.
5060 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5061 needed, and don't allocate stack.
5062 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5063 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5064 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5066 2017-11-03 Jeff Law <law@redhat.com>
5068 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5069 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5070 to probe at the start of a noreturn function.
5072 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5074 PR tree-optimization/78821
5075 * gimple-ssa-store-merging.c: Update the file comment.
5076 (MAX_STORE_ALIAS_CHECKS): Define.
5077 (struct store_operand_info): New type.
5078 (store_operand_info::store_operand_info): New constructor.
5079 (struct store_immediate_info): Add rhs_code and ops data members.
5080 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5081 and op1r arguments to the ctor, initialize corresponding data members.
5082 (struct merged_store_group): Add load_align_base and load_align
5084 (merged_store_group::merged_store_group): Initialize them.
5085 (merged_store_group::do_merge): Update them.
5086 (merged_store_group::apply_stores): Pick the constant for
5087 encode_tree_to_bitpos from one of the two operands, or skip
5088 encode_tree_to_bitpos if neither operand is a constant.
5089 (class pass_store_merging): Add process_store method decl. Remove
5090 bool argument from terminate_all_aliasing_chains method decl.
5091 (pass_store_merging::terminate_all_aliasing_chains): Remove
5092 var_offset_p argument and corresponding handling.
5093 (stmts_may_clobber_ref_p): New function.
5094 (compatible_load_p): New function.
5095 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5096 if there is overlap and rhs_code is not INTEGER_CST. For
5097 non-overlapping stores terminate group if rhs is not mergeable.
5098 (get_alias_type_for_stmts): Change first argument from
5099 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5100 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5101 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5103 (get_location_for_stmts): Change first argument from
5104 auto_vec<gimple *> & to vec<gimple *> &.
5105 (struct split_store): Remove orig_stmts data member, add orig_stores.
5106 (split_store::split_store): Create orig_stores rather than orig_stmts.
5107 (find_constituent_stmts): Renamed to ...
5108 (find_constituent_stores): ... this. Change second argument from
5109 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5110 to info structures rather than the statements.
5111 (split_group): Rename ALLOW_UNALIGNED argument to
5112 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5113 it. Adjust find_constituent_stores caller.
5114 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5115 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5116 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5117 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5118 (mem_valid_for_store_merging): New function.
5119 (handled_load): New function.
5120 (pass_store_merging::process_store): New method.
5121 (pass_store_merging::execute): Use process_store method. Adjust
5122 terminate_all_aliasing_chains caller.
5124 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5126 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5127 Return true for more constants, symbols and label references.
5128 (aarch64_valid_floating_const): Remove unused function.
5130 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5133 * config/aarch64/aarch64.c (aarch64_layout_frame):
5134 Undo forcing of LR at bottom of frame.
5136 2017-11-03 Jeff Law <law@redhat.com>
5139 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5140 for int_registers_saved.
5142 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5143 extracted from tree-ssa-dom.c.
5144 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5145 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5146 (record_equivalences_from_incoming_edge): Add additional argument
5147 to single_pred_edge_ignoring_loop_edges call.
5148 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5149 (uncprop_dom_walker::before_dom_children): Add additional argument
5150 to single_pred_edge_ignoring_loop_edges call.
5151 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5152 single_pred_edge_ignoring_loop_edges rather than open coding.
5153 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5155 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5157 * match.pd (-(-A)): Rewrite.
5159 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5161 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5162 (rs6000_emit_int_cmove): New declaration.
5163 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5164 (rs6000_emit_sISEL): Delete.
5165 (rs6000_emit_int_cmove): Make non-static.
5166 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5167 instead of rs6000_emit_sISEL.
5169 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5171 * asan.c (create_cond_insert_point): Maintain profile.
5172 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5174 * basic-block.h (struct basic_block_def): Remove frequency.
5175 (EDGE_FREQUENCY): Use to_frequency
5176 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5178 (find_traces): Update to use to_frequency.
5179 (find_traces_1_round): Likewise; use only IPA counts.
5180 (bb_to_key): Likewise.
5181 (connect_traces): Use IPA counts only.
5182 (copy_bb_p): Update to use to_frequency.
5183 (fix_up_crossing_landing_pad): Likewise.
5184 (sanitize_hot_paths): Likewise.
5185 * bt-load.c (basic_block_freq): Likewise.
5186 * cfg.c (init_flow): Set count_max to uninitialized.
5187 (check_bb_profile): Remove frequencies; check counts.
5188 (dump_bb_info): Do not dump frequencies.
5189 (update_bb_profile_for_threading): Update counts only.
5190 (scale_bbs_frequencies_int): Likewise.
5191 (MAX_SAFE_MULTIPLIER): Remove.
5192 (scale_bbs_frequencies_gcov_type): Update counts only.
5193 (scale_bbs_frequencies_profile_count): Update counts only.
5194 (scale_bbs_frequencies): Update counts only.
5195 * cfg.h (struct control_flow_graph): Add count-max.
5196 (update_bb_profile_for_threading): Update prototype.
5197 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5198 (find_many_sub_basic_blocks): Likewise.
5199 * cfgcleanup.c (try_forward_edges): Likewise.
5200 (try_crossjump_to_edge): Likewise.
5201 * cfgexpand.c (expand_gimple_cond): Likewise.
5202 (expand_gimple_tailcall): Likewise.
5203 (construct_init_block): Likewise.
5204 (construct_exit_block): Likewise.
5205 * cfghooks.c (verify_flow_info): Check consistency of counts.
5206 (dump_bb_for_graph): Do not dump frequencies.
5207 (split_block_1): Do not update frequencies.
5208 (split_edge): Do not update frequencies.
5209 (make_forwarder_block): Do not update frequencies.
5210 (duplicate_block): Do not update frequencies.
5211 (account_profile_record): Do not update frequencies.
5212 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5213 for global heuristics.
5214 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5215 (expected_loop_iterations_unbounded): Use counts only.
5216 * cfgloopmanip.c (scale_loop_profile): Simplify.
5217 (create_empty_loop_on_edge): Simplify
5219 (duplicate_loop_to_header_edge): Simplify
5220 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5221 (update_br_prob_note): Take care of removing note when profile
5223 (relink_block_chain): Do not dump frequency.
5224 (rtl_account_profile_record): Use to_frequency.
5225 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5226 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5227 (cgraph_update_edges_for_call_stmt_node): Likewise.
5228 (cgraph_edge::verify_count_and_frequency): Update.
5229 (cgraph_node::verify_node): Temporarily disable frequency verification.
5230 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5231 to_cgraph_frequency.
5232 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5233 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5235 (cgraph_node::expand_thunk): Update profile.
5236 * except.c (dw2_build_landing_pads): Do not update frequency.
5237 * final.c (compute_alignments): Use to_frequency.
5238 (dump_basic_block_info): Do not dump frequency.
5239 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5240 (dump_gimple_bb_header): Do not dump frequency.
5241 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5243 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5244 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5245 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5246 (init_before_recovery): Do not update frequency.
5247 (sched_create_recovery_edges): Do not update frequency.
5248 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5249 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5250 (ipa_cp_c_finalize): Set max_count to uninitialized.
5251 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5252 (param_change_prob): Use counts.
5253 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5255 * ipa-split.c (consider_split): Use to_frequency.
5256 (split_function): Use to_frequency.
5257 * ira-build.c (loop_compare_func): Likewise.
5258 (mark_loops_for_removal): Likewise.
5259 (mark_all_loops_for_removal): Likewise.
5260 * loop-doloop.c (doloop_modify): Do not update frequency.
5261 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5263 * lto-streamer-in.c (input_function): Update count_max.
5264 * omp-expand.c (expand_omp_taskreg): Update count_max.
5265 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5266 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5267 (maybe_hot_count_p): Use ipa counts only.
5268 (maybe_hot_bb_p): Simplify.
5269 (maybe_hot_edge_p): Simplify.
5270 (probably_never_executed): Do not take frequency argument.
5271 (probably_never_executed_bb_p): Do not pass frequency.
5272 (probably_never_executed_edge_p): Likewise.
5273 (combine_predictions_for_bb): Check that profile is nonzero.
5274 (propagate_freq): Do not set frequency.
5275 (drop_profile): Simplify.
5276 (counts_to_freqs): Simplify.
5277 (expensive_function_p): Use to_frequency.
5278 (propagate_unlikely_bbs_forward): Simplify.
5279 (determine_unlikely_bbs): Simplify.
5280 (estimate_bb_frequencies): Add hack to silence graphite issues.
5281 (compute_function_frequency): Use ipa counts.
5282 (pass_profile::execute): Update.
5283 (rebuild_frequencies): Use counts only.
5284 (force_edge_cold): Use counts only.
5285 * profile-count.c (profile_count::dump): Dump new count types.
5286 (profile_count::differs_from_p): Check compatiblity.
5287 (profile_count::to_frequency): New function.
5288 (profile_count::to_cgraph_frequency): New function.
5289 * profile-count.h (struct function): Declare.
5290 (enum profile_quality): Add profile_guessed_local and
5291 profile_guessed_global0.
5292 (class profile_proability): Decrease number of bits to 29;
5293 update from_reg_br_prob_note and to_reg_br_prob_note.
5294 (class profile_count: Update comment; decrease number of bits
5295 to 61. Check compatibility.
5296 (profile_count::compatible_p): New private member function.
5297 (profile_count::ipa_p): New member function.
5298 (profile_count::operator<): Handle global zero correctly.
5299 (profile_count::operator>): Handle global zero correctly.
5300 (profile_count::operator<=): Handle global zero correctly.
5301 (profile_count::operator>=): Handle global zero correctly.
5302 (profile_count::nonzero_p): New member function.
5303 (profile_count::force_nonzero): New member function.
5304 (profile_count::max): New member function.
5305 (profile_count::apply_scale): Handle IPA scalling.
5306 (profile_count::guessed_local): New member function.
5307 (profile_count::global0): New member function.
5308 (profile_count::ipa): New member function.
5309 (profile_count::to_frequency): Declare.
5310 (profile_count::to_cgraph_frequency): Declare.
5311 * profile.c (OVERLAP_BASE): Delete.
5312 (compute_frequency_overlap): Delete.
5313 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5314 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5315 * sched-ebb.c (rank): Use counts only.
5316 * shrink-wrap.c (handle_simple_exit): Use counts only.
5317 (try_shrink_wrapping): Use counts only.
5318 (place_prologue_for_one_component): Use counts only.
5319 * tracer.c (find_best_predecessor): Use to_frequency.
5320 (find_trace): Use to_frequency.
5321 (tail_duplicate): Use to_frequency.
5322 * trans-mem.c (expand_transaction): Do not update frequency.
5323 * tree-call-cdce.c: Do not update frequency.
5324 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5325 (gimple_merge_blocks): Likewise.
5326 (gimple_split_edge): Likewise.
5327 (gimple_duplicate_sese_region): Likewise.
5328 (gimple_duplicate_sese_tail): Likewise.
5329 (move_sese_region_to_fn): Likewise.
5330 (gimple_account_profile_record): Likewise.
5331 (insert_cond_bb): Likewise.
5332 * tree-complex.c (expand_complex_div_wide): Likewise.
5333 * tree-eh.c (lower_resx): Update profile.
5334 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5336 (initialize_cfun): Do not initialize frequencies
5337 (freqs_to_counts): Delete.
5338 (copy_cfg_body): Ignore count parameter.
5339 (copy_body): Update.
5340 (expand_call_inline): Update count_max.
5341 (optimize_inline_calls): Update count_max.
5342 (tree_function_versioning): Update count_max.
5343 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5344 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5346 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5347 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5348 (try_peel_loop): Likewise.
5349 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5351 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5352 (tree_transform_and_unroll_loop): Do not use frequencies
5353 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5354 Use reliable prediction only.
5355 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5356 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5357 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5358 probability scaling.
5359 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5360 not update frequency
5361 (any_remaining_duplicated_blocks): Likewise.
5362 (update_profile): Likewise.
5363 (estimated_freqs_path): Delete.
5364 (freqs_to_counts_path): Delete.
5365 (clear_counts_path): Delete.
5366 (ssa_fix_duplicate_block_edges): Likewise.
5367 (duplicate_thread_path): Likewise.
5368 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5369 * tree-tailcall.c (decrease_profile): Do not update frequency.
5370 (eliminate_tail_call): Likewise.
5371 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5372 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5373 (optimize_mask_stores): Likewise.
5374 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5375 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5376 (ubsan_expand_ptr_ifn): Update profile.
5377 * value-prof.c (gimple_ic): Simplify.
5378 * value-prof.h (gimple_ic): Update prototype.
5379 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5380 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5382 (want_inline_self_recursive_call_p): Likewise.
5383 (resolve_noninline_speculation): Only cummulate defined counts.
5384 (inline_small_functions): Use nonzero_p.
5385 (ipa_inline): Do not access freed node.
5387 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5389 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5390 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5392 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5394 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5395 non-legitimate address.
5397 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5399 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5400 (*lt0_<mode>di, *lt0_<mode>si): New.
5402 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5404 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5405 TARGET_PAIRED_FLOAT.
5407 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5408 Jim Wilson <jim.wilson@linaro.org>
5410 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5411 * config/aarch64/aarch64-tune.md: Regenerated.
5412 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5413 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5415 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5417 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5418 (arc_expand_prologue): Restore blink for millicode.
5419 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5421 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5424 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5425 gen_vec_duplicate after forcing the scalar into a register.
5427 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5429 * combine (try_combine): Print the insns input to try_combine to the
5432 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5435 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5436 Remove second argument from aarch64_process_target_attr call.
5437 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5439 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5441 (aarch64_handle_attr_arch): Remove second argument.
5442 (aarch64_handle_attr_cpu): Ditto.
5443 (aarch64_handle_attr_tune): Ditto.
5444 (aarch64_handle_attr_isa_flags): Ditto.
5445 (aarch64_process_one_target_attr): Ditto.
5446 (aarch64_process_target_attr): Ditto.
5447 (aarch64_option_valid_attribute_p): Remove second argument.
5448 on aarch64_process_target_attr call.
5450 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5452 * diagnostic.c: Include "selftest-diagnostic.h".
5453 (selftest::assert_location_text): New function.
5454 (selftest::test_diagnostic_get_location_text): New function.
5455 (selftest::diagnostic_c_tests): Call it.
5457 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5459 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5460 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5461 (class selftest::test_diagnostic_context): Move to...
5462 * selftest-diagnostic.c: New file.
5463 * selftest-diagnostic.h: New file.
5465 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5467 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5468 offset range for FT32B.
5469 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5470 * config/ft32/ft32.md: Add TARGET_NOPM.
5471 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5472 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5474 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5476 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5478 2017-11-02 Jeff Law <law@redhat.com>
5480 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5481 virtual keyword on FINAL OVERRIDE members.
5483 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5484 virtuals together. Add virtual destructor.
5485 (substitute_and_fold_engine): Similarly.
5487 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5489 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5491 2017-11-02 Richard Biener <rguenther@suse.de>
5493 PR tree-optimization/82795
5494 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5496 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5498 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5500 (gcc_GAS_CHECK_FEATURE): Remove.
5501 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5502 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5503 * configure: Regenerate.
5505 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5507 * config/arc/arc.c (hwloop_optimize): Account for empty
5510 2017-11-02 Richard Biener <rguenther@suse.de>
5513 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5514 Truncate ARRAY_REF index and element size.
5516 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5518 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5520 2017-11-01 Jeff Law <law@redhat.com>
5522 * tree-ssa-ccp.c (ccp_folder): New class derived from
5523 substitute_and_fold_engine.
5524 (ccp_folder::get_value): New member function.
5525 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5526 (ccp_fold_stmt): Remove prototype.
5527 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5528 * tree-ssa-copy.c (copy_folder): New class derived from
5529 substitute_and_fold_engine.
5530 (copy_folder::get_value): Renamed from get_value.
5531 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5532 * tree-vrp.c (vrp_folder): New class derived from
5533 substitute_and_fold_engine.
5534 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5535 (vrp_folder::get_value): New member function.
5536 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5537 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5538 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5539 provide a class interface to folder/substitute routines.
5540 (ssa_prop_fold_stmt_fn): Remove typedef.
5541 (ssa_prop_get_value_fn): Likewise.
5542 (subsitute_and_fold): Remove prototype.
5543 (replace_uses_in): Likewise.
5544 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5545 Renamed from replace_uses_in. Call the virtual member function
5546 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5547 (substitute_and_fold_dom_walker): Remove initialization of
5548 data member entries for calbacks. Add substitute_and_fold_engine
5549 member and initialize it.
5550 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5551 member functions for get_value, replace_phi_args_in c
5552 replace_uses_in, and fold_stmt calls.
5553 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5554 substitute_and_fold. Remove assert. Update ctor call.
5556 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5557 (ssa_prop_visit_phi_fn): Likewise.
5558 (class ssa_propagation_engine): New class to provide an interface
5560 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5562 (ssa_prop_visit_phi): Likewise.
5563 (ssa_propagation_engine::simulate_stmt): Moved into class.
5564 Call visit_phi/visit_stmt from the class rather than via
5565 file scoped static variables.
5566 (ssa_propagation_engine::simulate_block): Moved into class.
5567 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5568 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5569 set file scoped statics for the visit_stmt/visit_phi callbacks.
5570 * tree-complex.c (complex_propagate): New class derived from
5571 ssa_propagation_engine.
5572 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5573 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5574 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5576 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5577 ssa_propagation_engine.
5578 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5579 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5580 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5581 * tree-ssa-copy.c (copy_prop): New class derived from
5582 ssa_propagation_engine.
5583 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5584 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5585 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5586 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5587 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5588 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5589 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5591 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5593 PR rtl-optimization/82778
5594 PR rtl-optimization/82597
5595 * compare-elim.c (struct comparison): Add in_a_setter field.
5596 (find_comparison_dom_walker::before_dom_children): Remove killed
5597 bitmap and df_simulate_find_defs call, instead walk the defs.
5598 Compute last_setter and initialize in_a_setter. Merge definitions
5599 with first initialization for a few variables.
5600 (try_validate_parallel): Use insn_invalid_p instead of
5601 recog_memoized. Return insn rather than just the pattern.
5602 (try_merge_compare): Fix up comment. Don't uselessly test if
5603 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
5605 (execute_compare_elim_after_reload): Remove df_chain_add_problem
5608 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5609 Alan Hayward <alan.hayward@arm.com>
5610 David Sherwood <david.sherwood@arm.com>
5612 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
5613 aarch64_hard_regno_nregs to get the number of registers
5616 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5617 Alan Hayward <alan.hayward@arm.com>
5618 David Sherwood <david.sherwood@arm.com>
5620 * config/aarch64/constraints.md (Upl): Rename to...
5622 * config/aarch64/aarch64.md
5623 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
5626 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5627 Alan Hayward <alan.hayward@arm.com>
5628 David Sherwood <david.sherwood@arm.com>
5630 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
5631 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
5634 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5635 Alan Hayward <alan.hayward@arm.com>
5636 David Sherwood <david.sherwood@arm.com>
5638 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
5639 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
5640 (aarch64_evpc_dup): Generate rtl direcly, rather than using
5642 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
5643 of a single element.
5644 * config/aarch64/iterators.md: Add a comment above the permute
5645 unspecs to say that they are generated directly by
5646 aarch64_expand_vec_perm_const.
5647 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
5649 2017-11-01 Nathan Sidwell <nathan@acm.org>
5651 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
5653 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5655 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
5656 and medany code models, and describe what they do.
5658 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5660 Revert accidental duplicate:
5662 * combine.c (can_change_dest_mode): Reject changes in
5663 REGMODE_NATURAL_SIZE.
5665 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
5667 PR rtl-optimization/64682
5668 PR rtl-optimization/69567
5669 PR rtl-optimization/69737
5670 PR rtl-optimization/82683
5671 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
5672 register mentioned in the note, drop the note, unless it came from I3,
5673 in which case it should go to I3 again.
5675 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5677 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
5678 and return false if not.
5679 (clear_bytes_written_by, live_bytes_read): Update accordingly.
5681 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5683 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
5684 range is known to be empty.
5686 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5687 Alan Hayward <alan.hayward@arm.com>
5688 David Sherwood <david.sherwood@arm.com>
5690 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
5691 and CONST_VECTOR_NUNITS instead of computing the number of units from
5692 the byte sizes of the vector and element.
5693 (simplify_binary_operation_1): Likewise.
5694 (simplify_const_binary_operation): Likewise.
5695 (simplify_ternary_operation): Likewise.
5697 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5698 Alan Hayward <alan.hayward@arm.com>
5699 David Sherwood <david.sherwood@arm.com>
5701 * var-tracking.c (INT_MEM_OFFSET): Replace with...
5702 (int_mem_offset): ...this new function.
5703 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
5704 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
5705 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
5708 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5709 Alan Hayward <alan.hayward@arm.com>
5710 David Sherwood <david.sherwood@arm.com>
5712 * lower-subreg.c (interesting_mode_p): New function.
5713 (compute_costs, find_decomposable_subregs, decompose_register)
5714 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
5715 (resolve_clobber, dump_choices): Use it.
5717 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5718 Alan Hayward <alan.hayward@arm.com>
5719 David Sherwood <david.sherwood@arm.com>
5721 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
5723 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5724 Alan Hayward <alan.hayward@arm.com>
5725 David Sherwood <david.sherwood@arm.com>
5727 * alias.c (find_base_value, find_base_term): Only process integer
5728 truncations. Check the precision rather than the size.
5730 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5731 Alan Hayward <alan.hayward@arm.com>
5732 David Sherwood <david.sherwood@arm.com>
5734 * machmode.h (is_narrower_int_mode): New function
5735 * optabs.c (expand_float, expand_fix): Use it.
5736 * dwarf2out.c (rotate_loc_descriptor): Likewise.
5738 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5739 Alan Hayward <alan.hayward@arm.com>
5740 David Sherwood <david.sherwood@arm.com>
5742 * rtl.h (narrower_subreg_mode): New function.
5743 * ira-color.c (update_costs_from_allocno): Use it.
5745 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5746 Alan Hayward <alan.hayward@arm.com>
5747 David Sherwood <david.sherwood@arm.com>
5749 * optabs-query.h (convert_optab_p): New function, split out from...
5750 (convert_optab_handler): ...here.
5751 (widening_optab_handler): Delete.
5752 (find_widening_optab_handler): Remove permit_non_widening parameter.
5753 (find_widening_optab_handler_and_mode): Likewise. Provide an
5754 override that operates on mode class wrappers.
5755 * optabs-query.c (widening_optab_handler): Delete.
5756 (find_widening_optab_handler_and_mode): Remove permit_non_widening
5757 parameter. Assert that the two modes are the same class and that
5758 the "from" mode is narrower than the "to" mode. Use
5759 convert_optab_handler instead of widening_optab_handler.
5760 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
5761 instead of widening_optab_handler.
5762 * expr.c (expand_expr_real_2): Update calls to
5763 find_widening_optab_handler.
5764 * optabs.c (expand_widen_pattern_expr): Likewise.
5765 (expand_binop_directly): Take the insn_code as a parameter.
5766 (expand_binop): Only call find_widening_optab_handler for
5767 conversion optabs; use optab_handler otherwise. Update calls
5768 to find_widening_optab_handler and expand_binop_directly.
5769 Use convert_optab_handler instead of widening_optab_handler.
5770 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
5771 find_widening_optab_handler and use scalar_mode rather than
5773 (convert_plusminus_to_widen): Likewise.
5775 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5776 Alan Hayward <alan.hayward@arm.com>
5777 David Sherwood <david.sherwood@arm.com>
5779 * machmode.h (fixed_size_mode): New class.
5780 * rtl.h (get_pool_mode): Return fixed_size_mode.
5781 * gengtype.c (main): Add fixed_size_mode.
5782 * target.def (get_raw_result_mode): Return a fixed_size_mode.
5783 (get_raw_arg_mode): Likewise.
5784 * doc/tm.texi: Regenerate.
5785 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
5786 * targhooks.c (default_get_reg_raw_mode): Likewise.
5787 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
5788 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
5789 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
5790 (msp430_get_raw_result_mode): Likewise.
5791 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
5792 * dbxout.c (dbxout_parms): Require fixed-size modes.
5793 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
5794 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5795 * omp-low.c (lower_oacc_reductions): Likewise.
5796 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
5797 (simplify_subreg): Update accordingly.
5798 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
5799 (force_const_mem): Update accordingly. Return NULL_RTX for modes
5800 that aren't fixed-size.
5801 (get_pool_mode): Return a fixed_size_mode.
5802 (output_constant_pool_2): Take a fixed_size_mode.
5804 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5805 Alan Hayward <alan.hayward@arm.com>
5806 David Sherwood <david.sherwood@arm.com>
5808 * doc/rtl.texi (vec_series): Document.
5809 (const): Say that the operand can be a vec_series.
5810 * rtl.def (VEC_SERIES): New rtx code.
5811 * rtl.h (const_vec_series_p_1): Declare.
5812 (const_vec_series_p): New function.
5813 * emit-rtl.h (gen_const_vec_series): Declare.
5814 (gen_vec_series): Likewise.
5815 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
5816 (gen_vec_series): Likewise.
5817 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
5818 * simplify-rtx.c (simplify_unary_operation): Handle negations
5820 (simplify_binary_operation_series): New function.
5821 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
5822 (test_vector_ops_series): New function.
5823 (test_vector_ops): Call it.
5824 * config/powerpcspe/altivec.md (altivec_lvsl): Use
5825 gen_const_vec_series.
5826 (altivec_lvsr): Likewise.
5827 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
5829 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5830 Alan Hayward <alan.hayward@arm.com>
5831 David Sherwood <david.sherwood@arm.com>
5833 * doc/rtl.texi (const): Update description of address constants.
5834 Say that vector constants are allowed too.
5835 * common.md (E, F): Use CONSTANT_P instead of checking for
5837 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
5838 checking for CONST_VECTOR.
5839 * expmed.c (make_tree): Use build_vector_from_val for a CONST
5841 * expr.c (expand_expr_real_2): Check for vector modes instead
5842 of checking for CONST_VECTOR.
5843 * rtl.h (const_vec_p): New function.
5844 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
5845 (unwrap_const_vec_duplicate): Handle them here too.
5847 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5848 David Malcolm <dmalcolm@redhat.com>
5849 Alan Hayward <alan.hayward@arm.com>
5850 David Sherwood <david.sherwood@arm.com>
5852 * rtl.h (vec_duplicate_p): New function.
5853 * selftest-rtl.c (assert_rtx_eq_at): New function.
5854 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
5855 (assert_rtx_eq_at): Declare.
5856 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
5857 * selftest-run-tests.c (selftest::run_tests): Call it.
5858 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
5859 (simplify_unary_operation_1): Recursively handle vector duplicates.
5860 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
5862 (simplify_subreg): Handle subregs of vector duplicates.
5863 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
5864 (selftest::simplify_rtx_c_tests): New functions.
5866 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5867 Alan Hayward <alan.hayward@arm.com>
5868 David Sherwood <david.sherwood@arm.com>
5870 * emit-rtl.h (gen_const_vec_duplicate): Declare.
5871 (gen_vec_duplicate): Likewise.
5872 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
5874 (gen_const_vector): ...here.
5875 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
5876 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
5877 whose elements are all equal.
5878 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
5879 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
5880 (simplify_relational_operation): Likewise.
5881 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5883 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
5884 (aarch64_expand_vector_init): Likewise.
5885 * config/arm/arm.c (neon_vdup_constant): Likewise.
5886 (neon_expand_vector_init): Likewise.
5887 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
5888 (arm_block_set_unaligned_vect): Likewise.
5889 (arm_block_set_aligned_vect): Likewise.
5890 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
5891 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
5892 (expand_vec_perm_even_odd_pack): Likewise.
5893 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
5894 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
5895 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
5896 gen_const_vec_duplicate.
5897 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
5898 * config/mips/mips.c (mips_gen_const_int_vector): Use
5899 gen_const_vec_duplicate.
5900 (mips_expand_vector_init): Use CONST0_RTX.
5901 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
5902 (define_split): Use gen_const_vec_duplicate.
5903 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
5904 (define_split): Use gen_const_vec_duplicate.
5905 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
5906 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
5907 * config/spu/spu.c (spu_const): Likewise.
5909 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5910 Alan Hayward <alan.hayward@arm.com>
5911 David Sherwood <david.sherwood@arm.com>
5913 * combine.c (can_change_dest_mode): Reject changes in
5914 REGMODE_NATURAL_SIZE.
5916 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
5918 * configure.ac (--enable-libssp): New.
5919 (gcc_cv_libc_provides_ssp): Check for explicit setting before
5920 trying to determine target-specific default. Adjust indentation.
5921 * configure: Regenerated.
5922 * doc/install.texi (Configuration): Expand --disable-libssp
5925 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
5927 config/i386/i386.c (ix86_expand_epilogue): Correct stack
5930 2017-10-31 Martin Jambor <mjambor@suse.cz>
5933 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
5935 2017-10-31 David Malcolm <dmalcolm@redhat.com>
5937 * auto-profile.c (autofdo_source_profile::read): Use
5938 UNKNOWN_LOCATION rather than 0.
5939 * diagnostic-core.h (warning_at_rich_loc): Rename to...
5940 (warning_at): ...this overload.
5941 (warning_at_rich_loc_n): Rename to...
5942 (warning_n): ...this overload.
5943 (error_at_rich_loc): Rename to...
5944 (error_at): ...this overload.
5945 (pedwarn_at_rich_loc): Rename to...
5946 (pedwarn): ...this overload.
5947 (permerror_at_rich_loc): Rename to...
5948 (permerror): ...this overload.
5949 (inform_at_rich_loc): Rename to...
5950 (inform): ...this overload.
5951 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
5952 (diagnostic_n_impl_richloc): Rename to...
5953 (diagnostic_n_impl): ...this rich_location *-based decl.
5954 (inform_at_rich_loc): Rename to...
5955 (inform): ...this, and add an assertion.
5956 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
5957 (warning_at_rich_loc): Rename to...
5958 (warning_at): ...this, and add an assertion.
5959 (warning_at_rich_loc_n): Rename to...
5960 (warning_n): ...this, and add an assertion.
5961 (warning_n): Update location_t-based implementation for removal of
5962 location_t-based diagnostic_n_impl.
5963 (pedwarn_at_rich_loc): Rename to...
5964 (pedwarn): ...this, and add an assertion.
5965 (permerror_at_rich_loc): Rename to...
5966 (permerror): ...this, and add an assertion.
5967 (error_n): Update for removal of location_t-based diagnostic_n_impl.
5968 (error_at_rich_loc): Rename to...
5969 (error_at): ...this, and add an assertion.
5970 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
5971 (driver::do_spec_on_infiles): Likewise.
5972 * substring-locations.c (format_warning_va): Update for renaming
5973 of inform_at_rich_loc.
5975 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
5977 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
5978 _Float<N>X built-in functions so that the variant without the
5979 "__builtin_" prefix is only enabled for the GNU C and Objective C
5980 languages when they are in non-strict ANSI/ISO mode.
5981 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
5982 * target.def (floatn_builtin_p): Add a target hook to control
5983 whether _Float<N> and _Float<N>X built-in functions without the
5984 "__builtin_" prefix are enabled, and return true for C and
5985 Objective C in the default hook. Include langhooks.h in
5987 * targhooks.h (default_floatn_builtin_p): Likewise.
5988 * targhooks.c (default_floatn_builtin_p): Likewise.
5989 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
5990 floatn_builtin_p target hook.
5991 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
5993 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
5994 Eric Botcazou <ebotcazou@adacore.com>
5996 PR rtl-optimization/81803
5997 * lra-constraints.c (curr_insn_transform): Also reload the whole
5998 register for a strict subreg no wider than a word if this is for
5999 a WORD_REGISTER_OPERATIONS target.
6001 2017-10-31 Jason Merrill <jason@redhat.com>
6003 * gdbinit.in: Skip over inlines from timevar.h.
6005 2017-10-31 Martin Liska <mliska@suse.cz>
6007 * doc/gcov.texi: Document new option.
6008 * gcov.c (print_usage): Likewise print it.
6009 (process_args): Support the argument.
6010 (format_count): New function.
6011 (format_gcov): Use the function.
6013 2017-10-31 Martin Liska <mliska@suse.cz>
6015 * gcov.c (struct name_map): do not use typedef.
6016 Define operator== and operator<.
6017 (name_search): Remove.
6018 (name_sort): Remove.
6019 (main): Do not allocate names.
6020 (process_file): Add vertical space.
6021 (generate_results): Use std::find.
6022 (release_structures): Do not release memory.
6023 (find_source): Use std::find.
6025 2017-10-31 Martin Liska <mliska@suse.cz>
6027 * gcov.c (struct line_info): Remove it's typedef.
6028 (line_info::line_info): Add proper ctor.
6029 (line_info::has_block): Do not use a typedef.
6030 (struct source_info): Do not use typedef.
6031 (circuit): Likewise.
6032 (get_cycles_count): Likewise.
6033 (output_intermediate_file): Iterate via vector iterator.
6034 (add_line_counts): Use std::vector methods.
6035 (accumulate_line_counts): Likewise.
6036 (output_lines): Likewise.
6038 2017-10-31 Martin Liska <mliska@suse.cz>
6040 * gcov.c (struct source_info): Remove typedef.
6041 (source_info::source_info): Add proper ctor.
6042 (accumulate_line_counts): Use struct, not it's typedef.
6043 (output_gcov_file): Likewise.
6044 (output_lines): Likewise.
6045 (main): Do not allocate an array.
6046 (output_intermediate_file): Use size of vector container.
6047 (process_file): Resize the vector.
6048 (generate_results): Do not preallocate, use newly added vector
6050 (release_structures): Do not release sources.
6051 (find_source): Use vector methods.
6052 (add_line_counts): Do not use typedef.
6054 2017-10-31 Martin Liska <mliska@suse.cz>
6056 * doc/gcov.texi: Document that.
6057 * gcov.c (add_line_counts): Mark lines with a non-executed
6059 (output_line_beginning): Handle such lines.
6060 (output_lines): Pass new argument.
6061 (output_intermediate_file): Print it in intermediate format.
6063 2017-10-31 Martin Liska <mliska@suse.cz>
6065 * color-macros.h: New file.
6066 * diagnostic-color.c: Factor out color related to macros to
6068 * doc/gcov.texi: Document -k option.
6069 * gcov.c (INCLUDE_STRING): Include string.h.
6070 (print_usage): Add -k option.
6071 (process_args): Parse it.
6072 (pad_count_string): New function.
6073 (output_line_beginning): Likewise.
6074 (DEFAULT_LINE_START): New macro.
6075 (output_lines): Support color output.
6077 2017-10-31 Martin Liska <mliska@suse.cz>
6079 PR gcov-profile/82633
6080 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6081 their interaction with GCOV infrastructure.
6082 * configure.ac: Add -fkeep-{inline,static}-functions to
6084 * configure: Regenerate.
6086 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6089 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6091 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6094 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6095 into a register if it does not fit into an immediate offset field.
6097 2017-10-31 Olivier Hainque <hainque@adacore.com>
6099 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6101 2017-10-31 Julia Koval <julia.koval@intel.com>
6103 * config.gcc: Add gfniintrin.h.
6104 * config/i386/gfniintrin.h: New.
6105 * config/i386/i386-builtin-types.def
6106 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6107 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6108 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6109 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6110 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6111 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6112 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6113 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6114 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6115 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6116 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6117 * config/i386/immintrin.h: Include gfniintrin.h.
6118 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6120 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6122 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6124 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6126 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6127 (arm_ashldi3_1bit): Remove pattern.
6128 (ashrdi3): Remove shift by 1 expansion.
6129 (arm_ashrdi3_1bit): Remove pattern.
6130 (lshrdi3): Remove shift by 1 expansion.
6131 (arm_lshrdi3_1bit): Remove pattern.
6132 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6133 cost of ashldi3 by 1.
6134 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6135 (<shift>di3_neon): Likewise.
6137 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6139 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6140 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6141 and (*aarch64_simd_mov<VQ:mode>).
6142 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6143 pattern alternative.
6144 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6145 attributes to match pattern alternative.
6147 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6149 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6150 * config/rs6000/emmintrin.h: New file.
6151 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6153 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6155 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6156 * (movdi_vfp_cortexa8): Remove pattern.
6158 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6160 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6162 (Specific, alpha*-dec-osf5.1): Remove.
6163 (Specific, mips-sgi-irix5): Remove.
6164 (Specific, mips-sgi-irix6): Remove.
6166 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6169 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6170 arguments to clear_bit_region_be.
6172 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6174 * gimplify.c: Include memmodel.h.
6176 2017-10-30 Martin Jambor <mjambor@suse.cz>
6178 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6179 condition whether loop should be executed at all.
6181 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6183 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6184 gimple folding of vec_madd() intrinsics.
6185 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6186 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6187 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6189 2017-10-30 Richard Biener <rguenther@suse.de>
6191 PR tree-optimization/82762
6193 2017-10-23 Richard Biener <rguenther@suse.de>
6195 PR tree-optimization/82129
6197 2017-08-01 Richard Biener <rguenther@suse.de>
6199 PR tree-optimization/81181
6200 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6201 (compute_antic): ... end of iteration here.
6203 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6205 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6206 -std=iso9899:2017 and -std=gnu17.
6207 * doc/standards.texi (C Language): Document C17 support.
6208 * doc/cpp.texi (Overview): Mention -std=c17.
6209 (Standard Predefined Macros): Document C11 and C17 values of
6210 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6211 * doc/extend.texi (Inline): Do not list individual options for
6212 standards newer than C99.
6213 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6215 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6218 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6220 * asan.c (asan_finish_file): Align asan globals array by shadow
6223 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6226 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6227 (struct store_immediate_info): Add bitregion_start and bitregion_end
6229 (store_immediate_info::store_immediate_info): Add brs and bre
6230 arguments and initialize bitregion_{start,end} from those.
6231 (struct merged_store_group): Add bitregion_start, bitregion_end,
6232 align_base and mask fields. Drop unnecessary struct keyword from
6233 struct store_immediate_info. Add do_merge method.
6234 (clear_bit_region_be): Use memset instead of loop storing zeros.
6235 (merged_store_group::do_merge): New method.
6236 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6237 stores as long as the surrounding bitregions have no gaps.
6238 (merged_store_group::merge_overlapping): Use do_merge.
6239 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6240 is byte aligned, rather than requiring that start and width are
6241 byte aligned. Drop unnecessary struct keyword from
6242 struct store_immediate_info. Allocate and populate also mask array.
6243 Make start of the arrays relative to bitregion_start rather than
6244 start and size them according to bitregion_{end,start} difference.
6245 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6246 struct store_immediate_info.
6247 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6248 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6249 struct keyword from struct store_immediate_info.
6250 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6251 between stores as long as the surrounding bitregions have no gaps.
6253 (struct split_store): Add orig non-static data member.
6254 (split_store::split_store): Initialize orig to false.
6255 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6256 if there is exactly a single original stmt. Change stmts argument
6257 to pointer from reference, if NULL, don't push anything to it. Add
6258 first argument, use it to optimize skipping over orig stmts that
6259 are known to be before bitpos already. Simplify.
6260 (split_group): Return unsigned int count how many stores are or
6261 would be needed rather than a bool. Add allow_unaligned argument.
6262 Change split_stores argument from reference to pointer, if NULL,
6263 only do a dry run computing how many stores would be produced.
6264 Rewritten algorithm to use both alignment and misalign if
6265 !allow_unaligned and handle bitfield stores with gaps.
6266 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6267 from bitregion_start instead of start. Compute allow_unaligned
6268 here, if true, do 2 split_group dry runs to compute which one
6269 produces fewer stores and prefer aligned if equal. Punt if
6270 new count is bigger or equal than original before emitting any
6271 statements, rather than during that. Remove no longer needed
6272 new_ssa_names tracking. Replace num_stmts with
6273 split_stores.length (). Use 32-bit stack allocated entries
6274 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6275 if possible. Handle bitfields with gaps.
6276 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6277 Compute bitregion_{start,end} for the stores and construct
6278 store_immediate_info with that. Formatting fixes.
6280 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6283 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6284 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6286 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6288 * gimplify.c: Include tm_p.h.
6290 * common.opt (gcoff): Re-add as ignored option.
6291 (gcoff1, gcoff2, gcoff3): Likewise.
6293 * Makefile.in (OBJS): Delete sdbout.o.
6294 (GTFILES): Delete $(srcdir)/sdbout.c.
6295 * debug.h: Delete sdb_debug_hooks.
6296 * final.c: Delete sdbout.h include.
6297 (final_scan_insn): Delete SDB_DEBUG check.
6298 (rest_of_clean_state): Likewise.
6299 * output.h: Delete sdb_begin_function_line.
6302 * toplev.c: Delete sdbout.h include.
6303 (process_options): Delete SDB_DEBUG check.
6304 * tree-core.h (tree_type_common): Delete pointer field of
6306 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6307 TYPE_SYMTAB_POINTER.
6308 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6309 (TYPE_SYMTAB_IS_POINTER): Delete.
6310 (TYPE_SYMTAB_IS_DIE): Renumber.
6311 * xcoffout.c: Refer to former sdbout.c file.
6312 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6314 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6315 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6316 (Debugging Options): Delete -gcoff.
6317 (-gstabs): Delete SDB reference.
6319 (-gcoff@var{level}): Delete.
6320 * doc/passes.texi (Debugging information output): Delete SDB and
6321 sdbout.c references.
6322 * doc/tm.texi: Regenerate.
6323 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6324 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6326 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6327 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6328 (SDB_DEBUGGING_INFO): Delete.
6329 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6330 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6331 * target.def (output_source_filename): Delete COFF reference.
6333 * common.opt (gcoff): Delete.
6334 (gxcoff+): Update Negative chain.
6335 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6337 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6338 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6339 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6341 (expand_function_start): Change sdb reference to past tense.
6342 (expand_function_end): Change sdb reference to past tense.
6343 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6344 * opts.c (debug_type_names): Delete coff entry.
6345 (common_handle_option): Delete OPT_gcoff case.
6346 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6348 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6349 * config/cris/cris.h: Delete SDB reference in comment.
6350 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6351 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6352 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6353 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6355 (ix86_expand_prologue): Likewise.
6356 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6357 * config/ia64/ia64.h: Likewise.
6358 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6359 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6361 * config/mmix/mmix.h: Likewise.
6362 * config/nds32/nds32.c: Likewise.
6363 * config/stormy/storym16.h: Likewise.
6364 * config/visium/visium.h: Likewise.
6365 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6367 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6369 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6370 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6371 FRAME_POINTER_REGNUM point at high end of local var area.
6373 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6375 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6376 Move comment around. Do not reset best_edge for a copiable
6377 destination if the copy would cause a partition change.
6378 (better_edge_p): Remove redundant check.
6380 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6382 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6384 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6386 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6387 for math functions that have _Float<N> and _Float<N>X variants.
6388 (mathfn_built_in_2): Add support for math functions that have
6389 _Float<N> and _Float<N>X variants.
6390 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6391 (expand_builtin_mathfn_ternary): Add support for fma with
6392 _Float<N> and _Float<N>X variants.
6393 (expand_builtin): Likewise.
6394 (fold_builtin_3): Likewise.
6395 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6396 create math function _Float<N> and _Float<N>X variants as external
6398 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6399 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6400 the __builtin_ prefix and if not strict ansi, without the prefix.
6401 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6402 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6403 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6404 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6405 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6406 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6407 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6408 function signatures for fma _Float<N> and _Float<N>X variants.
6409 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6410 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6411 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6412 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6413 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6414 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6415 * gencfn-macros.c (print_case_cfn): Add support for math functions
6416 that have _Float<N> and _Float<N>X variants.
6417 (print_define_operator_list): Likewise.
6418 (fltfn_suffixes): Likewise.
6420 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6421 for math functions that have _Float<N> and _Float<N>X variants.
6422 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6423 and _Float<N>X variants.
6424 (COPYSIGN): Likewise.
6427 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6428 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6430 (integer_valued_read_call_p): Likewise.
6431 * fold-const-call.c (fold_const_call_ss): Likewise.
6432 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6433 _Float<N> and _Float<N>X variants.
6434 (fold_const_call_ssss): Add support for fma _Float<N> and
6435 _Float<N>X variants.
6436 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6437 support for copysign and fma _Float<N> and _Float<N>X variants.
6438 (backprop::process_builtin_call_use): Likewise.
6439 * tree-call-cdce.c (can_test_argument_range); Add support for
6440 sqrt _Float<N> and _Float<N>X variants.
6441 (edom_only_function): Likewise.
6442 (get_no_error_domain): Likewise.
6443 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6444 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6445 copysign _Float<N> and _Float<N>X variants.
6446 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6447 handled by machine independent code.
6448 (FMAF128): Likewise.
6449 * doc/cpp.texi (Common Predefined Macros): Document defining
6450 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6451 fma _Float<N> and _Float<N>X variants.
6453 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6456 * config/i386/i386-modes.def (CCFPU): Remove definition.
6457 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6458 (ix86_cc_modes_compatible): Ditto.
6459 (ix86_expand_carry_flag_compare): Ditto.
6460 (ix86_expand_int_movcc): Ditto.
6461 (ix86_expand_int_addcc): Ditto.
6462 (ix86_reverse_condition): Ditto.
6463 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6464 Return true/false for unordered/ordered fp comparisons.
6465 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6466 (ix86_prepare_fp_compare_args): Update for rename.
6467 (ix86_expand_fp_compare): Update for rename. Generate unordered
6468 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6469 (ix86_expand_sse_compare_and_jump): Ditto.
6470 * config/i386/predicates.md (fcmov_comparison_operator):
6471 Remove CCFPU mode handling.
6472 (ix86_comparison_operator): Ditto.
6473 (ix86_carry_flag_operator): Ditto.
6474 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6475 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6476 (*cmpu<mode>_cc_i387): Ditto.
6477 (FPCMP): Remove mode iterator.
6478 (unord): Remove mode attribute.
6479 (unord_subst): New define_subst transformation
6480 (unord): New define_subst attribute.
6482 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6483 (*cmpi<unord>xf_i387): Ditto.
6484 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6485 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6486 using unord_subst transformation.
6487 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6488 (round_saeonly): Also handle CCFP mode.
6489 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6490 Remove UNSPEC_SAHF unspec handling.
6492 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6494 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6496 2017-10-27 Jeff Law <law@redhat.com>
6498 * gimple-ssa-sprintf.c: Include domwalk.h.
6499 (class sprintf_dom_walker): New class, derived from dom_walker.
6500 (sprintf_dom_walker::before_dom_children): New function.
6501 (struct call_info): Moved into sprintf_dom_walker class
6502 (compute_formath_length, handle_gimple_call): Likewise.
6503 (sprintf_length::execute): Call the dominator walker rather
6504 than walking the statements.
6506 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6507 gimple statement locations.
6508 (check_array_bounds): Corresponding changes. Get the statement's
6509 location directly from wi->stmt.
6511 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6514 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6516 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6518 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6519 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6521 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6524 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6525 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6526 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6527 maybe_get_pool_constant.
6528 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6531 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6533 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6534 2.26 caveat. Update gas and gld versions.
6535 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6538 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6540 * cgraph.h (set_malloc_flag): Declare.
6541 * cgraph.c (set_malloc_flag_1): New function.
6542 (set_malloc_flag): Likewise.
6543 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6544 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6546 (read_ipa_call_summary): Add support for reading is_return_callee.
6547 (write_ipa_call_summary): Stream is_return_callee.
6548 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6549 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6550 ipa-prop.h, ipa-fnsummary.h.
6551 (pure_const_names): Change to static.
6552 (malloc_state_e): Define.
6553 (malloc_state_names): Define.
6554 (funct_state_d): Add field malloc_state.
6555 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6556 (check_retval_uses): New function.
6557 (malloc_candidate_p): Likewise.
6558 (analyze_function): Add support for malloc attribute.
6559 (pure_const_write_summary): Stream malloc_state.
6560 (pure_const_read_summary): Add support for reading malloc_state.
6561 (dump_malloc_lattice): New function.
6562 (propagate_malloc): New function.
6563 (warn_function_malloc): New function.
6564 (ipa_pure_const::execute): Call propagate_malloc and
6565 ipa_free_fn_summary.
6566 (pass_local_pure_const::execute): Add support for malloc attribute.
6567 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6568 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6570 2017-10-27 Martin Liska <mliska@suse.cz>
6572 PR gcov-profile/82457
6573 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6574 for fork-like functions to be properly instrumented.
6576 2017-10-27 Richard Biener <rguenther@suse.de>
6579 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6580 info when we redirected EH.
6582 2017-10-26 Michael Collison <michael.collison@arm.com>
6584 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6586 (<optab>_trunchf<GPI:mode>2: New pattern.
6587 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6588 * config/aarch64/iterators.md (wv): New mode attribute.
6589 (vf, VF): New mode attributes.
6590 (vgp, VGP): New mode attributes.
6591 (s): Update attribute with SImode and DImode prefixes.
6593 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6595 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
6596 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
6597 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
6598 (nios2_option_overide): Initialize it. Don't allow R0-relative
6599 addressing with PIC.
6600 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
6601 (nios2_symbolic_constant_p): Likewise.
6602 (nios2_legitimate_address_p): Likewise.
6603 (nios2_r0rel_section_name_p): New.
6604 (nios2_symbol_ref_in_r0rel_data_p): New.
6605 (nios2_emit_move_sequence): Handle r0rel_constant_p.
6606 (r0rel_constant_p): New.
6607 (nios2_print_operand_address): Handle r0rel_constant_p.
6608 (nios2_cdx_narrow_form_p): Likewise.
6609 * config/nios2/nios2.opt (mr0rel-sec=): New option.
6610 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
6611 (Nios II Options): Document -mr0rel-sec.
6613 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6615 * config/nios2/nios2.c: Include xregex.h.
6616 (nios2_gprel_sec_regex): New.
6617 (nios2_option_overide): Initialize it. Don't allow GP-relative
6618 addressing with PIC.
6619 (nios2_small_section_name_p): Check for regex match.
6620 * config/nios2/nios2.opt (mgprel-sec=): New option.
6621 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
6622 (Nios II Options): Document -mgprel-sec.
6624 2017-10-26 Jim Wilson <wilson@tuliptree.org>
6626 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
6628 2017-10-26 Tom de Vries <tom@codesourcery.com>
6630 PR tree-optimization/82707
6631 * gimple.c (gimple_copy): Fix unsharing of
6632 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
6634 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
6636 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
6637 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
6638 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
6639 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
6640 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
6641 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
6642 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
6643 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
6644 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
6645 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
6646 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
6647 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
6648 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
6649 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
6650 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
6651 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
6652 _mm512_mask_cmpunord_ps_mask): New intrinsics.
6654 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6656 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
6658 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
6659 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
6660 warning to rs6000.c. Remove the Undocumented flag, since it has
6662 (-mabi=ibmlongdouble): Likewise.
6663 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
6664 already set, set the default format for long double.
6665 (rs6000_debug_reg_global): Print whether long double is IBM or
6667 (rs6000_option_override_internal): Rework setting long double
6668 format. Only warn if the user is changing the long double default
6669 and they did not use -Wno-psabi.
6670 * doc/invoke.texi (PowerPC options): Update the documentation for
6671 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
6673 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6674 Alan Hayward <alan.hayward@arm.com>
6675 David Sherwood <david.sherwood@arm.com>
6677 * rtl.h (wider_subreg_mode): New function.
6678 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
6679 rather than an unsigned int *.
6680 * ira-color.c (regno_max_ref_width): Replace with...
6681 (regno_max_ref_mode): ...this new variable.
6682 (coalesced_pseudo_reg_slot_compare): Update accordingly.
6683 Use wider_subreg_mode.
6684 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
6685 rather than an unsigned int *.
6686 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
6687 (process_alt_operands): Likewise.
6688 (invariant_p): Likewise.
6689 * lra-spills.c (assign_mem_slot): Likewise.
6690 (add_pseudo_to_slot): Likewise.
6691 * lra.c (collect_non_operand_hard_regs): Likewise.
6692 (add_regs_to_insn_regno_info): Likewise.
6693 * reload1.c (regno_max_ref_width): Replace with...
6694 (regno_max_ref_mode): ...this new variable.
6695 (reload): Update accordingly. Update call to
6696 ira_sort_regnos_for_alter_reg.
6697 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
6698 (init_eliminable_invariants): Update to use regno_max_ref_mode.
6699 (scan_paradoxical_subregs): Likewise.
6701 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6703 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
6704 (aarch64_frame): Add emit_frame_chain boolean.
6705 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6706 Move eh_return case to aarch64_layout_frame.
6707 (aarch64_layout_frame): Initialize emit_frame_chain.
6708 (aarch64_expand_prologue): Use emit_frame_chain.
6710 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6712 * config/aarch64/aarch64.c (aarch64_layout_frame):
6713 Ensure LR is always stored at the bottom of the callee-saves.
6714 Remove rarely used frame layout which saves callee-saves at top of
6715 frame, so the store of LR can be used as a valid probe in all cases.
6717 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6719 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
6720 Improve unaligned TImode/TFmode base/offset split.
6722 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6723 Alan Hayward <alan.hayward@arm.com>
6724 David Sherwood <david.sherwood@arm.com>
6726 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
6727 * combine.c (find_single_use_1): Likewise.
6728 (expand_field_assignment): Likewise.
6729 (move_deaths): Likewise.
6730 * lra-constraints.c (simplify_operand_subreg): Likewise.
6731 (curr_insn_transform): Likewise.
6732 * lra.c (collect_non_operand_hard_regs): Likewise.
6733 (add_regs_to_insn_regno_info): Likewise.
6734 * rtlanal.c (reg_referenced_p): Likewise.
6735 (covers_regno_no_parallel_p): Likewise.
6737 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6739 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
6740 Don't print any bits outside the precision of the value.
6741 * wide-int.cc (test_printing): Add some new tests.
6743 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6745 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
6746 supports -xbrace_comment option.
6747 * configure: Regenerate.
6748 * config.in: Regenerate.
6749 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
6750 (ASM_CPU_SPEC): Use it.
6752 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6754 * target.def (static_rtx_alignment): New hook.
6755 * targhooks.h (default_static_rtx_alignment): Declare.
6756 * targhooks.c (default_static_rtx_alignment): New function.
6757 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
6758 * doc/tm.texi: Regenerate.
6759 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
6760 instead of targetm.constant_alignment. Remove call to
6762 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6763 (cris_preferred_mininum_alignment): New function, split out from...
6764 (cris_constant_alignment): ...here.
6765 (cris_static_rtx_alignment): New function.
6766 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
6768 (ix86_constant_alignment): ...here.
6769 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6770 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6771 (mmix_static_rtx_alignment): New function.
6772 * config/spu/spu.c (spu_static_rtx_alignment): New function.
6773 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6775 2017-10-26 Tamar Christina <tamar.christina@arm.com>
6778 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
6779 Add flag_trapping_math and flag_fp_int_builtin_inexact.
6781 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
6783 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
6784 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
6785 mark as a sign-extending load.
6786 (local_pic_load_u): Define.
6788 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
6791 * fold-const.c (operand_equal_for_comparison_p): Also return true
6792 if ARG0 is a simple variant of ARG1 with narrower precision.
6793 (fold_ternary_loc): Always pass unstripped operands to the predicate.
6795 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
6797 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
6799 * i386.h (processor_costs): Add gather_static, gather_per_elt,
6800 scatter_static, scatter_per_elt.
6801 * x86-tune-costs.h: Add new cost entries.
6803 2017-10-25 Richard Biener <rguenther@suse.de>
6805 * tree-ssa-sccvn.h (vn_eliminate): Declare.
6806 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
6807 class pass_fre): Move to ...
6808 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
6809 class pass_fre): ... here and adjust for statistics.
6811 2017-10-25 Jakub Jelinek <jakub@redhat.com>
6814 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
6815 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
6816 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
6817 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
6820 2017-10-25 Richard Biener <rguenther@suse.de>
6822 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
6823 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
6824 eliminate_push_avail, eliminate_insert): Move inside...
6825 (class eliminate_dom_walker): ... this class in preparation
6827 (fini_eliminate): Remove by merging with ...
6828 (eliminate): ... this function. Adjust for class changes.
6829 (pass_pre::execute): Remove fini_eliminate call.
6830 (pass_fre::execute): Likewise.
6832 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6835 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
6836 (VPERMI2, VPERMI2I): New mode iterators.
6837 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
6838 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
6840 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
6841 mode iterator. Remove 3 old define_insn patterns.
6842 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
6843 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
6844 VPERMI2 mode iterator, remove the other two expanders.
6845 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
6846 to use VPERMI2 mode iterator, add another alternative for vpermi2*
6847 instructions, remove the other two patterns.
6848 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
6849 mode iterator, remove the other two patterns.
6850 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
6851 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
6852 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
6853 and adjust argument order accordingly.
6854 (ix86_expand_vec_perm): Adjust caller.
6855 (expand_vec_perm_1): Likewise.
6856 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
6857 (expand_vec_perm_vpermt2_vpshub2): ... this.
6858 (ix86_expand_vec_perm_const_1): Adjust caller.
6859 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
6862 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
6863 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
6864 (vec_shl_<mode>): Remove unused expander.
6865 (avx512bw_<shift_insn><mode>3): New define_insn.
6866 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
6867 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
6869 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
6872 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
6875 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6877 PR rtl-optimization/82396
6878 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
6879 (autopref_multipass_init): Simplify initialization.
6880 (autopref_rank_data): Simplify sort order.
6881 * gcc/sched-int.h (autopref_multipass_data_): Remove
6882 multi_mem_insn_p, min_offset and max_offset.
6884 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6887 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6888 Check special value of flag_omit_frame_pointer.
6889 (aarch64_can_eliminate): Likewise.
6890 (aarch64_override_options_after_change_1): Simplify handling of
6891 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
6893 2017-10-24 Richard Biener <rguenther@suse.de>
6895 PR tree-optimization/82697
6896 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
6897 zero for conditional load and unconditional store.
6899 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6901 * doc/install.texi: Document bootstrap-cet.
6903 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6906 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
6907 ENDBR instruction at function entrance if function is only
6910 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6913 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
6914 patterns to better describe from which operation the CF is computed.
6915 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
6916 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
6917 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
6918 is 0, use _0 suffixed expanders instead of emitting a comparison
6921 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
6923 * config/i386/i386.md(*movsf_internal, *movdf_internal):
6924 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
6926 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
6929 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
6930 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
6931 * loop-iv.c (iv_get_reaching_def): Likewise.
6932 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
6933 variable is promoted and the partition contains undefined values.
6935 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6937 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
6939 (nios2_address_cost): Define.
6940 (nios2_legitimize_address): Recognize (exp + constant) directly.
6941 (TARGET_ADDRESS_COST): Define.
6943 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6945 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
6946 (nios2_symbolic_memory_operand_p): Declare.
6947 (nios2_split_large_constant): Declare.
6948 (nios2_split_symbolic_memory_operand): Declare.
6949 * config/nios2/nios2.c: Adjust includes.
6950 (nios2_symbolic_constant_allowed): New.
6951 (nios2_symbolic_constant_p): New.
6952 (nios2_plus_symbolic_constant_p): New.
6953 (nios2_valid_addr_expr_p): Recognize addresses involving
6955 (nios2_legitimate_address_p): Likewise, also LO_SUM.
6956 (nios2_symbolic_memory_operand_p): New.
6957 (nios2_large_constant_p): New.
6958 (nios2_split_large_constant): New.
6959 (nios2_split_plus_large_constant): New.
6960 (nios2_split_symbolic_memory_operand): New.
6961 (nios2_legitimize_address): Code refactoring. Handle addresses
6962 involving symbolic constants.
6963 (nios2_emit_move_sequence): Likewise.
6964 (nios2_print_operand): Improve error output.
6965 (nios2_print_operand_address): Handle LO_SUM.
6966 (nios2_cdx_narrow_form_p): Likewise.
6967 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
6968 operands involving symbolic constants.
6969 (movhi_internal, movsi_internal): Likewise.
6970 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
6971 (extendhisi2, extendqi<mode>2): Likewise.
6973 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6975 * tree-pass.h (PROP_rtl_split_insns): Define.
6976 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
6978 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6980 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
6982 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6985 * target.def (const_not_ok_for_debug_p): Default to
6986 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
6987 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
6988 * targhooks.c (default_const_not_ok_for_debug_p): New function.
6989 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
6990 which targetm.const_not_ok_for_debug_p returned true.
6991 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
6993 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
6995 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
6996 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
6997 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
6999 (ix86_const_not_ok_for_debug_p): New function.
7000 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7001 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7003 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7006 * system.h: Conditionally include "unique-ptr.h" if
7007 INCLUDE_UNIQUE_PTR is defined.
7008 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7009 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7011 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7013 * config/rl78/rl78.md: New define_expand "subdi3".
7015 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7018 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7019 DF_REF_INSN if DF_REF_INSN_INFO is false.
7021 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7023 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7024 xmm_move instead of sse_move.
7025 (sse_store_index): New function.
7026 (ix86_register_move_cost): Be more sensible about mismatch stall;
7027 model AVX moves correctly; make difference between sse->integer and
7029 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7030 moves; make difference between SSE and AVX.
7031 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7032 and zmm_move. Increase size of sse load and store tables;
7033 add unaligned load and store tables; add ssemmx_to_integer.
7034 * x86-tune-costs.h: Update all entries according to real
7035 move latencies from Agner Fog's manual and chip documentation.
7037 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7040 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7041 * config/i386/constraints.md (Wf): New constraint.
7042 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7043 (cmp<dwi>_doubleword): Removed.
7044 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7045 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7046 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7047 expand with cmp<dwi>_doubleword. For LTU and GEU use
7048 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7050 * common.opt (gcolumn-info): Enable by default.
7051 * doc/invoke.texi (gcolumn-info): Document new default.
7053 2017-10-23 Richard Biener <rguenther@suse.de>
7055 PR tree-optimization/82672
7056 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7057 Fold the stmt if we propagated into it.
7059 2017-10-23 Richard Biener <rguenther@suse.de>
7061 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7062 (bitmap_remove_expr_from_set): ... this. All callers call this
7063 for non-constant values.
7064 (bitmap_set_subtract): Rename to...
7065 (bitmap_set_subtract_expressions): ... this. Adjust and
7067 (bitmap_set_contains_value): Remove superfluous check.
7068 (bitmap_set_replace_value): Inline into single caller ...
7069 (bitmap_value_replace_in_set): ... here and simplify.
7070 (dependent_clean): Merge into ...
7071 (clean): ... this using an overload. Adjust.
7072 (prune_clobbered_mems): Adjust.
7073 (compute_antic_aux): Likewise.
7074 (compute_partial_antic_aux): Likewise.
7076 2017-10-23 Richard Biener <rguenther@suse.de>
7078 PR tree-optimization/82129
7080 2017-08-01 Richard Biener <rguenther@suse.de>
7082 PR tree-optimization/81181
7083 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7084 (compute_antic): ... end of iteration here.
7086 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7088 * target.def (starting_frame_offset): New hook.
7089 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7090 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7091 * doc/tm.texi.in: Regenerate.
7092 * hooks.h (hook_hwi_void_0): Declare.
7093 * hooks.c (hook_hwi_void_0): New function.
7094 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7095 STARTING_FRAME_OFFSET.
7096 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7097 * reload1.c (reload): Likewise.
7098 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7099 instead of STARTING_FRAME_OFFSET.
7100 * function.c (try_fit_stack_local): Likewise.
7101 (assign_stack_local_1): Likewise
7102 (instantiate_virtual_regs): Likewise.
7103 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7104 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7105 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7106 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7107 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7108 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7109 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7110 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7111 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7112 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7113 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7114 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7115 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7116 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7117 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7118 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7119 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7120 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7121 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7122 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7123 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7124 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7125 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7126 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7127 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7128 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7129 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7130 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7131 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7132 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7133 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7134 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7135 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7136 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7137 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7138 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7139 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7140 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7141 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7142 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7143 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7144 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7145 return a HOST_WIDE_INT.
7146 (avr_builtin_setjmp_frame_value): Use it instead of
7147 STARTING_FRAME_OFFSET.
7148 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7149 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7150 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7152 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7153 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7154 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7155 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7156 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7157 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7158 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7159 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7160 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7161 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7162 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7163 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7165 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7166 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7167 * config/mips/mips.c (mips_compute_frame_info): Refer to
7168 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7169 (mips_starting_frame_offset): New function.
7170 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7171 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7172 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7173 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7174 and return a HOST_WIDE_INT.
7175 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7176 (mmix_initial_elimination_offset): Refer to
7177 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7178 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7179 * config/pa/pa.c (pa_starting_frame_offset): New function.
7180 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7181 (pa_expand_prologue): Likewise.
7182 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7183 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7184 !FRAME_GROWS_DOWNWARD handling to...
7185 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7186 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7187 !FRAME_GROWS_DOWNWARD handling to...
7188 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7189 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7190 !FRAME_GROWS_DOWNWARD handling to...
7191 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7192 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7194 (rs6000_starting_frame_offset): New function.
7195 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7196 !FRAME_GROWS_DOWNWARD handling to...
7197 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7198 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7199 !FRAME_GROWS_DOWNWARD handling to...
7200 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7201 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7202 !FRAME_GROWS_DOWNWARD handling to...
7203 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7204 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7205 (rs6000_starting_frame_offset): New function.
7206 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7207 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7208 * config/vax/vax.c (vax_starting_frame_offset): New function.
7209 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7210 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7211 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7212 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7213 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7214 * system.h (STARTING_FRAME_OFFSET): Poison.
7216 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7218 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7219 SCALAR_TYPE_MODE instead of TYPE_MODE.
7221 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7222 Alan Hayward <alan.hayward@arm.com>
7223 David Sherwood <david.sherwood@arm.com>
7225 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7227 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7228 Alan Hayward <alan.hayward@arm.com>
7229 David Sherwood <david.sherwood@arm.com>
7231 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7233 2017-10-23 Richard Biener <rguenther@suse.de>
7235 PR tree-optimization/82129
7236 * tree-ssa-pre.c (bitmap_set_and): Remove.
7237 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7238 canonicalizing expressions in the set to those with lowest
7239 ID rather than taking that from the first edge.
7241 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7243 * combine.c (rtx_equal_for_field_assignment_p): Use
7244 byte_lowpart_offset.
7246 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7247 Alan Hayward <alan.hayward@arm.com>
7248 David Sherwood <david.sherwood@arm.com>
7250 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7251 to a SUBREG_PROMOTED_VAR.
7253 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7254 Alan Hayward <alan.hayward@arm.com>
7255 David Sherwood <david.sherwood@arm.com>
7257 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7258 (expand_debug_source_expr): Likewise.
7259 * combine.c (combine_simplify_rtx): Likewise.
7260 * cse.c (fold_rtx): Likewise.
7261 * optabs.c (expand_float): Likewise.
7262 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7263 (simplify_binary_operation_1): Likewise.
7265 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7266 Alan Hayward <alan.hayward@arm.com>
7267 David Sherwood <david.sherwood@arm.com>
7269 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7270 (record_promoted_value): Likewise.
7271 * expr.c (expand_expr_real_2): Likewise.
7272 * ree.c (update_reg_equal_equiv_notes): Likewise.
7273 (combine_set_extension): Likewise.
7274 * rtlanal.c (low_bitmask_len): Likewise.
7275 * simplify-rtx.c (neg_const_int): Likewise.
7276 (simplify_binary_operation_1): Likewise.
7278 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7279 Alan Hayward <alan.hayward@arm.com>
7280 David Sherwood <david.sherwood@arm.com>
7282 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7283 * regcprop.c (maybe_mode_change): Likewise.
7284 * reload1.c (alter_reg): Likewise.
7286 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7288 * inchash.h (inchash::hash::add_wide_int): New function.
7289 * lto-streamer-out.c (hash_tree): Use it.
7291 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7293 * inchash.h (inchash::hash::add_wide_int): Rename to...
7294 (inchash::hash::add_hwi): ...this.
7295 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7296 (polymorphic_call_target_hasher::hash): Likewise.
7297 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7298 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7299 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7300 * lto-streamer-out.c (hash_tree): Likewise.
7301 * optc-save-gen.awk: Likewise.
7302 * tree.c (add_expr): Likewise.
7304 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7307 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7308 for ordered inequality comparisons even with TARGET_IEEE_FP.
7310 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7313 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7314 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7315 Expand with cmp<dwi>_doubleword.
7317 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7319 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7320 List CET intrinsics.
7321 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7322 specific to -fcf-protection option.
7324 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7326 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7327 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7328 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7329 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7330 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7331 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7332 (extra_objs): Add cet.o for Linux/x86 targets.
7333 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7334 * config/i386/cet.c: New file.
7335 * config/i386/cetintrin.h: Likewise.
7336 * config/i386/t-cet: Likewise.
7337 * config/i386/cpuid.h (bit_SHSTK): New.
7338 (bit_IBT): Likewise.
7339 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7340 pass IBT and SHSTK bits.
7341 * config/i386/i386-builtin-types.def
7342 (VOID_FTYPE_UNSIGNED_PVOID): New.
7343 (VOID_FTYPE_UINT64_PVOID): Likewise.
7344 * config/i386/i386-builtin.def: Add CET intrinsics.
7345 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7346 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7347 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7348 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7350 * config/i386/i386.c (rest_of_insert_endbranch): New.
7351 (pass_data_insert_endbranch): Likewise.
7352 (pass_insert_endbranch): Likewise.
7353 (make_pass_insert_endbranch): Likewise.
7354 (ix86_notrack_prefixed_insn_p): Likewise.
7355 (ix86_target_string): Add -mibt, -mshstk flags.
7356 (ix86_option_override_internal): Add flag_cf_protection
7358 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7359 (ix86_print_operand): Add 'notrack' prefix output.
7360 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7361 (ix86_expand_builtin): Expand CET intrinsics.
7362 (x86_output_mi_thunk): Add 'endbranch' instruction.
7363 * config/i386/i386.h (TARGET_IBT): New.
7364 (TARGET_IBT_P): Likewise.
7365 (TARGET_SHSTK): Likewise.
7366 (TARGET_SHSTK_P): Likewise.
7367 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7368 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7369 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7370 (builtin_setjmp_setup): New pattern.
7371 (builtin_longjmp): Likewise.
7372 (rdssp<mode>): Likewise.
7373 (incssp<mode>): Likewise.
7374 (saveprevssp): Likewise.
7375 (rstorssp): Likewise.
7376 (wrss<mode>): Likewise.
7377 (wruss<mode>): Likewise.
7378 (setssbsy): Likewise.
7379 (clrssbsy): Likewise.
7380 (nop_endbr): Likewise.
7381 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7383 * config/i386/immintrin.h: Include <cetintrin.h>.
7384 * config/i386/linux-common.h
7385 (file_end_indicate_exec_stack_and_cet): New prototype.
7386 (TARGET_ASM_FILE_END): New.
7388 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7390 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7391 latencies instead of having separate table; make difference between
7392 integer and float costs.
7393 * i386.h (processor_costs): Remove scalar_stmt_cost,
7394 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7395 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7397 * x86-tune-costs.h: Remove entries which has been removed in
7398 procesor_costs from all tables; make cond_taken_branch_cost
7399 and cond_not_taken_branch_cost COST_N_INSNS based.
7401 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7403 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7405 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7407 * config/i386/i386.md (isa): Remove fma_avx512f.
7408 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7409 <avx512>_fmadd_<mode>_mask3<round_name>,
7410 <avx512>_fmsub_<mode>_mask<round_name>,
7411 <avx512>_fmsub_<mode>_mask3<round_name>,
7412 <avx512>_fnmadd_<mode>_mask<round_name>,
7413 <avx512>_fnmadd_<mode>_mask3<round_name>,
7414 <avx512>_fnmsub_<mode>_mask<round_name>,
7415 <avx512>_fnmsub_<mode>_mask3<round_name>,
7416 <avx512>_fmaddsub_<mode>_mask<round_name>,
7417 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7418 <avx512>_fmsubadd_<mode>_mask<round_name>,
7419 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7420 (*vec_widen_umult_even_v16si<mask_name>,
7421 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7422 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7424 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7426 * extend.texi: Add 'nocf_check' documentation.
7427 * gimple.texi: Add second parameter to
7428 gimple_build_call_from_tree.
7429 * invoke.texi: Add -fcf-protection documentation.
7430 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7432 2017-10-20 Richard Biener <rguenther@suse.de>
7434 PR tree-optimization/82473
7435 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7436 the largest input type.
7438 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7440 * c-attribs.c (handle_nocf_check_attribute): New function.
7441 (c_common_attribute_table): Add 'nocf_check' handling.
7442 * gimple-parser.c: Add second argument NULL to
7443 gimple_build_call_from_tree.
7444 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7445 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7447 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7448 * common.opt: Add fcf-protection flag.
7449 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7450 * flag-types.h: Add enum cf_protection_level.
7451 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7452 Add 'nocf_check' attribute propagation to gimple call.
7453 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7454 (gimple_build_call_from_tree): Update prototype.
7455 (gimple_call_nocf_check_p): New function.
7456 (gimple_call_set_nocf_check): Likewise.
7457 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7458 * ipa-icf.c: Add nocf_check attribute in statement hash.
7459 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7460 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7461 * toplev.c (process_options): Add flag_cf_protection handling.
7463 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7465 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7467 2017-10-20 Richard Biener <rguenther@suse.de>
7469 PR tree-optimization/82603
7470 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7471 remove false predicated stores.
7473 2017-10-20 Richard Biener <rguenther@suse.de>
7475 * graphite-isl-ast-to-gimple.c
7476 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7477 Remove return value and simplify, dump copied stmt after lhs
7479 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7480 Reduce dump verbosity.
7481 (gsi_insert_earliest): Likewise.
7482 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7483 * graphite.c (print_global_statistics): Adjust dumping.
7484 (print_graphite_scop_statistics): Likewise.
7485 (print_graphite_statistics): Do not dump loops here.
7486 (graphite_transform_loops): But here.
7488 2017-10-20 Nicolas Roche <roche@adacore.com>
7490 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7491 * configure: Regenerate.
7493 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7496 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7497 functions when optimizing replace GIMPLE_RETURN stmts with
7498 calls to __builtin_unreachable ().
7501 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7502 for -fsanitize=thread link of executables.
7503 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7504 link of executables.
7507 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7509 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7510 define_insns for logical vector shifts to use VI248_AVX512BW
7511 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7512 condition, useless isa and prefix attributes. Change the first
7513 2 of these define_insns to ...
7514 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7515 define_insn for avx512vl.
7516 (<shift_insn><mode>3): ... and this, new define_insn without
7517 masking for non-avx512vl.
7520 * config/i386/sse.md (*andnot<mode>3,
7521 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7522 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7523 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7524 not applied use empty suffix even for TARGET_AVX512VL.
7525 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7526 is applied, supply evex,evex or evex,evex,evex instead of just
7529 2017-10-20 Julia Koval <julia.koval@intel.com>
7531 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7532 (OPTION_MASK_ISA_GFNI_UNSET): New.
7533 (ix86_handle_option): Handle OPT_mgfni.
7534 * config/i386/cpuid.h (bit_GFNI): New.
7535 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7536 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7537 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7538 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7539 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7540 * config/i386/i386.opt: Add mgfni.
7542 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7544 * config/msp430/msp430.c (msp430_option_override): Disable
7545 -fdelete-null-pointer-checks.
7546 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7548 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7550 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7551 of x87 and SSE instructions.
7553 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7555 * asan.c (create_cond_insert_point): Do not update edge count.
7556 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7557 (afdo_propagate_circuit): Likewise.
7558 (afdo_calculate_branch_prob): Likewise.
7559 (afdo_annotate_cfg): Likewise.
7560 * basic-block.h (struct edge_def): Remove count.
7561 (edge_def::count): New accessor.
7562 * bb-reorder.c (rotate_loop): Update.
7563 (find_traces_1_round): Update.
7564 (connect_traces): Update.
7565 (sanitize_hot_paths): Update.
7566 * cfg.c (unchecked_make_edge): Update.
7567 (make_single_succ_edge): Update.
7568 (check_bb_profile): Update.
7569 (dump_edge_info): Update.
7570 (update_bb_profile_for_threading): Update.
7571 (scale_bbs_frequencies_int): Update.
7572 (scale_bbs_frequencies_gcov_type): Update.
7573 (scale_bbs_frequencies_profile_count): Update.
7574 (scale_bbs_frequencies): Update.
7575 * cfganal.c (connect_infinite_loops_to_exit): Update.
7576 * cfgbuild.c (compute_outgoing_frequencies): Update.
7577 (find_many_sub_basic_blocks): Update.
7578 * cfgcleanup.c (try_forward_edges): Update.
7579 (try_crossjump_to_edge): Update
7580 * cfgexpand.c (expand_gimple_cond): Update
7581 (expand_gimple_tailcall): Update
7582 (construct_exit_block): Update
7583 * cfghooks.c (verify_flow_info): Update
7584 (redirect_edge_succ_nodup): Update
7585 (split_edge): Update
7586 (make_forwarder_block): Update
7587 (duplicate_block): Update
7588 (account_profile_record): Update
7589 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7590 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7591 * cfgloopmanip.c (scale_loop_profile): Update.
7593 (lv_adjust_loop_entry_edge): Update.
7594 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7595 (force_nonfallthru_and_redirect): Update.
7596 (purge_dead_edges): Update.
7597 (rtl_flow_call_edges_add): Update.
7598 * cgraphunit.c (init_lowered_empty_function): Update.
7599 (cgraph_node::expand_thunk): Update.
7600 * gimple-pretty-print.c (dump_probability): Update.
7601 (dump_edge_probability): Update.
7602 * gimple-ssa-isolate-paths.c (isolate_path): Update.
7603 * haifa-sched.c (sched_create_recovery_edges): Update.
7604 * hsa-gen.c (convert_switch_statements): Update.
7605 * ifcvt.c (dead_or_predicable): Update.
7606 * ipa-inline-transform.c (inline_transform): Update.
7607 * ipa-split.c (split_function): Update.
7608 * ipa-utils.c (ipa_merge_profiles): Update.
7609 * loop-doloop.c (add_test): Update.
7610 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
7611 * lto-streamer-in.c (input_cfg): Update.
7612 (input_function): Update.
7613 * lto-streamer-out.c (output_cfg): Update.
7614 * modulo-sched.c (sms_schedule): Update.
7615 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7616 * predict.c (maybe_hot_edge_p): Update.
7617 (unlikely_executed_edge_p): Update.
7618 (probably_never_executed_edge_p): Update.
7619 (dump_prediction): Update.
7620 (drop_profile): Update.
7621 (propagate_unlikely_bbs_forward): Update.
7622 (determine_unlikely_bbs): Update.
7623 (force_edge_cold): Update.
7624 * profile.c (compute_branch_probabilities): Update.
7625 * reg-stack.c (better_edge): Update.
7626 * shrink-wrap.c (handle_simple_exit): Update.
7627 * tracer.c (better_p): Update.
7628 * trans-mem.c (expand_transaction): Update.
7629 (split_bb_make_tm_edge): Update.
7630 * tree-call-cdce.c: Update.
7631 * tree-cfg.c (gimple_find_sub_bbs): Update.
7632 (gimple_split_edge): Update.
7633 (gimple_duplicate_sese_region): Update.
7634 (gimple_duplicate_sese_tail): Update.
7635 (gimple_flow_call_edges_add): Update.
7636 (insert_cond_bb): Update.
7637 (execute_fixup_cfg): Update.
7638 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
7639 * tree-complex.c (expand_complex_div_wide): Update.
7640 * tree-eh.c (lower_resx): Update.
7641 (unsplit_eh): Update.
7642 (cleanup_empty_eh_move_lp): Update.
7643 * tree-inline.c (copy_edges_for_bb): Update.
7644 (freqs_to_counts): Update.
7645 (copy_cfg_body): Update.
7646 * tree-ssa-dce.c (remove_dead_stmt): Update.
7647 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7648 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7649 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7650 (unloop_loops): Update.
7651 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7652 * tree-ssa-loop-split.c (connect_loops): Update.
7653 (split_loop): Update.
7654 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7655 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7656 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7657 * tree-ssa-reassoc.c (branch_fixup): Update.
7658 * tree-ssa-tail-merge.c (replace_block_by): Update.
7659 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7660 (compute_path_counts): Update.
7661 (update_profile): Update.
7662 (recompute_probabilities): Update.
7663 (update_joiner_offpath_counts): Update.
7664 (estimated_freqs_path): Update.
7665 (freqs_to_counts_path): Update.
7666 (clear_counts_path): Update.
7667 (ssa_fix_duplicate_block_edges): Update.
7668 (duplicate_thread_path): Update.
7669 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7670 (case_bit_test_cmp): Update.
7671 (collect_switch_conv_info): Update.
7672 (gen_inbound_check): Update.
7673 (do_jump_if_equal): Update.
7674 (emit_cmp_and_jump_insns): Update.
7675 * tree-tailcall.c (decrease_profile): Update.
7676 (eliminate_tail_call): Update.
7677 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7678 (vect_do_peeling): Update.
7679 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7680 * ubsan.c (ubsan_expand_null_ifn): Update.
7681 (ubsan_expand_ptr_ifn): Update.
7682 * value-prof.c (gimple_divmod_fixed_value): Update.
7683 (gimple_mod_pow2): Update.
7684 (gimple_mod_subtract): Update.
7685 (gimple_ic): Update.
7686 (gimple_stringop_fixed_value): Update.
7688 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7691 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
7693 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
7695 PR rtl-optimization/82395
7696 * ira-color.c (allocno_priority_compare_func): Fix comparison step
7697 based on non_spilled_static_chain_regno_p.
7699 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7701 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
7702 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
7703 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
7705 2017-10-19 Martin Sebor <msebor@redhat.com>
7707 PR tree-optimization/82596
7708 * tree.c (array_at_struct_end_p): Handle STRING_CST.
7710 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7712 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
7713 (get_mem_refs_of_builtin_call): Likewise.
7714 * builtins.c (expand_builtin_apply): Adjust call to
7715 allocate_dynamic_stack_space.
7716 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
7717 the third argument to allocate_dynamic_stack_space, otherwise -1.
7718 (expand_builtin): Deal with all alloca variants.
7719 (is_inexpensive_builtin): Likewise.
7720 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
7721 * calls.c (special_function_p): Deal with all alloca variants.
7722 (initialize_argument_information): Adjust call to
7723 allocate_dynamic_stack_space.
7724 (expand_call): Likewise.
7725 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
7726 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
7727 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
7728 use it for the stack usage computation.
7729 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
7730 * function.c (gimplify_parameters): Call build_alloca_call_expr.
7731 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
7732 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
7733 (in_loop_p): Remove first argument and useless check.
7734 (pass_walloca::execute): Remove useless test and adjust call to above.
7735 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
7736 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
7737 (gimplify_call_expr): Deal with all alloca variants.
7738 * hsa-gen.c (gen_hsa_alloca): Likewise.
7739 (gen_hsa_insns_for_call): Likewise.
7740 * ipa-pure-const.c (special_builtin_state): Likewise.
7741 * tree-chkp.c (chkp_build_returned_bound): Likewise.
7742 * tree-object-size.c (alloc_object_size): Likewise.
7743 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
7744 (call_may_clobber_ref_p_1): Likewise.
7745 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
7746 (ccp_fold_stmt): Likewise.
7747 (optimize_stack_restore): Likewise.
7748 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
7749 (mark_all_reaching_defs_necessary_1): Likewise.
7750 (propagate_necessity): Likewise.
7751 (eliminate_unnecessary_stmts): Likewise.
7752 * tree.c (build_common_builtin_nodes): Build
7753 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
7754 (build_alloca_call_expr): New function.
7755 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
7756 (CASE_BUILT_IN_ALLOCA): Likewise.
7757 (build_alloca_call_expr): Declare.
7758 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
7760 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7763 * dwarf2out.c (new_die_raw): New static inline function.
7764 (new_die): Use it to create the DIE.
7765 (add_AT_external_die_ref): Likewise.
7766 (clone_die): Likewise.
7767 (clone_as_declaration): Likewise.
7768 (dwarf2out_vms_debug_main_pointer): Likewise.
7769 (base_type_die): Likewise. Remove early return for corner cases.
7770 Do not call add_pubtype on the DIE here.
7771 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
7772 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
7773 typedefs for base types with DW_AT_endianity. Make sure a DIE with
7774 native order exists for base types, attach the DIE manually and call
7775 add_pubtype on it. Do not equate a reverse order DIE to the type.
7777 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
7779 * config/arm/arm.c (align_ok_ldrd_strd): New function.
7780 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
7782 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
7784 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7786 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
7787 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
7788 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
7789 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
7790 * opts.c (sanitizer_opts): Add builtin.
7791 * ubsan.c (instrument_builtin): New function.
7792 (pass_ubsan::execute): Call it.
7793 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
7794 * doc/invoke.texi: Document -fsanitize=builtin.
7796 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
7797 builtins, store max (log2 (align), 0) into uchar field instead of
7798 align into uptr field.
7799 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
7800 store uchar 0 field instead of uptr 0 field.
7801 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
7802 instead of passing one address of struct with 2 locations pass
7803 two addresses of structs with 1 location each.
7804 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
7805 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
7806 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
7807 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
7808 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
7809 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
7810 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
7811 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
7813 2017-10-19 Martin Liska <mliska@suse.cz>
7816 * file-find.c (remove_prefix): Remove.
7817 * file-find.h (remove_prefix): Likewise.
7818 * gcc-ar.c: Remove smartness of lookup.
7820 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
7822 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
7823 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
7824 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
7826 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7829 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
7830 (setcc + and to xor + setcc): New peephole2.
7832 2017-10-19 Tom de Vries <tom@codesourcery.com>
7834 * doc/sourcebuild.texi (Test Directives, Variants of
7835 dg-require-support): Add dg-require-stack-size.
7837 2017-10-19 Martin Liska <mliska@suse.cz>
7840 * gimplify.c (gimplify_decl_expr): Do not instrument variables
7841 that have a large alignment.
7842 (gimplify_target_expr): Likewise.
7844 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
7846 PR rtl-optimization/82602
7847 * ira.c (rtx_moveable_p): Return false for volatile asm.
7849 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
7852 * config/i386/i386-modes.def (CCGZ): New CC mode.
7853 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
7854 * config/i386/predicates.md (ix86_comparison_operator):
7856 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7857 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
7858 with double-word subtraction.
7859 (put_condition_code): Handle CCGZmode.
7861 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
7863 * wide-int.cc (debug (const wide_int &)): New.
7864 (debug (const wide_int *)): New.
7865 (debug (const widest_int &)): New.
7866 (debug (const widest_int *)): New.
7868 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
7871 * lra-constraints.c (curr_insn_transform): Use non-input operand
7872 instead of output one for matched reload.
7874 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7876 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
7877 (tree-ssa-loop-ivopts.h): New header file.
7878 (struct builtin_info): New fields.
7879 (classify_builtin_1): Compute and record base and offset parts for
7880 memset builtin partition by calling strip_offset.
7881 (offset_cmp, fuse_memset_builtins): New functions.
7882 (finalize_partitions): Fuse adjacent memset partitions by calling
7884 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
7885 Expose the interface.
7886 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
7888 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7890 PR tree-optimization/82574
7891 * tree-loop-distribution.c (find_single_drs): New parameter. Check
7892 that data reference must be executed exactly once per iteration
7893 against the outermost loop in nest.
7894 (classify_partition): Update call to above function.
7896 2017-10-18 Richard Biener <rguenther@suse.de>
7898 PR tree-optimization/82591
7899 * graphite.c (graphite_transform_loops): Move code gen message
7901 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7902 Here. Handle scop_to_isl_ast failing.
7903 (scop_to_isl_ast): Limit the number of ISL operations.
7905 2017-10-18 Richard Biener <rguenther@suse.de>
7907 * graphite-isl-ast-to-gimple.c
7908 (translate_isl_ast_to_gimple::set_rename): Simplify.
7909 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
7910 (graphite_copy_stmts_from_block): ... here.
7911 (copy_bb_and_scalar_dependences): Simplify.
7912 (add_parameters_to_ivs_params): Canonicalize.
7913 (generate_entry_out_of_ssa_copies): Simplify.
7914 * graphite-sese-to-poly.c (extract_affine_name): Simplify
7915 by passing in ISL dimension.
7916 (parameter_index_in_region_1): Rename to ...
7917 (parameter_index_in_region): ... this.
7918 (extract_affine): Adjust assert, pass down parameter index.
7919 (add_param_constraints): Use range-info when available.
7920 (build_scop_context): Adjust.
7921 * sese.c (new_sese_info): Adjust.
7922 (free_sese_info): Likewise.
7923 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
7924 Remove unused typedefs.
7925 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
7927 2017-10-18 Martin Liska <mliska@suse.cz>
7929 * combine.c (simplify_compare_const): Add gcc_fallthrough.
7931 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7933 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
7934 (s390_sched_init): Do not reset s390_sched_state if we entered the
7935 current basic block via a fallthru edge and all others are unlikely.
7937 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7939 * config/s390/s390.c (NUM_SIDES): New variable.
7940 (LONGRUNNING_THRESHOLD): New variable.
7941 (LATENCY_FACTOR): New variable.
7942 (s390_sched_score): Decrease score for long-running instructions on
7944 (s390_sched_variable_issue): Perform bookkeeping for long-running
7947 2017-10-18 Richard Biener <rguenther@suse.de>
7949 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
7950 Simplify with removal of the parameter rename map.
7951 (set_rename): Likewise.
7952 (should_copy_to_new_region): Likewise.
7953 (graphite_copy_stmts_from_block): Likewise.
7954 (copy_bb_and_scalar_dependences): Remove initialization of
7955 unused copied_bb_map.
7957 (copy_internal_parameters): Likewise.
7958 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
7959 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
7960 Use INTEGRAL_TYPE_P.
7961 (parameter_index_in_region_1): Rename to ...
7962 (assign_parameter_index_in_region): ... this. Assert we have
7963 a parameter we handle.
7964 (scan_tree_for_params): Adjust.
7965 * sese.h (parameter_rename_map_t): Remove.
7966 (struct sese_info_t): Remove unused parameter_rename_map and
7967 copied_bb_map members.
7968 * sese.c (new_sese_info): Adjust.
7969 (free_sese_info): Likewise.
7971 2017-10-18 Martin Liska <mliska@suse.cz>
7974 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
7975 on an abnormal edge.
7977 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
7979 * doc/invoke.texi (ffunction-sections and fdata-sections):
7982 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7984 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
7985 the use statement can throw internally.
7987 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7989 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
7990 any RTX present on the RHS of a SET.
7991 * compare-elim.c (try_eliminate_compare): Restore comment.
7993 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7995 * langhooks.h (struct lang_hooks): Document that tree_size langhook
7996 may be also called on tcc_type nodes.
7997 * langhooks.c (lhd_tree_size): Likewise.
7999 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8001 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8002 format_warning_at_substring.
8003 (maybe_warn): Convert source_range * param to a location_t. Pass
8004 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8005 (format_directive): Remove code to extract source_ranges and
8006 source_range * in favor of just a location_t.
8007 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8009 * substring-locations.c (format_warning_va): Convert
8010 source_range * param to a location_t.
8011 (format_warning_at_substring): Likewise.
8012 * substring-locations.h (format_warning_va): Likewise.
8013 (format_warning_at_substring): Likewise.
8015 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8017 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8019 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8020 and scatter/gather ops.
8022 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8023 vec_gather_load and vec_scatter_store.
8024 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8025 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8026 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8027 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8028 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8029 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8031 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8033 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8034 Detect FTST instruction and handle its register pops. Only pop
8035 second operand if can_pop_second_op is true.
8036 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8037 set can_pop_second_op to false in the compare_for_stack_reg call.
8039 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8040 output_fp_compare for stack register operands.
8041 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8042 instructions here. Do not emit stack register pops here. Assert
8043 that FCOMPP pops next to top stack register. Rewrite function.
8045 2017-10-17 Nathan Sidwell <nathan@acm.org>
8048 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8049 use DECL_ASSEMBLER_NAME_RAW.
8052 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8055 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8056 Wilco Dijkstra <wilco.dijkstra@arm.com>
8058 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8059 converstion to Pmode from the buf_addr.
8061 2017-10-17 Richard Biener <rguenther@suse.de>
8063 * graphite-dependences.c (scop_get_reads_and_writes): Change
8064 output parameters to references.
8066 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8068 PR 71026/tree-optimization
8069 * fold-const.c (distribute_real_division): Removed.
8070 (fold_binary_loc): Remove calls to distribute_real_divison.
8072 2017-10-17 Richard Biener <rguenther@suse.de>
8074 * graphite-scop-detection.c
8075 (scop_detection::stmt_has_simple_data_refs_p): Always use
8076 the full nest as region.
8077 (try_generate_gimple_bb): Likewise.
8078 * sese.c (scalar_evolution_in_region): Simplify now that
8079 SCEV can handle instantiation in regions.
8080 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8081 in the non-loop part of a function if requested.
8083 2017-10-17 Richard Biener <rguenther@suse.de>
8085 PR tree-optimization/82563
8086 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8088 (graphite_regenerate_ast_isl): Call it.
8089 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8091 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8093 PR tree-optimization/82549
8094 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8095 Formatting fixes. Instead of calling make_bit_field_ref with negative
8098 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8100 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8101 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8102 _mm_maskz_reduce_ss): New.
8103 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8104 __builtin_ia32_reducess_mask): Ditto..
8105 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8106 * config/i386/sse.md (reduces<mode>): Renamed to ...
8107 (reduces<mode><mask_scalar_name>): ... this.
8108 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8110 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8111 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8113 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8115 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8116 * selftest-run-tests.c (selftest::run_tests): Call
8117 selftest::unique_ptr_tests_cc_tests.
8118 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8119 * unique-ptr-tests.cc: New file.
8121 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8124 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8126 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8127 (make_hard_regno_born, make_hard_regno_dead): Update
8128 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8130 2017-10-16 Jeff Law <law@redhat.com>
8132 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8134 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8136 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8138 2017-10-16 Olivier Hainque <hainque@adacore.com>
8140 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8141 with_cpu if we were configured for an e500v2 target cpu name.
8143 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8145 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8146 * doc/invoke.texi: Document +nodsp as a valid extension for
8149 2017-10-16 Martin Liska <mliska@suse.cz>
8151 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8152 (test_set_range): Likewise.
8153 (test_range_functions): Rename to ...
8154 (test_bit_in_range): ... this.
8155 (sbitmap_c_tests): Add new test.
8157 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8159 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8161 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8162 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8164 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8166 * config/aarch64/aarch64-builtins.c
8167 (aarch64_types_quadopu_lane_qualifiers): New.
8168 (TYPES_QUADOPU_LANE): New.
8169 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8170 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8171 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8172 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8173 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8174 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8175 (Vdottype, DOTPROD): New.
8176 (sur): Add SDOT and UDOT.
8178 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8180 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8181 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8182 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8184 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8185 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8186 Enable TARGET_DOTPROD.
8187 (cortex-a75.cortex-a55): Likewise.
8188 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8190 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8192 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8193 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8195 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8197 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8198 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8199 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8200 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8201 * config/arm/types.md (neon_dot, neon_dot_q): New.
8202 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8204 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8206 * config/arm/arm.h (TARGET_DOTPROD): New.
8207 * config/arm/arm.c (arm_arch_dotprod): New.
8208 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8209 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8210 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8211 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8212 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8213 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8214 * doc/invoke.texi (armv8.2-a): Document dotprod
8216 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8218 * i386.c (ix86_vec_cost): New function.
8219 (ix86_rtx_costs): Handle vector operations better.
8220 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8221 * x86-tune-costs.h: Add new costs to all tables.
8223 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8225 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8227 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8228 divsd, sqrtss and sqrtsd
8229 * x86-tune-costs.h: Add new entries to all costs.
8230 (znver1_cost): Fix to match real instruction latencies.
8232 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8233 Michael Collison <michael.collison@arm.com>
8235 * compare-elim.c: Include emit-rtl.h.
8236 (can_merge_compare_into_arith): New function.
8237 (try_validate_parallel): Likewise.
8238 (try_merge_compare): Likewise.
8239 (try_eliminate_compare): Call the above when no previous clobber
8241 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8244 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8248 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8249 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8253 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8254 to be any operand_equal_p operands. For & (B - 1) require
8255 B to be power of 2. Recognize
8256 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8258 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8261 * optabs.c (expand_memory_blockage): Fix call of
8262 targetm.have_memory_blockage.
8264 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8267 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8268 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8269 objects to extra_objs instead of overwriting it.
8271 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8273 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8274 Use any_fp_register_operand as operand[3] predicate. Simplify
8275 equality test for operands[2] and operands[4] memory location.
8276 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8277 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8278 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8279 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8280 any_fp_register_operand as operand[1] predicate. Simplify
8281 equality test for operands[0] and operands[3] memory location.
8282 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8283 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8284 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8286 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8288 * target-insns.def: Add memory_blockage.
8289 * optabs.c (expand_memory_blockage): New function.
8290 (expand_asm_memory_barrier): Rename ...
8291 (expand_asm_memory_blockage): ... to this.
8292 (expand_mem_thread_fence): Call expand_memory_blockage
8293 instead of expand_asm_memory_barrier.
8294 (expand_mem_singnal_fence): Ditto.
8295 (expand_atomic_load): Ditto.
8296 (expand_atomic_store): Ditto.
8297 * doc/md.texi (Standard Pattern Names For Generation):
8298 Document memory_blockage instruction pattern.
8300 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8302 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8303 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8304 * config/rl78/rl78.md: New define_expand "adddi3".
8306 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8308 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8311 2017-10-13 Jeff Law <law@redhat.com>
8313 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8315 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8318 * internal-fn.c (expand_mul_overflow): If both operands have
8319 the same highpart of -1 or 0 and the topmost bit of lowpart
8320 is different, overflow is if res <= 0 rather than res < 0.
8322 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8324 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8325 TARGET_P9_VECTOR code for unaligned_load case.
8327 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8329 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8331 2017-10-13 Nathan Sidwell <nathan@acm.org>
8333 * tree-core.h (tree_contains_struct): Make bool.
8334 * tree.c (tree_contains_struct): Likewise.
8335 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8336 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8337 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8338 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8339 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8341 2017-10-13 Richard Biener <rguenther@suse.de>
8343 * graphite-isl-ast-to-gimple.c
8344 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8345 parameters and dominance check.
8346 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8347 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8348 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8349 Do not update SSA form here or do intermediate IL verification.
8350 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8351 (graphite_initialize): Remove check on the number of loops in
8352 the function and inline into graphite_transform_loops.
8353 (graphite_finalize): Inline into graphite_transform_loops.
8354 (graphite_transform_loops): Perform SSA update and IL verification
8356 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8358 2017-10-13 Richard Biener <rguenther@suse.de>
8360 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8361 graphite_expression_type_precision): Avoid global constructor
8363 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8364 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8365 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8366 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8367 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8368 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8370 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8373 * config/i386/i386.h (ix86_red_zone_size): New.
8374 * config/i386/i386.md (push peephole2s): Replace
8375 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8377 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8378 Alan Hayward <alan.hayward@arm.com>
8379 David Sherwood <david.sherwood@arm.com>
8381 * combine.c (can_change_dest_mode): Reject changes in
8382 REGMODE_NATURAL_SIZE.
8384 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8385 Alan Hayward <alan.hayward@arm.com>
8386 David Sherwood <david.sherwood@arm.com>
8388 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8389 (expand_debug_source_expr): Likewise.
8390 * combine.c (combine_simplify_rtx): Likewise.
8391 * cse.c (fold_rtx): Likewise.
8392 * fwprop.c (canonicalize_address): Likewise.
8393 * targhooks.c (default_shift_truncation_mask): Likewise.
8395 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8396 Alan Hayward <alan.hayward@arm.com>
8397 David Sherwood <david.sherwood@arm.com>
8399 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8400 (widened_mode): Likewise.
8401 (expand_unop): Likewise.
8402 * ree.c (transform_ifelse): Likewise.
8403 (merge_def_and_ext): Likewise.
8404 (combine_reaching_defs): Likewise.
8405 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8407 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8408 Alan Hayward <alan.hayward@arm.com>
8409 David Sherwood <david.sherwood@arm.com>
8411 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8412 * combine.c (gen_lowpart_for_combine): Likewise.
8413 * dwarf2out.c (rtl_for_decl_location): Likewise.
8414 * final.c (alter_subreg): Likewise.
8415 * rtlhooks.c (gen_lowpart_general): Likewise.
8416 (gen_lowpart_if_possible): Likewise.
8418 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8419 Alan Hayward <alan.hayward@arm.com>
8420 David Sherwood <david.sherwood@arm.com>
8422 * calls.c (expand_call): Use subreg_lowpart_offset.
8423 * cse.c (cse_insn): Likewise.
8424 * regcprop.c (copy_value): Likewise.
8425 (copyprop_hardreg_forward_1): Likewise.
8427 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8430 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8431 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8432 =Q constraints to +Q and into insn condition add check
8433 that operands[0] and operands[1] are equal.
8434 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8435 =Q constraints to +Q and into insn condition add check
8436 that operands[0] is equal to either operands[1] or operands[2].
8439 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8440 instead of handling MINUS_EXPR twice (once for each argument),
8441 canonicalize operand order and handle just once, use rtype where
8442 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8445 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8446 any values of __C while still being pattern recognizable as a simple
8449 2017-10-13 Richard Biener <rguenther@suse.de>
8451 PR tree-optimization/82451
8453 2017-10-02 Richard Biener <rguenther@suse.de>
8455 PR tree-optimization/82355
8456 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8457 a mapping for the enclosing loop but avoid generating one for
8459 (copy_bb_and_scalar_dependences): Remove premature codegen
8460 error on PHIs in blocks duplicated into multiple places.
8461 * graphite-scop-detection.c
8462 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8463 in the region use it as loop and nest to analyze the DR in.
8464 (try_generate_gimple_bb): Likewise.
8465 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8466 (add_loop_constraints): For blocks in a loop not in the region
8467 create a dimension with a single iteration.
8468 * sese.h (gbb_loop_at_index): Remove assert.
8470 * cfgloop.c (loop_preheader_edge): For the loop tree root
8471 return the single successor of the entry block.
8472 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8473 Reset the SCEV hashtable and niters.
8474 * graphite-scop-detection.c
8475 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8476 assert that we only have POLYNOMIAL_CHREC that vary in loops
8477 contained in the region.
8478 (scop_detection::graphite_can_represent_expr): Adjust.
8479 (scop_detection::stmt_has_simple_data_refs_p): For loops
8480 not in the region set loop to NULL. The nest is now the
8481 entry edge to the region.
8482 (try_generate_gimple_bb): Likewise.
8483 * sese.c (scalar_evolution_in_region): Adjust for
8484 instantiate_scev change.
8485 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8486 Make nest parameter the edge into the region.
8487 (create_data_ref): Likewise.
8488 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8489 entry edge into a region and adjust instantiate_scev calls.
8490 (create_data_ref): Likewise.
8491 (graphite_find_data_references_in_stmt): Likewise.
8492 (find_data_references_in_stmt): Pass the loop preheader edge
8493 from the nest argument.
8494 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8495 parameter the edge into the region.
8496 (instantiate_parameters): Use the loop preheader edge as entry.
8497 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8499 (get_instantiated_value_entry): Make instantiate_below parameter
8500 the edge into the region.
8501 (instantiate_scev_name): Likewise. Adjust dominance checks,
8502 when we cannot use loop-based instantiation instantiate by
8503 walking use-def chains.
8504 (instantiate_scev_poly): Adjust.
8505 (instantiate_scev_binary): Likewise.
8506 (instantiate_scev_convert): Likewise.
8507 (instantiate_scev_not): Likewise.
8508 (instantiate_array_ref): Remove.
8509 (instantiate_scev_3): Likewise.
8510 (instantiate_scev_2): Likewise.
8511 (instantiate_scev_1): Likewise.
8512 (instantiate_scev_r): Do not blindly handle N-operand trees.
8513 Do not instantiate array-refs. Handle all constants and invariants.
8514 (instantiate_scev): Make instantiate_below parameter
8515 the edge into the region.
8516 (resolve_mixers): Use the loop preheader edge for the region
8517 parameter to instantiate_scev_r.
8518 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8520 2017-10-13 Richard Biener <rguenther@suse.de>
8522 PR tree-optimization/82525
8523 * graphite-isl-ast-to-gimple.c
8524 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8526 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8527 Fail code generation when we cannot represent the isl integer.
8528 (binary_op_to_tree): Elide modulo operations that are no-ops
8529 in the type we code generate. Remove now superfluous code
8532 2017-10-13 Richard Biener <rguenther@suse.de>
8534 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8535 (scop_detection::harmful_loop_in_region): Remove premature
8536 IV type restriction.
8537 (scop_detection::graphite_can_represent_scev): We can handle
8538 pointer IVs just fine.
8540 2017-10-13 Alan Modra <amodra@gmail.com>
8542 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8543 "Clobbers and Scratch Registers". Add paragraph on
8544 alternative to clobbers for scratch registers and OpenBLAS
8547 2017-10-13 Alan Modra <amodra@gmail.com>
8549 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8550 example of a memory input for a string of known length. Move
8551 commentary out of table. Add a number of new examples
8552 covering array memory inputs.
8554 2017-10-12 Martin Liska <mliska@suse.cz>
8556 PR tree-optimization/82493
8557 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8558 (test_range_functions): New function.
8559 (sbitmap_c_tests): Likewise.
8560 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8561 * selftest.h (sbitmap_c_tests): New function.
8563 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8565 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8567 * config/rs6000/amo.h: Fix spacing issue.
8569 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8572 * config/i386/i386.md (*ashl<mode>3_mask_1,
8573 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8574 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8577 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8579 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8580 (profile_probability): Set max_probability
8581 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8583 * profile-count.c (profile_probability::differs_from_p): Do not
8584 rely on max_probaiblity == 10000
8586 2017-10-12 Jeff Law <law@redhat.com>
8588 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8591 2017-10-12 Martin Sebor <msebor@redhat.com>
8595 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
8596 (handle_alias_pairs): Call it.
8597 * common.opt (-Wattribute-alias): New option.
8598 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
8599 * doc/invoke.texi (-Wattribute-alias): Document.
8601 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
8604 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8606 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8608 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8609 (make_hard_regno_born, make_hard_regno_dead): Update
8610 bb_killed_pseudos and bb_gen_pseudos.
8612 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8614 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
8616 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
8618 * config/alpha/alpha.c (alpha_split_conditional_move):
8619 Use std::swap instead of manually swapping.
8620 (alpha_stdarg_optimize_hook): Ditto.
8621 (alpha_canonicalize_comparison): Ditto.
8623 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8625 * tree-loop-distribution.c (struct builtin_info): New struct.
8626 (struct partition): Refactor fields into struct builtin_info.
8627 (partition_free): Free struct builtin_info.
8628 (build_size_arg_loc, build_addr_arg_loc): Delete.
8629 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
8630 information from struct builtin_info.
8631 (find_single_drs): New function refactored from classify_partition.
8632 Also moved builtin validity checks to this function.
8633 (compute_access_range, alloc_builtin): New functions.
8634 (classify_builtin_st, classify_builtin_ldst): New functions.
8635 (classify_partition): Refactor code into functions find_single_drs,
8636 classify_builtin_st and classify_builtin_ldst.
8637 (distribute_loop): Don't do runtime alias check when distributing
8639 (find_seed_stmts_for_distribution): New function.
8640 (pass_loop_distribution::execute): Refactor code finding seed
8641 stmts into above function. Support distribution for the innermost
8642 two-level loop nest. Adjust dump information.
8644 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8646 * tree-loop-distribution.c: Adjust the general comment.
8647 (NUM_PARTITION_THRESHOLD): New macro.
8648 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
8649 (classify_partition): Skip builtin pattern of loop nest's inner loop.
8650 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
8651 in call to build_partition_graph.
8652 (finalize_partitions): New parameter. Make loop distribution more
8653 conservative by fusing more partitions.
8654 (distribute_loop): Don't do runtime alias check in case of loop nest
8656 (find_seed_stmts_for_distribution): New function.
8657 (prepare_perfect_loop_nest): New function.
8658 (pass_loop_distribution::execute): Refactor code finding seed stmts
8659 and loop nest into above functions. Support loop nest distribution.
8660 Adjust dump information accordingly.
8662 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8664 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
8665 and set PTYPE_SEQUENTIAL for merged partition.
8667 2017-10-12 Richard Biener <rguenther@suse.de>
8669 PR tree-optimization/69728
8671 2017-09-19 Richard Biener <rguenther@suse.de>
8673 PR tree-optimization/69728
8674 * graphite-sese-to-poly.c (schedule_error): New global.
8675 (add_loop_schedule): Handle empty domain by failing the
8677 (build_original_schedule): Handle schedule_error.
8679 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
8680 domain by returning an unchanged schedule.
8682 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8684 * genrecog.c (validate_pattern): For VEC_SELECT verify that
8685 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
8687 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
8689 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
8692 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8695 * expr.c (store_field): Don't optimize away bitsize == 0 store
8696 from CALL_EXPR with addressable return type.
8698 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8700 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
8701 * config/rs6000/rs6000.md (sel): Delete mode attribute.
8702 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
8703 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
8704 TARGET_ISEL instead of TARGET_ISEL<sel>.
8706 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
8708 * config/rs6000/rs6000.c
8709 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
8711 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8713 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
8715 (reg_or_cint_operand): Fix comment.
8716 (reg_or_zero_operand): New predicate.
8717 * config/rs6000/rs6000-protos.h (output_isel): Delete.
8718 * config/rs6000/rs6000.c (output_isel): Delete.
8719 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
8720 instead of reg_or_cint_operand. Output instruction directly (not via
8722 (isel_unsigned_<mode>): Ditto.
8723 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
8724 gpc_reg_operand. Add an instruction alternative for this. Output
8725 instruction directly.
8726 (*isel_reversed_unsigned_<mode>): Ditto.
8728 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8730 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
8731 (TARGET_CANONICALIZE_COMPARISON): Define.
8733 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
8736 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8737 Check whether the dest is REG before adding REG_EQUIV note.
8739 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8742 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8744 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8745 (make_hard_regno_born, make_hard_regno_dead): Update
8746 bb_killed_pseudos and bb_gen_pseudos.
8748 2017-10-11 Nathan Sidwell <nathan@acm.org>
8750 * incpath.h (enum incpath_kind): Name enum, prefix values.
8751 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
8752 * incpath.c (heads, tails): Use INC_MAX.
8753 (add_env_var_paths, add_standard_paths): Use incpath_kind.
8754 (merge_include_chains, split_quote_chain,
8755 register_include_chains): Update incpath_kind names.
8756 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
8757 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
8759 (add_framework_path, darwin_register_objc_includes): Likewise.
8760 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
8762 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8764 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
8765 Do not use float_operator operator predicate.
8766 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
8767 * config/i386/predicates.md (float_operator): Remove predicate.
8769 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8771 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
8772 (*jccxf_i387): Ditto.
8773 (*jcc<mode>_i387): Ditto.
8774 (*jccu<mode>_i387): Ditto.
8775 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
8776 (*jcc_*_i387 splitters): Remove.
8777 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
8778 * config/i386/i386.c (ix86_split_fp_branch): Remove.
8779 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
8782 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8784 * profile-count.h (slow_safe_scale_64bit): New function.
8785 (safe_scale_64bit): New inline.
8786 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
8787 * profile-count.c: Include wide-int.h
8788 (slow_safe_scale_64bit): New.
8790 2017-10-11 Nathan Sidwell <nathan@acm.org>
8792 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
8793 HAS_DECL_ASSEMBLER_NAME_P.
8794 * gimple-expr.c (gimple_decl_printable_name: Check
8795 HAS_DECL_ASSEMBLER_NAME_P too.
8796 * ipa-utils.h (type_in_anonymous_namespace_p): Check
8797 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
8798 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
8799 * passes.c (rest_of_decl_compilation): Check
8800 HAS_DECL_ASSEMBLER_NAME_P too.
8801 * recog.c (verify_changes): Likewise.
8802 * tree-pretty-print.c (dump_decl_name): Likewise.
8803 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
8805 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
8806 (SET_DECL_ASSEMBLER_NAME): Use it.
8807 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
8808 (COPY_DECL_ASSEMBLER_NAME): Likewise.
8809 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
8811 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8813 * config.gcc (i386, x86_64): Add extra objects.
8814 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
8815 (ix86_min_insn_size): Declare.
8816 (ix86_issue_rate): Declare.
8817 (ix86_adjust_cost): Declare.
8818 (ia32_multipass_dfa_lookahead): Declare.
8819 (ix86_macro_fusion_p): Declare.
8820 (ix86_macro_fusion_pair_p): Declare.
8821 (ix86_bd_has_dispatch): Declare.
8822 (ix86_bd_do_dispatch): Declare.
8823 (ix86_core2i7_init_hooks): Declare.
8824 (ix86_atom_sched_reorder): Declare.
8825 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
8826 (COSTS_N_BYTES): Move to x86-tune-costs.h.
8827 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
8828 (rip_relative_addr_p): Rename to ...
8829 (ix86_rip_relative_addr_p): ... this one; export.
8830 (memory_address_length): Update.
8831 (ix86_issue_rate): Move to x86-tune-sched.c.
8832 (ix86_flags_dependent): Move to x86-tune-sched.c.
8833 (ix86_agi_dependent): Move to x86-tune-sched.c.
8834 (exact_dependency_1): Move to x86-tune-sched.c.
8835 (exact_store_load_dependency): Move to x86-tune-sched.c.
8836 (ix86_adjust_cost): Move to x86-tune-sched.c.
8837 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
8838 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
8839 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
8840 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
8841 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
8842 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
8843 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
8844 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
8845 (min_insn_size): Rename to ...
8846 (ix86_min_insn_size): ... this one; export.
8847 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
8848 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
8849 (core2i7_first_cycle_multipass_backtrack): Move to
8850 x86-tune-sched-core.c.
8851 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
8852 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
8853 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
8854 (ix86_avoid_jump_mispredicts): Update.
8855 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
8856 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
8857 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
8858 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
8859 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
8860 (MAX_INSN): Move to ix86-tune-sched-bd.c.
8861 (MAX_IMM): Move to ix86-tune-sched-bd.c.
8862 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
8863 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
8864 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
8865 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
8866 (MAX_STORE): Move to ix86-tune-sched-bd.c.
8867 (BIG): Move to ix86-tune-sched-bd.c.
8868 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
8869 (enum insn_path): Move to ix86-tune-sched-bd.c.
8870 (get_mem_group): Move to ix86-tune-sched-bd.c.
8871 (is_cmp): Move to ix86-tune-sched-bd.c.
8872 (dispatch_violation): Move to ix86-tune-sched-bd.c.
8873 (is_branch): Move to ix86-tune-sched-bd.c.
8874 (is_prefetch): Move to ix86-tune-sched-bd.c.
8875 (init_window): Move to ix86-tune-sched-bd.c.
8876 (allocate_window): Move to ix86-tune-sched-bd.c.
8877 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
8878 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
8879 (process_end_window): Move to ix86-tune-sched-bd.c.
8880 (allocate_next_window): Move to ix86-tune-sched-bd.c.
8881 (find_constant): Move to ix86-tune-sched-bd.c.
8882 (get_num_immediates): Move to ix86-tune-sched-bd.c.
8883 (has_immediate): Move to ix86-tune-sched-bd.c.
8884 (get_insn_path): Move to ix86-tune-sched-bd.c.
8885 (get_insn_group): Move to ix86-tune-sched-bd.c.
8886 (count_num_restricted): Move to ix86-tune-sched-bd.c.
8887 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
8888 (add_insn_window): Move to ix86-tune-sched-bd.c.
8889 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
8890 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
8891 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
8892 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
8893 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
8894 (do_dispatch): Move to ix86-tune-sched-bd.c.
8895 (has_dispatch): Move to ix86-tune-sched-bd.c.
8896 * config/i386/t-i386: Add new object files.
8897 * config/i386/x86-tune-costs.h: New file.
8898 * config/i386/x86-tune-sched-atom.c: New file.
8899 * config/i386/x86-tune-sched-bd.c: New file.
8900 * config/i386/x86-tune-sched-core.c: New file.
8901 * config/i386/x86-tune-sched.c: New file.
8903 2017-10-11 Liu Hao <lh_mouse@126.com>
8905 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
8906 the generic version below instead.
8907 (should_colorize): Recognize Windows consoles as terminals
8909 * pretty-print.c [__MINGW32__] (write_all): New function.
8910 [__MINGW32__] (find_esc_head): Likewise.
8911 [__MINGW32__] (find_esc_terminator): Likewise.
8912 [__MINGW32__] (eat_esc_sequence): Likewise.
8913 [__MINGW32__] (mingw_ansi_fputs): New function that handles
8915 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
8918 2017-10-11 Richard Biener <rguenther@suse.de>
8920 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
8921 Properly call analyze_scalar_evolution with the loop of the stmt.
8923 2017-10-11 Richard Biener <rguenther@suse.de>
8925 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
8926 * tree-core.h (tree_base): Add chrec_var union member.
8927 * tree.h (CHREC_VAR): Remove.
8928 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
8929 * tree-chrec.h (build_polynomial_chrec): Adjust.
8930 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
8931 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
8933 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
8935 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
8936 * match.pd: ... here.
8937 ((T) X == (T) Y): Relax condition.
8939 2017-10-11 Bin Cheng <bin.cheng@arm.com>
8941 PR tree-optimization/82472
8942 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
8944 (break_alias_scc_partitions): Update postorder number.
8946 2017-10-11 Martin Liska <mliska@suse.cz>
8949 * opts.c (parse_no_sanitize_attribute): Do not use error_value
8951 * opts.h (parse_no_sanitize_attribute): Remove last argument.
8953 2017-10-11 Martin Liska <mliska@suse.cz>
8955 * print-rtl.c (print_insn): Move declaration of idbuf
8956 to same scope as name.
8958 2017-10-11 Martin Liska <mliska@suse.cz>
8963 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8966 2017-10-11 Martin Liska <mliska@suse.cz>
8969 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8972 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
8974 * config.gcc (default_gnu_indirect_function): Default to yes for
8975 arm*-*-linux* with glibc.
8977 2017-10-11 Richard Biener <rguenther@suse.de>
8979 * tree-scalar-evolution.c (get_scalar_evolution): Handle
8980 default-defs and types we do not want to analyze.
8981 (interpret_loop_phi): Replace unreachable code with an assert.
8982 (compute_scalar_evolution_in_loop): Remove and inline ...
8983 (analyze_scalar_evolution_1): ... here, replacing condition with
8984 what makes the intent clearer. Remove handling of cases
8985 get_scalar_evolution now handles.
8987 2017-10-10 Jim Wilson <wilson@tuliptree.org>
8989 PR rtl-optimization/81434
8990 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
8991 comment for main loop. In sched_group_found if, also add checks for
8992 pass and min_cost_group.
8994 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
8996 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
8997 (rs6000_insn_cost): New function.
8998 * config/rs6000/rs6000.md (cost): New attribute.
9000 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9001 H.J. Lu <hongjiu.lu@intel.com>
9005 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9006 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9007 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9008 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9009 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9010 OPTION_MASK_ISA_MMX and some other bit set require both
9011 mmx and the ISAs without the mmx bit.
9012 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9013 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9014 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9015 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9016 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9017 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9018 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9019 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9020 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9021 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9022 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9023 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9024 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9025 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9026 Add OPTION_MASK_ISA_MMX.
9028 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9030 * config.gcc (armv7*-*-freebsd*): New target.
9031 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9033 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9035 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9036 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9039 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9041 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9042 a template parameter.
9043 (WIDE_INT_REF_FOR): Update accordingly.
9044 * tree.h (wi::int_traits <const_tree>): Delete.
9045 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9046 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9047 (wi::tree_to_wide_ref): New typedef.
9048 (wi::to_wide): New function.
9049 * calls.c (get_size_range): Use wi::to_wide when operating on
9051 * cgraph.c (cgraph_node::create_thunk): Likewise.
9052 * config/i386/i386.c (ix86_data_alignment): Likewise.
9053 (ix86_local_alignment): Likewise.
9054 * dbxout.c (stabstr_O): Likewise.
9055 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9056 * expr.c (const_vector_from_tree): Likewise.
9057 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9058 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9059 (fold_negate_expr_1, int_const_binop_1, const_binop)
9060 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9061 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9062 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9063 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9064 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9065 (fold_not_const, round_up_loc): Likewise.
9066 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9067 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9068 (alloca_call_type): Likewise.
9069 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9070 * godump.c (go_output_typedef): Likewise.
9071 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9072 * internal-fn.c (get_min_precision): Likewise.
9073 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9074 * ipa-polymorphic-call.c
9075 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9076 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9077 (ipa_modify_call_arguments): Likewise.
9078 * match.pd: Likewise.
9079 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9080 * print-tree.c (print_node_brief, print_node): Likewise.
9081 * stmt.c (expand_case): Likewise.
9082 * stor-layout.c (layout_type): Likewise.
9083 * tree-affine.c (tree_to_aff_combination): Likewise.
9084 * tree-cfg.c (group_case_labels_stmt): Likewise.
9085 * tree-data-ref.c (dr_analyze_indices): Likewise.
9086 (prune_runtime_alias_test_list): Likewise.
9087 * tree-dump.c (dequeue_and_dump): Likewise.
9088 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9089 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9090 * tree-pretty-print.c (dump_generic_node): Likewise.
9091 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9092 (simple_iv_with_niters): Likewise.
9093 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9094 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9095 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9096 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9097 (refine_value_range_using_guard, number_of_iterations_ne_max)
9098 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9099 (get_cst_init_from_scev, record_nonwrapping_iv)
9100 (scev_var_range_cant_overflow): Likewise.
9101 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9102 * tree-ssa-pre.c (compute_avail): Likewise.
9103 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9104 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9105 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9106 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9107 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9108 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9109 (dump_case_nodes, try_switch_expansion): Likewise.
9110 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9111 (vect_do_peeling): Likewise.
9112 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9113 * tree-vect-stmts.c (vectorizable_load): Likewise.
9114 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9115 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9116 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9117 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9118 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9119 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9120 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9121 (evrp_dom_walker::before_dom_children): Likewise.
9122 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9123 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9124 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9125 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9126 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9127 (get_range_pos_neg): Likewise.
9128 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9129 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9130 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9131 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9132 * config/avr/avr.c (avr_fold_builtin): Likewise.
9133 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9134 * config/msp430/msp430.c (msp430_attr): Likewise.
9135 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9136 * config/powerpcspe/powerpcspe-c.c
9137 (altivec_resolve_overloaded_builtin): Likewise.
9138 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9139 (rs6000_expand_ternop_builtin): Likewise.
9140 * config/rs6000/rs6000-c.c
9141 (altivec_resolve_overloaded_builtin): Likewise.
9142 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9143 (rs6000_expand_ternop_builtin): Likewise.
9144 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9146 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9148 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9149 when copying loop nest with only one inner loop.
9151 2017-10-10 Richard Biener <rguenther@suse.de>
9153 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9154 blocks if SCEV is active.
9155 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9157 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9158 (scev_initialize): Assert we are not yet initialized.
9160 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9162 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9163 inner loop's exit stmt by making it always exit the loop, otherwise
9164 we would generate an infinite empty loop.
9166 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9168 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9169 renaming variables in new preheader if it's deleted.
9171 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9173 * tree-loop-distribution.c (struct partition): Remove unused field
9174 loops of the structure.
9175 (partition_alloc, partition_free): Ditto.
9176 (build_rdg_partition_for_vertex): Ditto.
9178 2017-10-09 Jeff Law <law@redhat.com>
9180 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9181 return type to match prototype and documentation.
9183 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9185 * config/rs6000/rs6000.c (processor_costs): Move to ...
9186 * config/rs6000/rs6000.h: ... here.
9187 (rs6000_cost): Declare.
9189 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9191 * except.c (setjmp_fn): New global variable.
9192 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9193 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9194 if DONT_USE_BUILTIN_SETJMP is defined.
9196 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9198 * target.def (insn_cost): New hook.
9199 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9200 * doc/tm.texi: Regenerate.
9201 * rtlanal.c (insn_cost): Use the new hook.
9203 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9205 * combine.c (combine_validate_cost): Compute the new insn_cost,
9206 not just pattern_cost.
9207 (try_combine): Adjust comment.
9209 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9211 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9213 * combine.c (uid_insn_cost): Adjust comment.
9214 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9216 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9217 * dse.c (find_shift_sequence): Ditto.
9218 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9219 (bb_valid_for_noce_process_p): Use pattern_cost.
9220 * rtl.h (insn_rtx_cost): Delete.
9221 (pattern_cost): New prototype.
9222 (insn_cost): New prototype.
9223 * rtlanal.c (insn_rtx_cost): Rename to...
9224 (pattern_cost): ... this.
9227 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9229 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9230 (*jcc<mode>_0_r_i387): Ditto.
9231 (*jccxf_r_i387): Ditto.
9232 (*jcc<mode>_r_i387): Ditto.
9233 (*jccu<mode>_r_i387): Ditto.
9234 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9235 (*jcc): Rename from *jcc_1.
9237 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9239 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9240 deferred rescans after the lvx/stvx recombination pre-pass.
9242 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9244 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9245 memory operation instruction support.
9246 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9247 (rs6000-ibm-aix[789]*): Likewise.
9248 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9249 Document new functions.
9251 2017-10-09 Richard Biener <rguenther@suse.de>
9253 PR tree-optimization/82397
9254 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9255 equality only for semantically equal trees.
9257 2017-10-09 Richard Biener <rguenther@suse.de>
9259 PR tree-optimization/82449
9260 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9261 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9262 allow constant addresses.
9263 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9266 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9268 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9271 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9274 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9277 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9280 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9282 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9285 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9286 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9288 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9290 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9291 (WI_BINARY_PREDICATE_RESULT): Likewise.
9292 (wi::binary_traits::operator_result): New type.
9293 (wi::binary_traits::predicate_result): Likewise.
9294 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9295 (generic_wide_int::operator==, generic_wide_int::operator!=)
9296 (generic_wide_int::operator&, generic_wide_int::and_not)
9297 (generic_wide_int::operator|, generic_wide_int::or_not)
9298 (generic_wide_int::operator^, generic_wide_int::operator+
9299 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9301 (operator~, unary operator-, operator==, operator!=, operator&)
9302 (operator|, operator^, operator+, binary operator-, operator*): New
9304 * expr.c (get_inner_reference): Use wi::bit_and_not.
9305 * fold-const.c (fold_binary_loc): Likewise.
9306 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9307 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9308 (bit_value_binop): Likewise.
9309 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9310 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9311 (extract_range_from_binary_expr_1): Likewise.
9312 (masked_increment): Likewise.
9313 (simplify_bit_ops_using_ranges): Likewise.
9315 2017-10-09 Martin Jambor <mjambor@suse.cz>
9318 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9319 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9320 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9322 (hsa_fixup_mov_insn_type): New function.
9323 (hsa_op_with_type::get_in_type): Use it.
9324 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9325 immediates in an assert.
9326 (hsa_op_with_type::extend_int_to_32bit): New method.
9327 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9328 types. Convert to dest type if necessary.
9329 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9330 (reg_for_gimple_ssa): Pass false as min32int to
9331 hsa_type_for_scalar_tree_type.
9332 (gen_hsa_addr): Fixup type when creating addresable temporary.
9333 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9334 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9335 necessary. Call hsa_fixup_mov_insn_type.
9336 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9337 extend operands and convert to dest type if necessary.
9338 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9339 to dest type if necessary.
9340 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9341 if conversion nt necessary and size matches.
9342 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9343 to dest type if necessary.
9344 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9345 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9347 (gen_hsa_clrsb): Likewise.
9348 (gen_hsa_ffs): Likewise.
9349 (gen_hsa_divmod): Extend operands and convert to dest type if
9351 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9353 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9355 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9356 Remove empty default arguments. Use a brace block as output
9358 (conditional return): Ditto.
9360 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9361 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9363 (group_ending_nop): Ditto.
9364 (doloop_end): Ditto.
9365 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9366 (splitters for those): Ditto.
9368 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9370 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9371 a conditional jump (and the compare for it) so that pc_rtx is the
9373 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9374 for the deleted and renamed ctr<mode>_internal[234] patterns.
9375 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9376 Delete second conditional return pattern.
9377 (ctr<mode>_internal2): Delete this second bdnz pattern.
9378 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9379 (ctr<mode>_internal4): Delete this second bdz pattern.
9381 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9383 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9384 (always_initialized_rtx_for_ssa_name_p): New predicate.
9385 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9386 (finish_out_of_ssa): Free new field of SA.
9387 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9388 * tree-ssa-coalesce.c: Include tree-ssa.h.
9389 (get_parm_default_def_partitions): Remove extern keyword.
9390 (get_undefined_value_partitions): New function.
9391 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9392 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9394 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9396 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9398 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9400 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9402 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9403 for targets that preffer 128bit.
9405 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9407 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9409 2017-10-08 Olivier Hainque <hainque@adacore.com>
9411 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9412 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9414 (thumb_set_return_address): Likewise.
9416 2017-10-08 Olivier Hainque <hainque@adacore.com>
9418 * common/config/arm/arm-common.c (arm_except_unwind_info):
9419 Handle DWARF2_UNWIND_INFO.
9421 2017-10-07 Michael Collison <michael.collison@arm.com>
9423 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9426 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9428 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9429 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9430 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9431 defined, force the creation of a new block for a dispatch label.
9433 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9435 * invoke.texi (Wsuggest-attribute=cold): Document.
9436 * common.opt (Wsuggest-attribute=cold): New
9437 * ipa-pure-const.c (warn_function_cold): New function.
9438 * predict.c (compute_function_frequency): Use it.
9439 * predict.h (warn_function_cold): Declare.
9441 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9443 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9446 2017-10-06 Martin Liska <mliska@suse.cz>
9448 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9449 keyword for member functions.
9450 (struct sanopt_tree_couple): New struct.
9451 (struct sanopt_tree_couple_hash): New function.
9452 (struct sanopt_ctx): Add new hash_map.
9453 (has_dominating_ubsan_ptr_check): New function.
9454 (record_ubsan_ptr_check_stmt): Likewise.
9455 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9456 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9457 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9459 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9462 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9463 aarch64_simd_valid_immediate on CONST_VECTORs.
9464 (aarch64_reg_or_bic_imm): Likewise.
9466 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9468 PR rtl-optimization/82396
9469 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9471 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9473 * graphite-dependences.c (scop_get_reads): Move code to...
9474 (scop_get_must_writes): Move code to...
9475 (scop_get_may_writes): Move code to...
9476 (scop_get_reads_and_writes): ... here.
9477 (scop_get_dependences): Call scop_get_reads_and_writes.
9479 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9481 PR tree-optimization/82434
9482 * fold-const.h (can_native_encode_type_p,
9483 can_native_encode_string_p): Remove.
9484 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9485 don't encode anything, just return what would be otherwise returned.
9486 (native_encode_fixed, native_encode_complex, native_encode_vector):
9488 (native_encode_string): Likewise. Inline by hand
9489 can_native_encode_string_p.
9490 (can_native_encode_type_p): Remove.
9491 (can_native_encode_string_p): Remove.
9492 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9493 STRING_CSTs using can_native_encode_string_p, test all
9494 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9495 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9496 argument from native_encode_expr.
9497 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9498 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9501 2017-10-06 Richard Biener <rguenther@suse.de>
9503 PR tree-optimization/82397
9504 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9505 operand_equal_p but rely on data_ref_compare_tree for detecting
9507 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9508 to match up with dr_group_sort_cmp.
9510 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9513 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9515 * config/s390/s390-builtin-types.def: Regenerate.
9517 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9520 * config/s390/s390-builtin-types.def: Regenerate.
9521 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9522 Change flag from B_VXE to B_VX.
9523 (s390_vec_min_dbl): Remove B_VXE flag.
9525 2017-10-06 Richard Biener <rguenther@suse.de>
9527 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9528 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9529 translate_isl_ast_to_gimple::is_valid_rename,
9530 translate_isl_ast_to_gimple::get_rename,
9531 translate_isl_ast_to_gimple::get_def_bb_for_const,
9532 translate_isl_ast_to_gimple::get_new_name,
9533 translate_isl_ast_to_gimple::collect_all_ssa_names,
9534 translate_isl_ast_to_gimple::copy_loop_phi_args,
9535 translate_isl_ast_to_gimple::collect_all_ssa_names,
9536 translate_isl_ast_to_gimple::copy_loop_phi_args,
9537 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9538 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9539 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9540 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9541 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9542 translate_isl_ast_to_gimple::copy_cond_phi_args,
9543 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9544 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9545 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9546 translate_isl_ast_to_gimple::rename_uses,
9547 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9548 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9549 (set_rename_for_each_def): Likewise.
9550 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9551 here. Handle rewriting SCEV analyzable uses here.
9552 (copy_bb_and_scalar_dependences): Generate code for PHI
9554 (graphite_regenerate_ast_isl): Adjust.
9555 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9556 (add_write, add_read): New functions.
9557 (build_cross_bb_scalars_def): Use it and simplify.
9558 (build_cross_bb_scalars_use): Likewise.
9559 (graphite_find_cross_bb_scalar_vars): Inline into...
9560 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9561 simulating out-of-SSA. Compute liveout and add dependencies.
9562 (build_scops): Force an empty entry block.
9563 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9565 (sese_build_liveouts): Declare.
9566 (sese_trivially_empty_bb_p): Likewise.
9567 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9568 compute liveout and debug_liveout.
9569 (sese_bad_liveouts_use): Remove.
9570 (sese_reset_debug_liveouts_bb): Likewise.
9571 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9572 (sese_build_liveouts): Build liveout and debug_liveout and store
9574 (new_sese_info): Adjust.
9575 (free_sese_info): Likewise.
9576 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9577 do not build liveout here.
9578 (move_sese_in_condition): Adjust region entry.
9579 (scev_analyzable_p): Match up with chrec_apply requirements.
9580 (sese_trivially_empty_bb_p): New.
9581 * tree-into-ssa.c (get_reaching_def): Properly support generating
9582 default-defs for incremental rewrite of anonymous names.
9584 2017-10-06 Richard Biener <rguenther@suse.de>
9586 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9587 precision do not perform modulo reduction.
9589 2017-10-06 Richard Biener <rguenther@suse.de>
9591 PR tree-optimization/82436
9592 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9593 conservatively choose the vectorization factor when checking
9594 whether we can perform the required load permutation.
9595 (vect_transform_slp_perm_load): Assert when we may not fail.
9597 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
9599 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
9600 message for incompatible -msdata=* and -mcall-* options.
9602 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9604 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
9605 rate for post-reload scheduling.
9607 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9609 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
9611 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9613 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
9614 to improve monte carlo in scimark.
9616 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9618 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
9619 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
9620 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
9621 pentium4_cost, nocona_cost): Set reassociation width to 1.
9622 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
9623 width to 2 for fp operations and 1 otherwise.
9624 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
9626 (atom_cost): Set reassociation width to 2.
9627 (slm_cost, generic_cost): Set fp reassociation width
9628 to 2 and 1 otherwise.
9629 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
9630 (core_cost): Set fp reassociation width to 4 and vector to 2.
9631 (ix86_reassociation_width): Rewrite using cost table; special case
9632 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
9633 and TARGET_AVX128_OPTIMAL.
9634 * config/i386/i386.h (processor_costs): Add
9635 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
9636 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
9637 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
9638 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
9639 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
9640 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
9642 2017-10-05 Nathan Sidwell <nathan@acm.org>
9644 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
9646 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9648 * config/arm/arm.c (arm_test_fpu_data): New.
9649 (arm_run_selftests): Call arm_test_fpu_data.
9651 2017-10-04 Nathan Sidwell <nathan@acm.org>
9653 * toplev.c (toplev::main): Remove excess parens on pretty_printer
9655 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
9657 2017-10-04 Sudakshina Das <sudi.das@arm.com>
9659 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
9660 check type for aarch64_simd_valid_immediate.
9661 (aarch64_output_simd_mov_immediate): Update prototype.
9662 (aarch64_simd_valid_immediate): Update prototype.
9663 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
9664 support for ORR-immediate.
9665 (and<mode>3): modified pattern to add support for BIC-immediate.
9666 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
9667 now checks for valid immediate for BIC and ORR based on new enum
9669 (aarch64_output_simd_mov_immediate): Function now used to output
9670 BIC/ORR imm as well based on new enum argument.
9671 * config/aarch64/constraints.md (Do): New vector immediate constraint.
9673 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
9674 (aarch64_reg_or_bic_imm): Likewise.
9676 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9678 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
9679 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
9681 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9685 PR rtl-optimization/82396
9686 * haifa-sched.c (autopref_multipass_init): Simplify
9688 (autopref_rank_data): Simplify sort order.
9689 * sched-int.h (autopref_multipass_data_): Remove
9690 multi_mem_insn_p, min_offset and max_offset.
9692 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9694 * doc/sourcebuild.texi: Document vect_peeling_profitable.
9696 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9698 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
9701 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9703 * doc/sourcebuild.texi: Document vect_long_mult.
9705 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
9707 PR tree-optimization/82413
9708 * fold-const.c (build_range_check): Use widest_int when comparing
9709 the maximum ETYPE value with HIGH.
9711 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9713 PR rtl-optimization/82396
9714 * haifa-sched.c (autopref_multipass_init): Simplify
9716 (autopref_rank_data): Simplify sort order.
9717 * sched-int.h (autopref_multipass_data_): Remove
9718 multi_mem_insn_p, min_offset and max_offset.
9720 2017-10-04 Jakub Jelinek <jakub@redhat.com>
9722 PR tree-optimization/82381
9723 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
9724 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
9727 PR tree-optimization/82374
9728 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
9729 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
9730 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
9731 current_function_decl to the new decl.
9733 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9735 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
9736 helper macro for IEEE float128 hardware built-in functions.
9737 (SQRTF128_ODD): Add built-in functions with the round-to-odd
9739 (TRUNCF128_ODD): Likewise.
9740 (ADDF128_ODD): Likewise.
9741 (SUBF128_ODD): Likewise.
9742 (MULF128_ODD): Likewise.
9743 (DIVF128_ODD): Likewise.
9744 (FMAF128_ODD): Likewise.
9745 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
9746 UNSPEC_TRUNC_ROUND_TO_ODD.
9747 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
9748 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
9749 floating point round to odd instructions.
9750 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
9751 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
9752 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
9753 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
9754 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
9755 (trunc<mode>sf2_hw): Change the truncate with round to odd
9756 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
9757 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
9758 to odd hardware instructions.
9759 (sub<mode>3_odd): Likewise.
9760 (mul<mode>3_odd): Likewise.
9761 (div<mode>3_odd): Likewise.
9762 (sqrt<mode>2_odd): Likewise.
9763 (fma<mode>4_odd): Likewise.
9764 (fms<mode>4_odd): Likewise.
9765 (nfma<mode>4_odd): Likewise.
9766 (nfms<mode>4_odd): Likewise.
9767 (trunc<mode>df2_odd): Change the truncate with round to odd
9768 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
9770 * doc/extend.texi (PowerPC built-in functions): Update documentation
9771 for existing IEEE float128-bit built-in functions. Add built-in
9772 functions that generate the IEEE 128-bit floating point round to
9775 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
9777 PR rtl-optimization/77729
9778 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
9779 to (X&(C1&~C2))|C2 transformations.
9781 2017-10-03 Martin Jambor <mjambor@suse.cz>
9783 PR tree-optimization/82363
9784 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
9785 mismatch, mark lacc written regardless of racc.
9787 2017-10-03 Jakub Jelinek <jakub@redhat.com>
9789 PR tree-optimization/82381
9790 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
9791 stmt_to_insert nor wheather SSA_NAMEs are default defs.
9792 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
9793 fallthrough into reassoc_stmt_dominates_stmt_p.
9796 * combine.c (combine_instructions): Don't combine in unreachable
9799 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9802 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
9803 function to not use the have_cpu variable. Do not set cpu_index,
9804 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
9806 (rs6000_valid_attribute_p): Remove duplicate initializations of
9807 old_optimize and func_optimize.
9808 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
9809 (rs6000_activate_target_options): Make global.
9810 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
9813 2017-10-02 Jakub Jelinek <jakub@redhat.com>
9815 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
9816 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
9817 Set *poffset to 0 and *psize and *pmax_size to -1 if
9818 *poffset + *psize overflows in HOST_WIDE_INT.
9820 PR tree-optimization/82387
9821 PR tree-optimization/82388
9822 PR tree-optimization/82389
9823 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
9824 instead of live_bytes non-NULL.
9826 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
9829 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
9832 2017-10-02 Richard Biener <rguenther@suse.de>
9834 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
9835 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
9836 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
9838 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9840 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
9841 requested precision matches the type's.
9842 * calls.c (alloc_max_size): Calculate the new candidate size as
9843 a widest_int and use wi::to_widest when comparing it with the
9844 current candidate size.
9845 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
9846 zero rather than integer_zero_node.
9847 * match.pd: Check for a no-op conversion before using wi::add
9848 rather than after. Use tree_to_uhwi when summing small shift
9849 counts into an unsigned int.
9851 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9852 Alan Hayward <alan.hayward@arm.com>
9853 David Sherwood <david.sherwood@arm.com>
9856 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
9857 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
9858 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
9859 POINTER_AND_FP_REGS.
9861 2017-10-02 Richard Biener <rguenther@suse.de>
9863 PR tree-optimization/82355
9864 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9865 a mapping for the enclosing loop but avoid generating one for
9867 (copy_bb_and_scalar_dependences): Remove premature codegen
9868 error on PHIs in blocks duplicated into multiple places.
9869 * graphite-scop-detection.c
9870 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9871 in the region use it as loop and nest to analyze the DR in.
9872 (try_generate_gimple_bb): Likewise.
9873 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9874 (add_loop_constraints): For blocks in a loop not in the region
9875 create a dimension with a single iteration.
9876 * sese.h (gbb_loop_at_index): Remove assert.
9878 2017-10-01 Kevin Buettner <kevinb@redhat.com>
9880 * omp-expand.c (adjust_context_scope): New function.
9881 (expand_parallel_call): Call adjust_context_scope.
9883 2017-10-01 Jeff Law <law@redhat.com>
9885 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
9886 dom_opt_dom_walker class with direct access to private members.
9887 Add comments. Call test_for_singularity.
9888 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
9889 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
9890 m_dummy_cond anymore.
9891 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
9893 (pass_dominator:execute): Build the dummy_cond here and pass it
9894 to the dom_opt_dom_walker ctor.
9895 (test_for_singularity): New function.
9897 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
9898 Maya Rashish <coypu@sdf.org>
9900 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
9901 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
9902 (alpha*-*-netbsd*) Use nbsd_tm_file.
9903 (arm*-*-netbsdelf*) Likewise.
9904 (i[34567]86-*-netbsdelf*) Likewise.
9905 (x86_64-*-netbsd*) Likewise.
9906 (mips*-*-netbsd*) Likewise.
9907 (powerpc-*-netbsd*) Likewise.
9908 (sh*-*-netbsd*) Likewise.
9909 (sparc-*-netbsdelf*) Likewise.
9910 (sparc64-*-netbsd*) Likewise.
9911 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
9913 (vax-*-netbsdelf*) Likewise.
9914 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
9915 (UINT_FAST8_TYPE) Likewise.
9916 (INT_FAST16_TYPE) Check CHAR_FAST16.
9917 (UINT_FAST16_TYPE) Likewise.
9919 2017-09-30 Jakub Jelinek <jakub@redhat.com>
9922 * config/i386/i386.md
9923 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
9924 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
9925 *divmodsi4_zext_2): New define_insn_and_split.
9926 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
9927 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
9928 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
9929 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
9930 New define_insn_and_split.
9931 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
9932 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
9933 operands[1] having DImode when mode is SImode.
9935 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
9936 always SImode for DIV and MOD in REG_EQUAL notes.
9938 2017-09-29 Yury Gribov <tetra2005@gmail.com>
9941 * match.pd: Fix handling of NaNs in pattern.
9943 2017-09-29 Jeff Law <law@redhat.com>
9945 * sbitmap.c (bitmap_bit_in_range_p): New function.
9946 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
9947 * tree-ssa-dse.c (live_bytes_read): New function.
9948 (dse_classify_store): Ignore reads of dead bytes.
9950 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
9951 typos and whitespace errors.
9952 * config/i386/predicates.md (address_no_seg_operand): Likewise.
9953 * config/s390/s390.c (s390_emit_prologue): Likewise.
9955 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9958 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
9959 with a symbol for LRA.
9961 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9963 PR rtl-optimization/82338
9964 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
9966 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
9968 * genmodes.c (calc_wider_mode): Suppress qsort macro.
9969 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
9970 (qsort_chk): Declare.
9971 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
9972 (qsort_chk): New function.
9974 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9976 PR tree-optimization/82337
9977 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9978 phi definition if the PHI result appears in an abnormal PHI.
9979 (find_basis_for_base_expr): Don't record a basis if the LHS of the
9980 basis appears in an abnormal PHI.
9982 2017-09-29 Richard Biener <rguenther@suse.de>
9984 * graphite-isl-ast-to-gimple.c
9985 (translate_isl_ast_to_gimple::set_codegen_error): New function.
9986 (binary_op_to_tree): Use it.
9987 (get_rename_from_scev): Likewise.
9988 (copy_loop_phi_nodes): Likewise.
9989 (copy_bb_and_scalar_dependences): Likewise.
9990 (translate_pending_phi_nodes): Likewise.
9992 2017-09-29 Jakub Jelinek <jakub@redhat.com>
9995 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
9996 for movabsq $(i32 << shift), r64.
9998 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10000 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10001 index when encoding %esp as %rsp to avoid 0x67 prefix.
10003 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10005 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10006 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10008 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10010 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10011 Extensions with more than 16 double VFP registers.
10012 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10013 to_clear_mask and all code related to it. Replace the remaining
10014 entry by a sbitmap and adapt code accordingly.
10016 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10018 * brig-builtins.def: Change pure attributes to const.
10020 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10022 * config.gcc (default_gnu_indirect_function): Default to yes for
10023 sparc*-*-linux* with glibc.
10025 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10027 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10028 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10029 when creating .init_array and .fini_array sections with priority
10032 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10035 * config/aarch64/aarch64.c
10036 (aarch64_builtin_support_vector_misalignment): Always return false
10037 when misalignment is unknown.
10039 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10041 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10042 this function to return false if the definition used by the swap
10043 instruction is artificial, or if the memory address from which the
10044 constant value is loaded is not represented by a base address held
10045 in a register or if the base address register is a frame or stack
10046 pointer. Additionally, return false if the base address of the
10047 loaded constant is a SYMBOL_REF but is not considered to be a
10049 (replace_swapped_load_constant): New function.
10050 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10051 loaded constant vector with a load of a swapped constant vector.
10053 2017-09-27 Carl Love <cel@us.ibm.com>
10055 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10056 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10057 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10060 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10062 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10063 first, always call autopref_rank_data otherwise.
10065 2017-09-27 Richard Biener <rguenther@suse.de>
10067 * graphite-scop-detection.c (find_scop_parameters): Move
10068 loop bound handling ...
10069 (gather_bbs::before_dom_children): ... here, avoiding the need
10070 to build scop_info->loop_nest.
10071 (record_loop_in_sese): Remove.
10072 * sese.h (sese_info_t::loop_nest): Remove.
10073 * sese.c (new_sese_info): Do not allocate loop_nest.
10074 (free_sese_info): Do not free loop_nest.
10076 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10079 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10080 lhs from calls if the lhs has addressable type.
10082 2017-09-27 Richard Biener <rguenther@suse.de>
10084 * graphite.h (scop::max_alias_set): New member.
10085 * graphite-scop-detection.c: Remove references to non-existing
10086 --param in comments.
10087 (build_alias_sets): Record the maximum alias set used for drs.
10088 (build_scops): Support zero as unlimited for
10089 --param graphite-max-arrays-per-scop.
10090 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10091 and inline into ...
10092 (build_poly_sr_1): ... here. Compute alias set based on the
10093 maximum alias set used for drs rather than
10094 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10096 2017-09-27 Richard Biener <rguenther@suse.de>
10098 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10099 --param loop-block-tile-size=0 to disable tiling.
10101 2017-09-27 Richard Biener <rguenther@suse.de>
10103 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10104 (graphite-max-nb-scop-params): Document special value zero.
10105 * domwalk.h (dom_walker::STOP): New symbolical constant.
10106 (dom_walker::dom_walker): Add optional parameter for bb to
10108 (dom_walker::~dom_walker): Declare.
10109 (dom_walker::before_dom_children): Document STOP return value.
10110 (dom_walker::m_user_bb_to_rpo): New member.
10111 (dom_walker::m_bb_to_rpo): Likewise.
10112 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10113 mapping here if not provided by the user.
10114 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10115 provided by the user.
10116 (dom_walker::STOP): Define.
10117 (dom_walker::walk): Do not compute bb to RPO mapping here.
10118 Support STOP return value from before_dom_children to stop
10120 * graphite-optimize-isl.c (optimize_isl): If the schedule
10121 is the same still generate code if -fgraphite-identity
10122 or -floop-parallelize-all are given.
10123 * graphite-scop-detection.c: Include cfganal.h.
10124 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10126 (gather_bbs::before_dom_children): Return STOP for BBs
10128 (build_scops): Compute bb to RPO mapping and pass it to
10129 the domwalk. Treat --param graphite-max-nb-scop-params=0
10130 as not limiting the number of params.
10131 * graphite.c (graphite_initialize): Remove limit on the
10132 number of basic-blocks in a function.
10133 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10134 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10135 default value of 10.
10137 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10139 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10140 Adjust code to eliminate needing to do the shift right 32-bits
10141 operation after XSCVDPSPN.
10143 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10145 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10146 ((X / Y) != 0 -> X >= Y): Likewise.
10148 2017-09-26 Carl Love <cel@us.ibm.com>
10150 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10151 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10152 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10153 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10154 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10155 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10156 definitions and overloading.
10157 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10158 statement for P9V_BUILTIN_XST_LEN_R.
10159 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10160 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10161 define_expand and define_insn for the instructions and builtins.
10162 * doc/extend.texi: Update the built-in documentation file for the new
10163 built-in functions.
10164 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10165 define_insn for the instructions
10167 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10170 * gcc/config/netbsd-protos.h: New file.
10171 * gcc/config/netbsd.c: New file.
10172 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10173 * gcc/config/t-netbsd: New file.
10174 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10175 (tmake_file) Add t-netbsd.
10176 (extra_objs) Add netbsd.o.
10178 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10182 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10184 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10186 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10187 sign extension from a vector register to a GPR by doing a 32-bit
10188 direct move and then an EXTSW.
10189 (extendsi<mode>2 splitter): Likewise.
10190 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10191 right or vector extract after doing XSCVDPSPN. Use
10192 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10194 (movdi_from_sf_zero_ext): Likewise.
10195 (reload_gpr_from_vsxsf): Likewise.
10196 (p8_mfvsrd_4_disf): Delete, no longer used.
10197 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10198 then needing to move the SFmode to a GPR to use the XSCVDPSP
10199 instruction instead of FRSP and XSCVDPSPN.
10200 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10201 it is adjacent to the other XSCVSPDP insns.
10202 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10203 SFmode to be in traditional Altivec registers.
10204 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10205 (vsx_xscvspdpn): Likewise.
10206 (vsx_xscvspdpn_scalar): Likewise.
10208 2017-09-26 Martin Jambor <mjambor@suse.cz>
10210 * tree-sra.c (compare_access_positions): Put integral types first,
10211 stabilize sorting of integral types, remove conditions putting
10212 non-full-precision integers last.
10213 (sort_and_splice_var_accesses): Disable scalarization if a
10214 non-integert would be represented by a non-full-precision integer.
10216 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10218 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10219 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10220 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10221 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10222 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10223 conditionals inside the function instead of around it. Call
10224 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10225 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10227 2017-09-26 Richard Biener <rguenther@suse.de>
10229 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10231 (scop_detection::build_scop_breadth): ... this. Removed.
10232 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10233 (scop_detection::harmful_stmt_in_bb): Likewise.
10234 (scop_detection::graphite_can_represent_stmt): Likewise.
10235 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10236 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10237 (scop_detection::can_represent_loop_1): ... this. Removed.
10238 (scop_detection::harmful_loop_in_region): Simplify after inlining
10239 the above and remove more quadraticness.
10240 (build_scops): Adjust.
10241 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10244 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10247 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10248 REGNO (base) == SP_REG if base is a REG.
10250 PR middle-end/35691
10251 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10252 if it is different SSA_NAME.
10253 (optimize_range_tests_cmp_bitwise): New function.
10254 (optimize_range_tests): Call it.
10256 2017-09-26 Richard Biener <rguenther@suse.de>
10258 PR tree-optimization/82321
10259 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10260 for the def being inside the loop.
10262 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10264 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10266 * config/s390/s390-builtins.def: Fix constraint on op4.
10268 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10270 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10271 independent expanders.
10272 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10273 ("vec_ordered", "vec_unordered"): New expanders.
10275 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10277 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10280 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10282 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10283 vec_unpacks_lo_v16qi.
10284 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10286 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10288 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10289 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10290 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10292 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10294 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10296 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10298 ("*vec_slb<mode>"): New insn pattern.
10299 ("vec_shr_<mode>"): New expander.
10300 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10301 and force the shift count operand to V16QImode.
10302 ("vec_srb<mode>"): Set shift count mode to V16QI.
10304 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10306 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10307 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10308 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10310 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10313 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10315 2017-09-26 Richard Biener <rguenther@suse.de>
10317 PR tree-optimization/82320
10318 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10321 2017-09-25 Jeff Law <law@redhat.com>
10323 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10324 prototype for new argument.
10325 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10326 mostly extracted from rs6000_emit_allocate_stack.
10327 (rs6000_emit_probe_stack_range_stack_clash): New function.
10328 (rs6000_emit_allocate_stack): Call
10329 rs6000_emit_probe_stack_range_stack_clash as needed.
10330 (rs6000_emit_probe_stack_range): Add additional argument
10331 to call to gen_probe_stack_range{si,di}.
10332 (output_probe_stack_range): New.
10333 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10334 (output_probe_stack_range_stack_clash): New.
10335 (rs6000_emit_prologue): Emit notes into dump file as requested.
10336 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10337 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10338 Add additional operand and pass it to output_probe_stack_range.
10340 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10342 PR tree-optimization/82163
10343 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10344 (checking_verify_loop_closed_ssa): New parameter.
10345 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10346 (check_loop_closed_ssa_stmt): Delete.
10347 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10348 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10349 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10352 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10354 * brig-builtins.def: Treat HSAIL barrier builtins as
10355 setjmp/longjump style functions.
10357 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10359 * target.def (constant_alignment): New hook.
10360 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10361 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10362 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10363 * doc/tm.texi: Regenerate.
10364 * targhooks.h (default_constant_alignment): Declare.
10365 (constant_alignment_word_strings): Likewise.
10366 * targhooks.c (default_constant_alignment): New function.
10367 (constant_alignment_word_strings): Likewise.
10368 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10369 instead of CONSTANT_ALIGNMENT.
10370 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10371 (force_const_mem): Likewise.
10372 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10373 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10374 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10375 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10376 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10378 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10379 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10380 constant_alignment_word_strings.
10381 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10382 (CONSTANT_ALIGNMENT): Likewise.
10383 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10384 (arm_constant_alignment): New function.
10385 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10386 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10387 constant_alignment_word_strings.
10388 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10389 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10390 constant_alignment_word_strings.
10391 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10392 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10393 (cris_constant_alignment): New function.
10394 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10395 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10396 (epiphany_constant_alignment): New function.
10397 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10398 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10399 constant_alignment_word_strings.
10400 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10401 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10402 constant_alignment_word_strings.
10403 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10404 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10405 constant_alignment_word_strings.
10406 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10407 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10408 * config/i386/i386.c (ix86_constant_alignment): Make static.
10409 Use the same interface as the target hook.
10410 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10411 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10412 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10413 constant_alignment_word_strings.
10414 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10415 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10416 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10417 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10418 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10419 constant_alignment_word_strings.
10420 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10421 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10422 constant_alignment_word_strings.
10423 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10424 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10425 constant_alignment_word_strings.
10426 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10427 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10429 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10430 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10431 * config/mips/mips.c (mips_constant_alignment): New function.
10432 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10433 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10434 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10435 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10436 (mmix_constant_alignment): Make static. Use the same interface
10437 as the target hook.
10438 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10439 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10440 constant_alignment_word_strings.
10441 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10442 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10443 constant_alignment_word_strings.
10444 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10445 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10446 constant_alignment_word_strings.
10447 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10448 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10449 (rs6000_constant_alignment): New function.
10450 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10451 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10452 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10453 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10454 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10455 (rs6000_constant_alignment): New function.
10456 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10457 * config/s390/s390.c (s390_constant_alignment): New function.
10458 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10459 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10460 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10461 constant_alignment_word_strings.
10462 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10463 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10464 (sparc_constant_alignment): New function.
10465 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10466 * config/spu/spu.c (spu_constant_alignment): New function.
10467 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10468 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10469 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10470 constant_alignment_word_strings.
10471 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10472 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10473 constant_alignment_word_strings.
10474 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10475 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10476 constant_alignment_word_strings.
10477 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10478 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10479 (visium_constant_alignment): New function.
10480 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10481 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10482 (xtensa_constant_alignment): New function.
10483 * system.h (CONSTANT_ALIGNMENT): Poison.
10485 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10487 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10488 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10489 (rs6000_builtin_valid_without_lhs): New helper function.
10490 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10491 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10493 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10495 * target.h (vec_perm_indices): Use unsigned short rather than
10497 (auto_vec_perm_indices): Likewise.
10498 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10499 Use unsigned int rather than unsigned char.
10500 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10502 2017-09-25 Richard Biener <rguenther@suse.de>
10504 * cfgloop.h (sort_sibling_loops): Declare.
10505 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10506 (sort_sibling_loops): New function sorting the sibling loop list
10508 * graphite.c (graphite_transform_loops): Sort sibling loops.
10510 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10512 * target.def (vec_perm_const_ok): Change sel parameter to
10514 * optabs-query.c (can_vec_perm_p): Update accordingly.
10515 * doc/tm.texi: Regenerate.
10516 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10517 auto_vec_perm_indices and remove separate nelt field.
10518 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10519 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10520 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10521 (aarch64_expand_vec_perm_const): Update accordingly.
10522 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10523 to vec_perm_indices.
10524 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10525 auto_vec_perm_indices and remove separate nelt field.
10526 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10527 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10528 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10530 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10531 to vec_perm_indices.
10532 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10533 sel to vec_perm_indices.
10534 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10535 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10536 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10538 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10541 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10544 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10545 on the FUNCTION_DECL function context if it has a DIE that is a
10548 2017-09-25 Richard Biener <rguenther@suse.de>
10550 PR tree-optimization/82285
10551 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10554 2017-09-25 Tom de Vries <tom@codesourcery.com>
10558 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10561 2017-09-25 Richard Biener <rguenther@suse.de>
10563 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10564 ISL errors other than isl_error_quota happen. Dump if the
10565 schedule is the same.
10566 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10567 errors instead of aborting inside ISL.
10569 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10572 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10573 of libgcc_eh for m64.
10574 * config/i386/darwin64.h: Likewise.
10576 2017-09-25 Richard Biener <rguenther@suse.de>
10578 PR middle-end/82144
10579 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10580 attribute for incomplete types nor twice for complete ones.
10582 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10585 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10586 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10589 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10592 * config/i386/i386.opt (mprefer-avx256): Use
10593 ix86_target_flags variable.
10594 * config/i386/i386.c (ix86_target_string): Move
10595 -mprefer-avx256 to flag2_opts.
10597 2017-09-22 Jakub Jelinek <jakub@redhat.com>
10599 PR middle-end/35691
10600 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
10601 and x != -1 | y != -1 into (x & y) != -1.
10603 2017-09-22 Steve Ellcey <sellcey@cavium.com>
10605 * config.gcc: Add new case statement to set
10606 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
10607 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
10608 s390x-*-linux* case statements. Added aarch64 to the list of
10609 supported architectures.
10611 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10613 PR tree-optimization/82289
10614 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
10615 STMT_VINFO_RELEVANT_P.
10617 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10618 Alan Hayward <alan.hayward@arm.com>
10619 David Sherwood <david.sherwood@arm.com>
10621 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
10622 for VR_RANGE only; don't allow VR_ANTI_RANGE.
10623 (extract_range_from_binary_expr_1): Don't call
10624 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
10626 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10627 Alan Hayward <alan.hayward@arm.com>
10628 David Sherwood <david.sherwood@arm.com>
10630 * target.def (preferred_vector_alignment): New hook.
10631 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
10633 * doc/tm.texi: Regenerate.
10634 * targhooks.h (default_preferred_vector_alignment): Declare.
10635 * targhooks.c (default_preferred_vector_alignment): New function.
10636 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
10638 (DR_TARGET_ALIGNMENT): New macro.
10639 (aligned_access_p): Update commentary.
10640 (vect_known_alignment_in_bytes): New function.
10641 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
10643 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
10644 Calculate the misalignment based on the target alignment rather than
10646 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
10647 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
10648 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
10649 the target alignment, rather than masking the element misalignment
10650 with the number of elements in a vector. Also use the target
10651 alignment when calculating the maximum number of peels.
10652 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
10653 instead of TYPE_ALIGN_UNIT.
10654 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
10655 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
10656 (vect_create_addr_base_for_vector_ref): Update call accordingly.
10657 (vect_create_data_ref_ptr): Likewise.
10658 (vect_setup_realignment): Realign by ANDing with
10659 -DR_TARGET_MISALIGNMENT.
10660 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
10661 the number of peels based on DR_TARGET_ALIGNMENT.
10662 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
10663 with the guaranteed alignment boundary when deciding whether
10665 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
10666 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
10667 (ensure_base_align): Remove stmt_info parameter. Get the
10668 target base alignment from DR_TARGET_ALIGNMENT.
10669 (vectorizable_store): Update call accordingly. Interpret
10670 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
10672 (vectorizable_load): Likewise.
10674 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10675 Alan Hayward <alan.hayward@arm.com>
10676 David Sherwood <david.sherwood@arm.com>
10678 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
10679 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
10680 (vect_enhance_data_refs_alignment): Likewise.
10682 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10684 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10685 error. Only quit immediately if parsing is complete.
10686 (BEGIN): Initialize fatal_err and parse_done.
10687 (begin fpu, end fpu): Check number of arguments.
10688 (begin arch, end arch): Likewise.
10689 (begin cpu, end cpu): Likewise.
10690 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10691 (optalias): Likewise.
10693 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10695 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10696 * config/arm/arm-isa.h: Delete. Move definitions to ...
10697 * arm-cpus.in: ... here. Use new feature and fgroup values.
10698 * config/arm/arm.c (arm_option_override): Use lower case for feature
10700 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10701 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10702 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10704 (print_isa_bits_for): New function.
10705 (gen_isa): New function.
10706 (gen_comm_data): Use print_isa_bits_for.
10707 (define feature): New keyword.
10708 (define fgroup): New keyword.
10709 * config/arm/t-arm (TM_H): Remove.
10710 (GTM_H): Add arm-isa.h.
10711 (arm-isa.h): Add rule to generate file.
10712 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10713 case for feature bit names.
10715 2017-09-22 Richard Biener <rguenther@suse.de>
10717 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
10719 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
10720 print of no dependency loops ...
10721 * graphite.c (graphite_transform_loops): ... here.
10722 (canonicalize_loop_closed_ssa_form): Work from inner to outer
10724 (same_close_phi_node, remove_duplicate_close_phi,
10725 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
10726 (canonicalize_loop_closed_ssa): ... here and simplify.
10727 * graphite-optimize-isl.c: Include tree-vectorizer.h.
10728 (optimize_isl): Use dump_printf_loc to tell when we stopped
10729 optimizing because of an ISL timeout.
10731 2017-09-22 Richard Biener <rguenther@suse.de>
10733 PR tree-optimization/82291
10734 * tree-if-conv.c (predicate_mem_writes): Make sure to
10735 remove writes in blocks predicated with false.
10737 2017-09-22 Richard Biener <rguenther@suse.de>
10739 * sese.c: Include cfganal.h.
10740 (if_region_set_false_region): Remove.
10741 (create_if_region_on_edge): Likewise.
10742 (move_sese_in_condition): Re-implement without destroying
10745 2017-09-22 Richard Biener <rguenther@suse.de>
10747 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
10748 Verify both BBs contain loop PHI nodes before dispatching to
10749 copy_loop_phi_args.
10750 (graphite_regenerate_ast_isl): Do not recompute dominators,
10751 do not verify three times. Restructure for clarity.
10752 * graphite-scop-detection.c (same_close_phi_node,
10753 remove_duplicate_close_phi, make_close_phi_nodes_unique,
10754 defined_in_loop_p, canonicalize_loop_closed_ssa,
10755 canonicalize_loop_closed_ssa_form): Simplify, remove excess
10756 checking and SSA rewrite, move to ...
10757 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
10758 (graphite_initialize): Do not pass in ctx, do not reset the
10759 SCEV cache, compute only dominators.
10760 (graphite_transform_loops): Allocate ISL ctx after
10761 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
10762 Maintain post-dominators only around build_scops.
10763 * sese.c (if_region_set_false_region): Make static. Free
10764 and recompute dominators.
10765 (move_sese_in_condition): Assert we don't get called with
10766 post-dominators computed.
10767 * sese.h (if_region_set_false_region): Remove.
10769 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
10771 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
10772 mode attribute for TARGET_AVX512VL.
10774 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
10776 * config/i386/i386.opt (mprefer-avx256): New option.
10777 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
10779 (ix86_preferred_simd_mode): Return 256-bit AVX modes
10780 for TARGET_PREFER_AVX256.
10781 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
10783 2017-09-21 Jeff Law <law@redhat.com>
10785 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
10786 Fix dump output if the only stack space is for pushed registers.
10788 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10790 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
10793 2017-09-21 Martin Sebor <msebor@redhat.com>
10796 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
10797 code (in C++) or code that triggers warnings.
10799 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
10801 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
10803 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
10805 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
10806 * sched-rgn.c: Ditto.
10807 * sel-sched-ir.c: Ditto.
10809 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
10811 * toplev.h (set_random_seed): Adjust return type.
10812 * toplev.c (init_local_tick): Move eager initialization of random_seed
10813 to get_random_seed. Adjust comment.
10814 (init_random_seed): Inline to get_random_seed, delete.
10815 (get_random_seed): Initialize random_seed lazily.
10816 (set_random_seed): Do not return previous value.
10817 (print_switch_value): Do not call get_random_seed.
10819 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
10821 * cgraph.c (delete_function_version): New, broken out from...
10822 (cgraph_node::delete_function_version): ...here. Rename to
10823 cgraph_node::delete_function_version_by_decl. Update all uses.
10824 (cgraph_node::remove): Call delete_function_version.
10826 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10829 * tree-inline.c (expand_call_inline): Emit clobber stmts for
10830 VAR_DECLs to which addressable non-volatile parameters are mapped
10831 and for id->retvar after the return value assignment. Clear
10832 id->retval and id->retbnd after inlining.
10834 2017-09-21 Richard Biener <rguenther@suse.de>
10836 PR tree-optimization/82276
10837 PR tree-optimization/82244
10838 * tree-vrp.c (build_assert_expr_for): Set
10839 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
10841 (remove_range_assertions): Revert earlier change.
10843 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
10846 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
10848 2017-09-21 Richard Biener <rguenther@suse.de>
10850 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10851 Restore valid IL after code generation errors.
10852 * graphite.c (graphite_transform_loops): Diagnose code
10853 generation issues as MSG_MISSED_OPTIMIZATION and continue
10854 with processing SCOPs.
10856 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10857 Alan Hayward <alan.hayward@arm.com>
10858 David Sherwood <david.sherwood@arm.com>
10860 * calls.c (compute_argument_addresses): Use simplify_gen_binary
10861 rather than choosing between plus_constant and gen_rtx_<CODE>.
10862 * expr.c (emit_push_insn): Likewise.
10863 (expand_expr_real_2): Likewise.
10865 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10866 Alan Hayward <alan.hayward@arm.com>
10867 David Sherwood <david.sherwood@arm.com>
10869 * loop-unroll.c (split_iv): Call copy_rtx on the step.
10871 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10872 Alan Hayward <alan.hayward@arm.com>
10873 David Sherwood <david.sherwood@arm.com>
10875 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
10876 calling tree_to_uhwi.
10878 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10879 Alan Hayward <alan.hayward@arm.com>
10880 David Sherwood <david.sherwood@arm.com>
10882 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
10883 INTEGER_CST rather than a negative test for ADDR_EXPR.
10885 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10886 Alan Hayward <alan.hayward@arm.com>
10887 David Sherwood <david.sherwood@arm.com>
10889 * tree-vrp.c (extract_range_from_binary_expr_1): Check
10890 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
10892 2017-09-21 Richard Biener <rguenther@suse.de>
10894 PR tree-optimization/71351
10895 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
10896 graphite_create_new_loop_guard): Remove, fold remaining parts
10898 (translate_isl_ast_node_for): ... here and simplify.
10900 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10903 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
10904 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
10905 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
10906 alternative always use QI mode, for -Os imov (=R,R) alternative
10907 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
10910 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10911 Jeff Law <law@redhat.com>
10913 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
10914 (allocate_stack_space): New function, partially extracted from
10915 s390_emit_prologue.
10916 (s390_emit_prologue): Track offset to most recent stack probe.
10917 Code to allocate space moved into allocate_stack_space.
10918 Dump actions when no stack is allocated.
10919 (s390_prologue_plus_offset): New function.
10920 (s390_emit_stack_probe): Likewise.
10922 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
10924 * common.opt (Wa, Wl, Wp, g, gz=): Add
10926 (gno-column-info): Remove.
10927 (gcolumn-info): Drop RejectNegative.
10928 (gno-): New prefix.
10929 (gno-record-gcc-switches): Remove.
10930 (grecord-gcc-switches): Drop RejectNegative.
10931 (gno-split-dwarf): Remove.
10932 (gsplit-dwarf): Drop RejectNegative.
10933 (gno-strict-dwarf): Remove.
10934 (gstrict-dwarf): Drop RejectNegative.
10935 * config/darwin.opt (gfull, gused): Add RejectNegative.
10936 * dwarf2out.c (gen_producer_string): Drop
10937 gno-record-gcc-switches handler.
10938 * optc-gen.awk: Add g to prefixes with negative forms.
10939 * opts-common.c (remapping_prefix_p): New.
10940 (find_opt): Check it.
10941 (generate_canonical_option): Test g prefix.
10942 (option_map): Add -gno- mapping.
10943 (add_misspelling_candidates): Check remapping_prefix_p.
10945 2017-09-20 Jeff Law <law@redhat.com>
10947 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
10948 thinko in stack clash protection support.
10950 * explow.c (compute_stack_clash_protection_loop_data): Use
10951 CONST_INT_P instead of explicit test. Verify object is a
10952 CONST_INT_P before looking at INTVAL.
10953 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
10954 instead of explicit test.
10956 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
10959 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
10960 address instead of to r1 and r11.
10962 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
10964 * config.gcc: Support "knm".
10965 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
10966 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10968 * config/i386/i386.c (m_KNM): Define.
10969 (processor_target_table): Add "knm".
10971 (ix86_option_override_internal): Add "knm".
10972 (ix86_issue_rate): Add PROCESSOR_KNM.
10973 (ix86_adjust_cost): Ditto.
10974 (ia32_multipass_dfa_lookahead): Ditto.
10975 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
10976 (fold_builtin_cpu): Add M_INTEL_KNM.
10977 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
10978 (processor_type): Add PROCESSOR_KNM.
10979 * config/i386/x86-tune.def: Add m_KNM.
10980 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
10982 2017-09-20 Richard Biener <rguenther@suse.de>
10984 PR tree-optimization/80213
10985 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
10986 are allowed in empty BBs as well.
10987 (canonicalize_loop_closed_ssa): Also look for other complex
10989 (scop_detection::get_sese): Include the loop-closed PHI block
10991 (scop_detection::merge_sese): Remove code adding extra blocks.
10992 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
10993 (build_scops): Assert the final returned scop is invalid.
10995 2017-09-20 Richard Biener <rguenther@suse.de>
10997 PR tree-optimization/82264
10998 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11000 (vn_phi_lookup): Likewise.
11001 (vn_phi_insert): Likewise.
11003 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11005 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11006 that fits into uhwi or shwi, add DW_AT_const_value regardless
11007 of early_dwarf without going through RTL, using add_AT_unsigned
11010 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11011 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11012 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11013 (DEBUG_MACRO_SECTION): Likewise.
11014 (DEBUG_LTO_MACRO_SECTION): Likewise.
11015 (DEBUG_STR_DWO_SECTION): Likewise.
11016 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11017 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11018 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11019 (DEBUG_LTO_LINE_STR_SECTION): Define.
11020 (init_sections_and_labels): Initialize debug_line_str_section
11021 variable. Initialize debug_loc_section for -gdwarf-5 to
11022 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11024 2017-09-20 Richard Biener <rguenther@suse.de>
11026 * graphite-sese-to-poly.c (extract_affine): Properly handle
11027 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11029 2017-09-20 Richard Biener <rguenther@suse.de>
11031 PR tree-optimization/81373
11032 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11033 Force SESE live-out defs to be handled even if they are
11036 2017-09-19 Jeff Law <law@redhat.com>
11038 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11039 nothing for stack adjustments with REG_STACK_CHECK.
11040 * sched-deps.c (parse_add_or_inc): Reject insns with
11041 REG_STACK_CHECK from dependency breaking.
11042 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11043 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11044 * reg-notes.def (STACK_CHECK): New note.
11046 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11047 (ix86_expand_prologue): Dump stack clash info as needed.
11048 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11050 * function.c (dump_stack_clash_frame_info): New function.
11051 * function.h (dump_stack_clash_frame_info): Prototype.
11052 (enum stack_clash_probes): New enum.
11054 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11055 flag_stack_clash_protection.
11056 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11057 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11058 (arm_frame_pointer_required): Likewise.
11059 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11060 (ia64_expand_prologue): Likewise.
11061 * config/mips/mips.c (mips_expand_prologue): Likewise.
11062 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11063 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11064 (sparc_flat_expand_prologue): Likewise.
11065 * config/spu/spu.c (spu_expand_prologue): Likewise.
11067 * explow.c: Include "params.h".
11068 (anti_adjust_stack_and_probe_stack_clash): New function.
11069 (get_stack_check_protect): Likewise.
11070 (compute_stack_clash_protection_loop_data): Likewise.
11071 (emit_stack_clash_protection_loop_start): Likewise.
11072 (emit_stack_clash_protection_loop_end): Likewise.
11073 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11074 Use anti_adjust_stack_and_probe_stack_clash.
11075 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11076 (emit_stack_clash_protection_loop_start): Likewise.
11077 (emit_stack_clash_protection_loop_end): Likewise.
11078 * rtl.h (get_stack_check_protect): Prototype.
11079 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11080 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11081 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11083 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11085 * doc/tm.texi: Rebuilt.
11086 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11087 get_stack_check_protect.
11088 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11089 * config/arm/arm.c (arm_expand_prologue): Likewise.
11090 (arm_frame_pointer_required): Likewise.
11091 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11092 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11093 * config/mips/mips.c (mips_expand_prologue): Likewise.
11094 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11095 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11096 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11097 (sparc_flat_expand_prologue): Likewise.
11099 * common.opt (-fstack-clash-protection): New option.
11100 * flag-types.h (enum stack_check_type): Note difference between
11101 -fstack-check= and -fstack-clash-protection.
11102 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11103 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11104 * toplev.c (process_options): Issue warnings/errors for cases
11105 not handled with -fstack-clash-protection.
11106 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11107 (-fstack-check): Note additional problem with -fstack-check=generic.
11108 Note that -fstack-check is primarily for Ada and refer users
11109 to -fstack-clash-protection for stack-clash-protection.
11110 Document new params for stack clash protection.
11112 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11114 * config/i386/i386.c (ix86_split_long_move): Do not handle
11115 address used for LEA in a special way.
11117 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11119 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11121 2017-09-19 Martin Sebor <msebor@redhat.com>
11124 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11125 of incompatible types.
11127 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11129 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11130 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11131 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11132 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11134 2017-09-19 Richard Biener <rguenther@suse.de>
11136 PR tree-optimization/82244
11137 * tree-vrp.c (remove_range_assertions): Do not propagate
11138 a constant to abnormals but replace the assert with a copy.
11140 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11142 PR rtl-optimization/57878
11143 PR rtl-optimization/68988
11144 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11145 avoidance test involving non_reload_pseudos. Move frequency test
11146 below the general fragmentation avoidance test.
11148 2017-09-19 Richard Biener <rguenther@suse.de>
11150 PR tree-optimization/69728
11151 * graphite-sese-to-poly.c (schedule_error): New global.
11152 (add_loop_schedule): Handle empty domain by failing the
11154 (build_original_schedule): Handle schedule_error.
11156 2017-09-19 Richard Biener <rguenther@suse.de>
11158 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11159 Do not iterate to sibling loops but only to siblings of inner
11162 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11165 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11166 that the registers are different.
11168 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11170 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11171 to processor_model and "amdfam17h" to arch_names_table.
11172 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11174 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11177 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11179 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11180 Alan Hayward <alan.hayward@arm.com>
11181 David Sherwood <david.sherwood@arm.com>
11183 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11185 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11187 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11188 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11189 vect_schedule_slp_instance.
11190 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11191 Update call to vect_slp_analyze_node_operations. Simplify return
11193 (vect_slp_analyze_bb_1): Update call accordingly.
11194 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11195 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11196 (vect_schedule_slp): Update call accordingly.
11198 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11199 Alan Hayward <alan.hayward@arm.com>
11200 David Sherwood <david.sherwood@arm.com>
11202 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11203 with types that aren't in fact scalar.
11205 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11207 * tree-vect-slp.c (vect_record_max_nunits): New function,
11209 (vect_build_slp_tree_1): ...here.
11210 (vect_build_slp_tree_2): Call it for phis too.
11212 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11214 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11215 to vect_get_vec_def_for_operand when getting the mask operand.
11217 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11218 Alan Hayward <alan.hayward@arm.com>
11219 David Sherwood <david.sherwood@arm.com>
11221 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11224 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11225 Alan Hayward <alan.hayward@arm.com>
11226 David Sherwood <david.sherwood@arm.com>
11228 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11229 calculation for vector booleans.
11231 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11232 Alan Hayward <alan.hayward@arm.com>
11233 David Sherwood <david.sherwood@arm.com>
11235 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11237 (vect_transform_stmt): ...here.
11238 (vect_analyze_stmt): Use it instead of calling
11239 vectorizable_live_operation directly.
11241 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11243 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11244 non-SIMT targets in acc vector loops.
11246 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11248 * configure.ac: Add arc and check if assembler supports gdwarf2.
11249 * configure: Regenerate.
11251 2017-09-18 Richard Biener <rguenther@suse.de>
11253 PR tree-optimization/82220
11254 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11255 epilogue niters from the min_profitable_iters compute.
11257 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11260 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11261 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11262 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11264 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11267 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11268 switching to a new text section.
11270 2017-09-18 Richard Biener <rguenther@suse.de>
11272 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11274 (build_alias_set): Reject aliases with no access function.
11276 2017-09-18 Richard Biener <rguenther@suse.de>
11278 PR tree-optimization/79622
11279 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11281 (build_cross_bb_scalars_use): Likewise.
11283 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11285 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11287 2017-09-18 Alan Modra <amodra@gmail.com>
11290 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11291 stack_pointer_rtx for count 0. Update comments. Break up
11292 large rtl expression.
11294 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11296 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11297 Increase to 20 bytes.
11298 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11299 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11300 or avx version of the stub.
11302 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11305 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11306 compute the minimum stack alignment. Also update preferred stack
11307 boundary for leaf functions.
11309 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11311 PR tree-optimization/82228
11312 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11315 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11317 * common/config/nds32/nds32-common.c
11318 (nds32_option_optimization_table): Refine formatting.
11319 (nds32_option_optimization_table): Use -fsched-pressure and
11320 -fomit-frame-pointer for specific optimization level.
11322 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11324 * config/nds32/nds32.c: Refine formatting and comments.
11325 * config/nds32/nds32.h: Likewise.
11326 * config/nds32/nds32.md: Likewise.
11327 * config/nds32/nds32-cost.c: Likewise.
11328 * config/nds32/nds32-isr.c: Likewise.
11329 * config/nds32/nds32-md-auxiliary.c: Likewise.
11330 * config/nds32/nds32-multiple.md: Likewise.
11331 * config/nds32/nds32-predicates.c: Likewise.
11333 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11334 Jakub Jelinek <jakub@redhat.com>
11336 Add support for -std=c++2a.
11337 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11339 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11341 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11344 * doc/extend.texi (Common Function Attributes): Add
11345 references to ARM, AArch64, and S/390 specific attributes.
11346 (Function Specific Option Pragmas): Add AArch64 and S/390
11347 to list of back ends that support the target pragma.
11349 2017-09-15 Nathan Sidwell <nathan@acm.org>
11351 * doc/standards.texi: Fix C++17 description. Update URLs for
11354 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11356 * common.opt (Wcast-align=strict): New warning option.
11357 * doc/invoke.texi: Document -Wcast-align=strict.
11359 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11361 * cgraph.h (cgraph_thunk_info): Add comments.
11362 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11363 assert for VIRTUAL_* arguments stricter.
11365 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11367 PR tree-optimization/71026
11368 * match.pd: Move RDIV patterns from fold-const.c
11369 * fold-const.c (distribute_real_division): Removed.
11370 (fold_binary_loc): Remove calls to distribute_real_divison.
11372 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11374 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11375 c++1z and gnu++1z as deprecated. Change other references to
11376 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11377 Change -Wc++1z-compat to -Wc++17-compat.
11378 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11379 * dwarf2out.c (highest_c_language): Handle C++17.
11380 (gen_compile_unit_die): Likewise.
11382 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11384 PR rtl-optimization/82192
11385 * combine.c (make_extraction): Don't look through non-paradoxical
11386 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11389 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11390 Alan Hayward <alan.hayward@arm.com>
11391 David Sherwood <david.sherwood@arm.com>
11393 * target.def (function_arg_offset): New hook.
11394 * targhooks.h (default_function_arg_offset): Declare.
11395 * targhooks.c (default_function_arg_offset): New function.
11396 * function.c (locate_and_pad_parm): Use
11397 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11398 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11399 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11400 * doc/tm.texi: Regenerate.
11401 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11402 * config/spu/spu.c (spu_function_arg_offset): New function.
11403 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11404 * system.h (FUNCTION_ARG_OFFSET): Poison.
11406 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11407 Alan Hayard <alan.hayward@arm.com>
11408 David Sherwood <david.sherwood@arm.com>
11410 * target.def (truly_noop_truncation): New hook.
11411 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11412 than TRULY_NOOP_TRUNCATION.
11413 * hooks.h (hook_bool_uint_uint_true): Declare.
11414 * hooks.c (hook_bool_uint_uint_true): New function.
11415 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11416 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11417 * doc/tm.texi: Regenerate.
11418 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11419 rather than TRULY_NOOP_TRUNCATION in comments.
11420 (simplify_comparison): Likewise.
11421 (record_truncated_value): Likewise.
11422 * expmed.c (extract_bit_field_1): Likewise.
11423 (extract_split_bit_field): Likewise.
11424 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11425 instead of TRULY_NOOP_TRUNCATION.
11426 * function.c (assign_parm_setup_block): Likewise.
11427 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11428 * rtlhooks.c: Include target.h.
11429 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11430 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11431 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11432 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11433 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11434 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11435 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11436 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11437 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11438 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11439 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11440 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11441 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11442 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11443 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11444 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11445 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11446 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11447 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11448 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11449 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11450 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11451 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11452 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11453 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11454 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11455 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11456 rather than TRULY_NOOP_TRUNCATION in comments.
11457 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11458 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11459 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11460 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11461 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11462 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11463 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11464 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11465 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11466 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11467 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11468 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11469 rather than TRULY_NOOP_TRUNCATION in comments.
11470 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11471 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11472 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11473 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11474 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11475 TRULY_NOOP_TRUNCATION condition.
11476 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11477 (TRULY_NOOP_TRUNCATION): Delete.
11478 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11479 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11480 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11481 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11482 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11483 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11484 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11485 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11486 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11487 rather than TRULY_NOOP_TRUNCATION in comments.
11488 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11489 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11490 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11491 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11492 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11493 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11495 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11498 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11499 (*cmp_ior): Likewise.
11500 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11501 (*ior_scc_scc_cmp): Likewise.
11502 (*and_scc_scc): Likewise.
11503 (*and_scc_scc_cmp): Likewise.
11505 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11506 Alan Hayard <alan.hayward@arm.com>
11507 David Sherwood <david.sherwood@arm.com>
11509 * target.def (can_change_mode_class): New hook.
11510 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11511 (hard_regno_nregs): Likewise.
11512 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11513 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11514 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11515 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11516 (LOAD_EXTEND_OP): Update accordingly.
11517 * doc/tm.texi: Regenerate.
11518 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11519 CANNOT_CHANGE_MODE_CLASS.
11520 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11521 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11522 * combine.c (simplify_set): Update accordingly.
11523 * emit-rtl.c (validate_subreg): Likewise.
11524 * recog.c (general_operand): Likewise.
11525 * regcprop.c (mode_change_ok): Likewise.
11526 * reload1.c (choose_reload_regs): Likewise.
11527 (inherit_piecemeal_p): Likewise.
11528 * rtlanal.c (simplify_subreg_regno): Likewise.
11529 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11530 instead of CANNOT_CHANGE_MODE_CLASS.
11531 (reload_cse_simplify_operands): Likewise.
11532 * reload.c (push_reload): Use targetm.can_change_mode_class
11533 instead of CANNOT_CHANGE_MODE_CLASS.
11534 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11535 REG_CANNOT_CHANGE_MODE_P.
11536 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11537 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11538 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11539 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11540 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11541 (arm_can_change_mode_class): New function.
11542 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11543 than CANNOT_CHANGE_MODE_CLASS in comments.
11544 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11545 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11546 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11547 (ix86_can_change_mode_class): ...this new function, inverting the
11548 sense of the return value.
11549 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11550 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11551 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11552 (ia64_can_change_mode_class): New function.
11553 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11554 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11555 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11556 (m32c_can_change_mode_class): ...this new function, inverting the
11557 sense of the return value.
11558 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11559 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11560 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11561 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11562 (mips_can_change_mode_class): ...this new function, inverting the
11563 sense of the return value.
11564 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11565 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11566 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11567 (msp430_can_change_mode_class): New function.
11568 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11569 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11570 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11571 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11572 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11573 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11574 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11575 (pa_cannot_change_mode_class): Replace with...
11576 (pa_can_change_mode_class): ...this new function, inverting the
11577 sense of the return value.
11578 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11579 than CANNOT_CHANGE_MODE_CLASS in comments.
11580 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11581 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11582 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11583 (pdp11_cannot_change_mode_class): Replace with...
11584 (pdp11_can_change_mode_class): ...this new function, inverting the
11585 sense of the return value.
11586 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11587 * config/powerpcspe/powerpcspe-protos.h
11588 (rs6000_cannot_change_mode_class_ptr): Delete.
11589 * config/powerpcspe/powerpcspe.c
11590 (rs6000_cannot_change_mode_class_ptr): Delete.
11591 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11592 (rs6000_option_override_internal): Assign to
11593 targetm.can_change_mode_class instead of
11594 rs6000_cannot_change_mode_class_ptr.
11595 (rs6000_cannot_change_mode_class): Replace with...
11596 (rs6000_can_change_mode_class): ...this new function, inverting the
11597 sense of the return value.
11598 (rs6000_debug_cannot_change_mode_class): Replace with...
11599 (rs6000_debug_can_change_mode_class): ...this new function.
11600 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11601 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
11602 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11603 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11604 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
11606 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
11607 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11608 (rs6000_option_override_internal): Assign to
11609 targetm.can_change_mode_class instead of
11610 rs6000_cannot_change_mode_class_ptr.
11611 (rs6000_cannot_change_mode_class): Replace with...
11612 (rs6000_can_change_mode_class): ...this new function, inverting the
11613 sense of the return value.
11614 (rs6000_debug_cannot_change_mode_class): Replace with...
11615 (rs6000_debug_can_change_mode_class): ...this new function.
11616 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11617 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
11618 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
11619 (s390_can_change_mode_class): ...this new function, inverting the
11620 sense of the return value.
11621 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11622 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11623 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
11624 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11625 (sh_cannot_change_mode_class): Replace with...
11626 (sh_can_change_mode_class): ...this new function, inverting the
11627 sense of the return value.
11628 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11629 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11630 (sparc_can_change_mode_class): New function.
11631 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11632 * config/spu/spu.c (spu_can_change_mode_class): New function.
11633 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11634 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11635 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11636 (visium_can_change_mode_class): New function.
11637 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
11639 2017-09-15 Richard Biener <rguenther@suse.de>
11641 PR tree-optimization/82217
11642 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
11643 but not undefined case.
11645 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11648 * postreload.c (reload_cse_simplify_operands): Skip
11649 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
11651 2017-09-15 Richard Biener <rguenther@suse.de>
11653 PR tree-optimization/68823
11654 * graphite-scop-detection.c (build_alias_set): If we have a
11655 possible dependence check whether we can handle them by just
11656 looking at the DRs DR_ACCESS_FNs.
11657 (build_scops): If build_alias_set fails, fail the SCOP.
11659 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11661 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
11662 to support float128 built-in functions that require the ISA 3.0
11664 (BU_FLOAT128_3_HW): Likewise.
11665 (SQRTF128): Add support for the IEEE 128-bit square root and fma
11666 built-in functions.
11667 (FMAF128): Likewise.
11669 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
11670 support for built-in functions that need the ISA 3.0 IEEE 128-bit
11671 floating point instructions.
11672 (rs6000_invalid_builtin): Likewise.
11673 (rs6000_builtin_mask_names): Likewise.
11674 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
11675 (RS6000_BTM_FLOAT128_HW): Likewise.
11676 (RS6000_BTM_COMMON): Likewise.
11677 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
11679 * doc/extend.texi (RS/6000 built-in functions): Document the
11680 IEEE 128-bit floating point square root and fused multiply-add
11681 built-in functions.
11683 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
11685 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
11686 reg (r2) isn't in the set of registers defined in the prologue.
11688 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11689 Alan Hayward <alan.hayward@arm.com>
11690 David Sherwood <david.sherwood@arm.com>
11692 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
11693 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
11694 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
11695 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
11696 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
11698 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
11699 max_vectorization_factor.
11700 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
11702 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11703 Alan Hayward <alan.hayward@arm.com>
11704 David Sherwood <david.sherwood@arm.com>
11706 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
11707 (vect_worthwhile_without_simd_p): Declare.
11708 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
11709 (vectorizable_reduction): Use it.
11710 * tree-vect-stmts.c (vectorizable_shift): Likewise.
11711 (vectorizable_operation): Likewise.
11713 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11714 Alan Hayward <alan.hayward@arm.com>
11715 David Sherwood <david.sherwood@arm.com>
11717 * tree-vectorizer.h (vect_get_num_copies): New function.
11718 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
11719 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11720 (vectorizable_induction): Likewise.
11721 (vectorizable_live_operation): Likewise.
11722 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11723 (vectorizable_bswap): Likewise.
11724 (vectorizable_call): Likewise.
11725 (vectorizable_conversion): Likewise.
11726 (vectorizable_assignment): Likewise.
11727 (vectorizable_shift): Likewise.
11728 (vectorizable_operation): Likewise.
11729 (vectorizable_store): Likewise.
11730 (vectorizable_load): Likewise.
11731 (vectorizable_condition): Likewise.
11732 (vectorizable_comparison): Likewise.
11733 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
11735 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11736 Alan Hayward <alan.hayward@arm.com>
11737 David Sherwood <david.sherwood@arm.com>
11739 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
11740 of vect_init_vector.
11742 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11743 Alan Hayward <alan.hayward@arm.com>
11744 David Sherwood <david.sherwood@arm.com>
11746 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
11747 an inline wrapper that provides a location.
11748 (gimple_build_vector): Likewise.
11749 * gimple-fold.c (gimple_build_vector_from_val): New function.
11750 (gimple_build_vector): Likewise.
11751 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
11752 functions to build the initial value. Always return a gimple value.
11753 (get_initial_defs_for_reduction): Likewise. Only compute
11755 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
11756 vect_init_vector on the results from get_initial_def(s)_for_reduction.
11757 (vectorizable_induction): Use gimple_build_vector rather than
11760 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11761 Alan Hayward <alan.hayward@arm.com>
11762 David Sherwood <david.sherwood@arm.com>
11764 * target.h (vec_perm_indices): New typedef.
11765 (auto_vec_perm_indices): Likewise.
11766 * optabs-query.h: Include target.h
11767 (can_vec_perm_p): Take a vec_perm_indices *.
11768 * optabs-query.c (can_vec_perm_p): Likewise.
11769 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
11770 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11771 * tree-vect-generic.c (lower_vec_perm): Likewise.
11772 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
11773 (vect_grouped_load_supported): Likewise.
11774 (vect_shift_permute_load_chain): Likewise.
11775 (vect_permute_store_chain): Use auto_vec_perm_indices.
11776 (vect_permute_load_chain): Likewise.
11777 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
11778 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
11779 Update uses of can_vec_perm_p.
11780 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
11781 mode with a number of elements. Take a vec_perm_indices *.
11782 (vect_create_epilog_for_reduction): Update accordingly.
11783 Use auto_vec_perm_indices.
11784 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
11785 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
11786 (vect_transform_slp_perm_load): Likewise.
11787 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
11788 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
11789 (vect_gen_perm_mask_checked): Likewise.
11790 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
11791 (vect_gen_perm_mask_checked): Likewise.
11792 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
11793 (vectorizable_store): Likewise.
11794 (vectorizable_load): Likewise.
11795 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
11796 (vectorizable_bswap): Likewise.
11798 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11799 Alan Hayward <alan.hayward@arm.com>
11800 David Sherwood <david.sherwood@arm.com>
11802 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
11803 * tree.c (build_vector): Likewise.
11804 (build_vector_from_ctor): Update accordingly.
11805 (build_vector_from_val): Likewise.
11806 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
11807 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11808 * tree-vect-generic.c (add_rshift): Likewise.
11809 (expand_vector_divmod): Likewise.
11810 (optimize_vector_constructor): Likewise.
11811 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
11812 (vect_transform_slp_perm_load): Likewise.
11813 (vect_schedule_slp_instance): Likewise.
11814 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
11815 (vectorizable_call): Likewise.
11816 (vect_gen_perm_mask_any): Likewise. Add elements in order.
11817 * expmed.c (make_tree): Likewise.
11818 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
11819 a vector passed to build_vector.
11820 (fold_convert_const): Likewise.
11821 (exact_inverse): Likewise.
11822 (fold_ternary_loc): Likewise.
11823 (fold_relational_const): Likewise.
11824 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
11825 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
11826 (const_unop): Likewise. Store the reduction accumulator in a
11827 variable rather than an array.
11828 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
11829 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
11830 the new vector, rather than constructing it after the input arrays.
11831 (native_interpret_vector): Use auto_vec<tree> when building
11832 a vector passed to build_vector. Add elements in order.
11833 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
11834 auto_vec<tree> when building a vector passed to build_vector.
11835 (vect_create_epilog_for_reduction): Likewise.
11836 (vectorizable_induction): Likewise.
11837 (get_initial_def_for_reduction): Likewise. Fix indentation of
11839 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
11841 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
11842 passed to build_vector.
11844 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11845 Alan Hayward <alan.hayward@arm.com>
11846 David Sherwood <david.sherwood@arm.com>
11848 * tree-core.h (tree_base::u): Add an "nelts" field.
11849 (tree_vector): Use VECTOR_CST_NELTS as the length.
11850 * tree.c (tree_size): Likewise.
11851 (make_vector): Initialize VECTOR_CST_NELTS.
11852 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
11853 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
11854 TYPE_VECTOR_SUBPARTS.
11855 * expr.c (const_vector_mask_from_tree): Consistently use "units"
11856 as the number of units, setting it from VECTOR_CST_NELTS.
11857 (const_vector_from_tree): Likewise.
11858 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
11859 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
11860 (fold_negate_expr_1): Likewise.
11861 (fold_convert_const): Likewise.
11862 (const_binop): Likewise. Differentiate the number of output and
11864 (const_unop): Likewise.
11865 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
11866 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
11867 in cases that did the opposite.
11869 2017-09-14 Richard Biener <rguenther@suse.de>
11871 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
11874 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
11876 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
11878 2017-09-14 Jakub Jelinek <jakub@redhat.com>
11881 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
11882 if and where to split a bb, except for splitting before debug insn
11883 sequences followed by non-label real insn. Delete debug insns
11884 in between basic blocks.
11886 * combine.c (make_compound_operation_int): Formatting fixes.
11888 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
11889 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11890 * config/netbsd.h (LINK_EH_SPEC): Likewise.
11891 * config/sol2.h (LINK_EH_SPEC): Likewise.
11892 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11893 * config/s390/linux.h (LINK_SPEC): Likewise.
11894 * config/freebsd.h (LINK_EH_SPEC): Likewise.
11895 * config/openbsd.h (LINK_EH_SPEC): Likewise.
11896 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11897 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
11898 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
11899 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11900 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
11901 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
11903 2017-09-13 Jakub Jelinek <jakub@redhat.com>
11905 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
11907 (ENDFILE_LINUX_SPEC): Likewise.
11908 (LINK_EH_SPEC): Likewise.
11909 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
11910 (LINK_OS_LINUX_SPEC32): Likewise.
11911 (LINK_OS_LINUX_SPEC64): Likewise.
11912 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
11913 (LINK_OS_LINUX_SPEC): Likewise.
11915 2017-09-13 Martin Liska <mliska@suse.cz>
11917 PR middle-end/82154
11918 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
11919 CASE_HIGH is NULL_TREE.
11921 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11922 Alan Hayward <alan.hayward@arm.com>
11923 David Sherwood <david.sherwood@arm.com>
11925 * target.def (secondary_memory_needed): New hook.
11926 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
11927 instead of SECONDARY_MEMORY_NEEDED.
11928 (secondary_memory_needed_mode): Likewise.
11929 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
11930 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
11931 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
11932 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
11933 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
11934 * doc/tm.texi: Regenerate.
11935 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
11936 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
11937 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11938 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
11939 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
11940 * config/i386/i386.c (inline_secondary_memory_needed): Put the
11941 mode argument first and change the reg_class arguments to reg_class_t.
11942 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
11943 Make static. Update the call to inline_secondary_memory_needed.
11944 (ix86_register_move_cost): Update the call to
11945 inline_secondary_memory_needed.
11946 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11947 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
11949 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
11950 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11952 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
11953 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
11954 * config/mips/mips.c (mips_secondary_memory_needed): Make static
11955 and match hook interface. Add comment from mips.h.
11956 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11957 * config/mmix/mmix.md (truncdfsf2): Refer to
11958 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11960 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
11961 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
11962 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11963 (pa_secondary_memory_needed): New function.
11964 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
11965 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
11966 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11967 (pdp11_secondary_memory_needed): Make static and match hook interface.
11968 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
11969 * config/powerpcspe/powerpcspe-protos.h
11970 (rs6000_secondary_memory_needed_ptr): Delete.
11971 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
11973 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11974 (rs6000_option_override_internal): Assign to
11975 targetm.secondary_memory_needed rather than
11976 rs6000_secondary_memory_needed_ptr.
11977 (rs6000_secondary_memory_needed): Match hook interface.
11978 (rs6000_debug_secondary_memory_needed): Likewise.
11979 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
11980 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
11981 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
11982 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11983 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
11984 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
11986 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
11987 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11988 (rs6000_option_override_internal): Assign to
11989 targetm.secondary_memory_needed rather than
11990 rs6000_secondary_memory_needed_ptr.
11991 (rs6000_secondary_memory_needed): Match hook interface.
11992 (rs6000_debug_secondary_memory_needed): Likewise.
11993 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
11994 * config/s390/s390.c (s390_secondary_memory_needed): New function.
11995 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11996 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
11997 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11998 (sparc_secondary_memory_needed): New function.
11999 * lra-constraints.c (check_and_process_move): Refer to
12000 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12002 (curr_insn_transform): Likewise.
12003 (process_alt_operands): Use targetm.secondary_memory_needed
12004 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12005 (check_secondary_memory_needed_p): Likewise.
12006 (choose_split_class): Likewise.
12007 * reload.c: Unconditionally include code that was previously
12008 conditional on SECONDARY_MEMORY_NEEDED.
12009 (push_secondary_reload): Use targetm.secondary_memory_needed
12010 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12011 (push_reload): Likewise.
12012 * reload1.c: Unconditionally include code that was previously
12013 conditional on SECONDARY_MEMORY_NEEDED.
12014 (choose_reload_regs): Use targetm.secondary_memory_needed
12015 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12016 (gen_reload): Likewise.
12017 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12019 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12020 Alan Hayward <alan.hayward@arm.com>
12021 David Sherwood <david.sherwood@arm.com>
12023 * target.def (secondary_memory_needed_mode): New hook:
12024 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12025 * targhooks.h (default_secondary_memory_needed_mode): New function.
12026 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12027 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12028 * doc/tm.texi: Regenerate.
12029 * lra-constraints.c (check_and_process_move): Use
12030 targetm.secondary_memory_needed_mode instead of
12031 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12032 (curr_insn_transform): Likewise.
12033 * reload.c (get_secondary_mem): Likewise.
12034 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12035 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12037 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12038 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12039 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12040 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12041 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12043 * config/powerpcspe/powerpcspe-protos.h
12044 (rs6000_secondary_memory_needed_mode): Delete.
12045 * config/powerpcspe/powerpcspe.c
12046 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12047 (rs6000_secondary_memory_needed_mode): Make static.
12048 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12049 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12051 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12053 (rs6000_secondary_memory_needed_mode): Make static.
12054 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12055 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12056 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12057 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12058 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12060 (sparc_secondary_memory_needed_mode): New function.
12061 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12063 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12065 * config/aarch64/constraints.md (Umq): New constraint.
12066 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12068 (mov<mode>): Update condition.
12070 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12072 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12073 when bitposition is the same.
12075 2017-09-13 Richard Biener <rguenther@suse.de>
12077 * dwarf2out.c (output_die_symbol): Remove.
12078 (output_die): Do not output a DIEs symbol.
12080 2017-09-13 Richard Biener <rguenther@suse.de>
12082 PR middle-end/82128
12083 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12084 default-def to avoid breaking iterator update with the weird
12085 interaction with cgraph_update_edges_for_call_stmt_node.
12087 2017-09-13 Richard Biener <rguenther@suse.de>
12089 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12090 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12091 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12092 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12094 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12096 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12097 Disable pc relative literal load irrespective of
12098 TARGET_FIX_ERR_A53_84341 for default.
12100 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12102 * config/sparc/sparc.c (output_return): Output the source location of
12103 the insn in the delay slot, if any.
12104 (output_sibcall): Likewise.
12106 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12109 * common.opt (-static-pie): New alias.
12110 (shared): Negate static-pie.
12111 (-no-pie): Update help text.
12113 (static-pie): New option.
12114 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12115 -static-pie support.
12116 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12117 (LINK_EH_SPEC): Likewise.
12118 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12119 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12120 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12121 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12122 (init_gcc_specs): Likewise.
12123 (init_spec): Likewise.
12124 (display_help): Update help message for -pie.
12125 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12128 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12130 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12131 (movdi_aarch64): Likewise.
12132 (movti_aarch64): Likewise.
12134 2017-09-12 Simon Wright <simon@pushface.org>
12137 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12138 calculation of the minor version, always output as 0.
12140 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12143 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12144 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12145 on it early, rather than manual conversion late. For
12146 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12147 instead of performing manual conversion.
12149 2017-09-12 Carl Love <cel@us.ibm.com>
12151 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12152 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12154 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12155 VMULOSW): Add definitions.
12156 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12157 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12158 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12159 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12160 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12162 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12164 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12166 (movti_aarch64): Likewise.
12167 (movdf_aarch64): Likewise.
12168 (movtf_aarch64): Likewise.
12169 (load_pairdi): Likewise.
12170 (store_pairdi): Likewise.
12171 (load_pairdf): Likewise.
12172 (store_pairdf): Likewise.
12173 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12174 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12175 (ldr_got_small_<mode>): Likewise.
12176 (ldr_got_small_28k_<mode>): Likewise.
12177 (ldr_got_tiny): Likewise.
12178 * config/aarch64/iterators.md (ldst_sz): New.
12179 (ldpstp_sz): Likewise.
12180 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12182 (thunderx_load): Split load_8 to load_16.
12183 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12185 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12186 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12187 (xgene1_store_pair): Split store_8 to store_16.
12188 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12189 (falkor_st_0_st_sd): Split store_8 to store_16.
12191 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12193 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12194 and store1/2/3/4 to store_4/8/12/16.
12195 * config/aarch64/aarch64.md: Update for rename.
12196 * config/arm/arm.md: Likewise.: Likewise.
12197 * config/arm/arm.c: Likewise.
12198 * config/arm/thumb1.md: Likewise.
12199 * config/arm/thumb2.md: Likewise.
12200 * config/arm/vfp.md: Likewise.
12201 * config/arm/arm-generic.md: Likewise.
12202 * config/arm/arm1020e.md: Likewise.
12203 * config/arm/arm1026ejs.md: Likewise.
12204 * config/arm/arm1136jfs.md: Likewise.
12205 * config/arm/arm926ejs.md: Likewise.
12206 * config/arm/cortex-a15.md: Likewise.
12207 * config/arm/cortex-a17.md: Likewise.
12208 * config/arm/cortex-a5.md: Likewise.
12209 * config/arm/cortex-a53.md: Likewise.
12210 * config/arm/cortex-a57.md: Likewise.
12211 * config/arm/cortex-a7.md: Likewise.
12212 * config/arm/cortex-a8.md: Likewise.
12213 * config/arm/cortex-a9.md: Likewise.
12214 * config/arm/cortex-m4.md: Likewise.
12215 * config/arm/cortex-m7.md: Likewise.
12216 * config/arm/cortex-r4.md: Likewise.
12217 * config/arm/exynos-m1.md: Likewise.
12218 * config/arm/fa526.md: Likewise.
12219 * config/arm/fa606te.md: Likewise.
12220 * config/arm/fa626te.md: Likewise.
12221 * config/arm/fa726te.md: Likewise.
12222 * config/arm/fmp626.md: Likewise.
12223 * config/arm/iwmmxt.md: Likewise.
12224 * config/arm/ldmstm.md: Likewise.
12225 * config/arm/marvell-pj4.md: Likewise.
12226 * config/arm/xgene1.md: Likewise.
12227 * config/aarch64/thunderx.md: Likewise.
12228 * config/aarch64/thunderx2t99.md: Likewise.
12229 * config/aarch64/falkor.md: Likewise.
12231 2017-09-12 Martin Liska <mliska@suse.cz>
12233 * attribs.c (private_lookup_attribute): New function.
12234 * attribs.h (private_lookup_attribute): Declared here.
12235 (lookup_attribute): Called from this place.
12237 2017-09-12 Richard Biener <rguenther@suse.de>
12239 PR tree-optimization/82157
12240 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12241 stmts with side-effects.
12243 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12244 Alan Hayward <alan.hayward@arm.com>
12245 David Sherwood <david.sherwood@arm.com>
12247 * target.def (hard_regno_nregs): New hook.
12248 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12249 * targhooks.h (default_hard_regno_nregs): Declare.
12250 * targhooks.c (default_hard_regno_nregs): New function.
12251 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12252 (TARGET_HARD_REGNO_NREGS): ...this hook.
12253 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12254 (CLASS_MAX_NREGS): Likewise.
12255 * doc/tm.texi: Regenerate.
12256 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12257 instead of HARD_REGNO_NREGS.
12258 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12259 HARD_REGNO_NREGS in the comment.
12260 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12261 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12262 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12263 Return an unsigned int.
12264 (TARGET_HARD_REGNO_NREGS): Redefine.
12265 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12266 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12267 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12268 (arc_hard_regno_nregs): New function.
12269 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12270 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12271 (arm_hard_regno_nregs): New function.
12272 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12273 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12274 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12275 (TARGET_HARD_REGNO_NREGS): Redefine.
12276 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12277 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12278 (HARD_REGNO_NREGS): Delete.
12279 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12280 (cr16_hard_regno_nregs): New function.
12281 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12282 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12283 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12284 (cris_hard_regno_nregs): New function.
12285 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12286 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12287 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12288 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12289 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12290 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12291 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12292 (frv_hard_regno_nregs): Make static. Take and return an
12294 (frv_class_max_nregs): Remove outdated copy of documentation.
12295 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12296 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12297 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12298 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12299 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12300 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12301 (TARGET_HARD_REGNO_NREGS): Redefine.
12302 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12303 (CLASS_MAX_NREGS): Update comment.
12304 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12305 (ia64_hard_regno_nregs): New function.
12306 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12307 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12308 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12309 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12310 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12312 (m32c_hard_regno_nregs): Likewise. Make static.
12313 (TARGET_HARD_REGNO_NREGS): Redefine.
12314 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12315 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12316 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12317 (m68k_hard_regno_nregs): New function.
12318 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12319 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12320 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12321 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12322 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12323 Take and return an unsigned int.
12324 (TARGET_HARD_REGNO_NREGS): Redefine.
12325 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12326 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12327 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12328 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12329 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12330 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12331 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12332 (msp430_hard_regno_nregs): Make static. Take and return an
12334 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12335 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12336 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12337 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12338 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12339 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12340 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12341 (TARGET_HARD_REGNO_NREGS): Redefine.
12342 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12343 (PA_HARD_REGNO_NREGS): ...this.
12344 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12345 (PA_HARD_REGNO_NREGS): ...this.
12346 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12347 (pa_hard_regno_nregs): New function.
12348 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12349 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12350 (pdp11_hard_regno_nregs): New function.
12351 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12352 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12353 (rs6000_hard_regno_nregs_hook): New function.
12354 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12355 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12356 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12357 Take and return an unsigned int. Move earlier in file.
12358 (TARGET_HARD_REGNO_NREGS): Redefine.
12359 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12360 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12361 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12362 (rl78_hard_regno_nregs): Make static. Take and return an
12364 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12365 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12366 (rs6000_hard_regno_nregs_hook): New function.
12367 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12368 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12369 (TARGET_HARD_REGNO_NREGS): Redefine.
12370 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12371 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12372 instead of HARD_REGNO_NREGS.
12373 (s390_hard_regno_nregs): New function.
12374 (s390_hard_regno_mode_ok): Add comment from s390.h.
12375 (TARGET_HARD_REGNO_NREGS): Redefine.
12376 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12377 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12378 (sh_hard_regno_nregs): New function.
12379 (sh_pass_in_reg_p): Use it.
12380 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12381 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12382 (sparc_hard_regno_nregs): New function.
12383 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12384 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12385 (spu_function_arg_advance): Use it, supplying a valid register number.
12386 (TARGET_HARD_REGNO_NREGS): Redefine.
12387 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12388 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12389 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12390 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12391 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12392 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12393 (CLASS_MAX_NREGS): Remove copy of old documentation.
12394 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12395 (visium_hard_regno_nregs): New function.
12396 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12397 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12398 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12399 xtensa_hard_regno_nregs): New function.
12400 * system.h (HARD_REGNO_NREGS): Poison.
12402 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12404 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12405 hard_regno_nregs instead of HARD_REGNO_NREGS.
12406 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12407 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12408 (c6x_expand_epilogue): Likewise.
12409 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12410 (frv_read_iacc_argument): Likewise.
12411 * config/sh/sh.c: Include regs.h.
12412 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12413 (regs_used): Likewise.
12414 (output_stack_adjust): Likewise.
12415 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12416 * expmed.c: Include regs.h.
12417 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12418 * ree.c: Include regs.h.
12419 (combine_reaching_defs): Use hard_regno_nregs instead of
12421 (add_removable_extension): Likewise.
12423 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12425 * regs.h (hard_regno_nregs): Turn into a function.
12426 (end_hard_regno): Update accordingly.
12427 * caller-save.c (setup_save_areas): Likewise.
12428 (save_call_clobbered_regs): Likewise.
12429 (replace_reg_with_saved_mem): Likewise.
12430 (insert_restore): Likewise.
12431 (insert_save): Likewise.
12432 * combine.c (can_change_dest_mode): Likewise.
12433 (move_deaths): Likewise.
12434 (distribute_notes): Likewise.
12435 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12436 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12437 (rs6000_split_multireg_move): Likewise.
12438 (rs6000_register_move_cost): Likewise.
12439 (rs6000_memory_move_cost): Likewise.
12440 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12441 (rs6000_split_multireg_move): Likewise.
12442 (rs6000_register_move_cost): Likewise.
12443 (rs6000_memory_move_cost): Likewise.
12444 * cselib.c (cselib_reset_table): Likewise.
12445 (cselib_lookup_1): Likewise.
12446 * emit-rtl.c (set_mode_and_regno): Likewise.
12447 * function.c (aggregate_value_p): Likewise.
12448 * ira-color.c (setup_profitable_hard_regs): Likewise.
12449 (check_hard_reg_p): Likewise.
12450 (calculate_saved_nregs): Likewise.
12451 (assign_hard_reg): Likewise.
12452 (improve_allocation): Likewise.
12453 (calculate_spill_cost): Likewise.
12454 * ira-emit.c (modify_move_list): Likewise.
12455 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12456 (ira_hard_reg_in_set_p): Likewise.
12457 * ira.c (setup_reg_mode_hard_regset): Likewise.
12458 (clarify_prohibited_class_mode_regs): Likewise.
12459 (check_allocation): Likewise.
12460 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12461 (lra_setup_reg_renumber): Likewise.
12462 (setup_try_hard_regno_pseudos): Likewise.
12463 (spill_for): Likewise.
12464 (assign_hard_regno): Likewise.
12465 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12466 * lra-constraints.c (in_class_p): Likewise.
12467 (lra_constraint_offset): Likewise.
12468 (simplify_operand_subreg): Likewise.
12469 (lra_constraints): Likewise.
12470 (split_reg): Likewise.
12471 (split_if_necessary): Likewise.
12472 (invariant_p): Likewise.
12473 (inherit_in_ebb): Likewise.
12474 * lra-lives.c (process_bb_lives): Likewise.
12475 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12476 (get_hard_regs): Likewise.
12477 (do_remat): Likewise.
12478 * lra-spills.c (assign_spill_hard_regs): Likewise.
12479 * mode-switching.c (create_pre_exit): Likewise.
12480 * postreload.c (reload_combine_recognize_pattern): Likewise.
12481 * recog.c (peep2_find_free_register): Likewise.
12482 * regcprop.c (kill_value_regno): Likewise.
12483 (set_value_regno): Likewise.
12484 (copy_value): Likewise.
12485 (maybe_mode_change): Likewise.
12486 (find_oldest_value_reg): Likewise.
12487 (copyprop_hardreg_forward_1): Likewise.
12488 * regrename.c (check_new_reg_p): Likewise.
12489 (regrename_do_replace): Likewise.
12490 * reload.c (push_reload): Likewise.
12491 (combine_reloads): Likewise.
12492 (find_dummy_reload): Likewise.
12493 (operands_match_p): Likewise.
12494 (find_reloads): Likewise.
12495 (find_equiv_reg): Likewise.
12496 (reload_adjust_reg_for_mode): Likewise.
12497 * reload1.c (count_pseudo): Likewise.
12498 (count_spilled_pseudo): Likewise.
12499 (find_reg): Likewise.
12500 (clear_reload_reg_in_use): Likewise.
12501 (free_for_value_p): Likewise.
12502 (allocate_reload_reg): Likewise.
12503 (choose_reload_regs): Likewise.
12504 (reload_adjust_reg_for_temp): Likewise.
12505 (emit_reload_insns): Likewise.
12506 (delete_output_reload): Likewise.
12507 * rtlanal.c (subreg_get_info): Likewise.
12508 * sched-deps.c (sched_analyze_reg): Likewise.
12509 * sel-sched.c (init_regs_for_mode): Likewise.
12510 (mark_unavailable_hard_regs): Likewise.
12511 (choose_best_reg_1): Likewise.
12512 (verify_target_availability): Likewise.
12513 * valtrack.c (dead_debug_insert_temp): Likewise.
12514 * var-tracking.c (track_loc_p): Likewise.
12515 (emit_note_insn_var_location): Likewise.
12516 * varasm.c (make_decl_rtl): Likewise.
12517 * reginfo.c (choose_hard_reg_mode): Likewise.
12518 (init_reg_modes_target): Refer directly to
12519 this_target_regs->x_hard_regno_nregs.
12521 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12523 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12524 instead of hard_regno_nregs.
12526 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12528 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12529 end_hard_regno instead of hard_regno_nregs.
12530 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12531 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12532 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12533 * ira-color.c (improve_allocation): Likewise.
12534 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12535 * lra-lives.c (mark_regno_live): Likewise.
12536 (mark_regno_dead): Likewise.
12537 * lra-remat.c (operand_to_remat): Likewise.
12538 * lra.c (collect_non_operand_hard_regs): Likewise.
12539 * postreload.c (reload_combine_note_store): Likewise.
12540 (move2add_valid_value_p): Likewise.
12541 * reload.c (regno_clobbered_p): Likewise.
12543 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12545 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12547 * config/v850/v850.c (v850_reorg): Likewise.
12548 * reload.c (refers_to_regno_for_reload_p): Likewise.
12549 (find_equiv_reg): Likewise.
12550 * reload1.c (reload_reg_reaches_end_p): Likewise.
12552 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12554 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12556 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12557 * config/arm/arm.c (output_move_neon): Likewise.
12558 (arm_attr_length_move_neon): Likewise.
12559 (neon_split_vcombine): Likewise.
12560 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12561 (c6x_mark_reg_written): Likewise.
12562 (c6x_dwarf_register_span): Likewise.
12563 * config/i386/i386.c (ix86_save_reg): Likewise.
12564 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12565 (rws_access_reg): Likewise.
12566 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12567 * mode-switching.c (create_pre_exit): Likewise.
12568 * ree.c (combine_reaching_defs): Likewise.
12569 (add_removable_extension): Likewise.
12570 * regcprop.c (find_oldest_value_reg): Likewise.
12571 (copyprop_hardreg_forward_1): Likewise.
12572 * reload.c (reload_inner_reg_of_subreg): Likewise.
12573 (push_reload): Likewise.
12574 (combine_reloads): Likewise.
12575 (find_dummy_reload): Likewise.
12576 (reload_adjust_reg_for_mode): Likewise.
12577 * reload1.c (find_reload_regs): Likewise.
12578 (forget_old_reloads_1): Likewise.
12579 (reload_reg_free_for_value_p): Likewise.
12580 (reload_adjust_reg_for_temp): Likewise.
12581 (emit_reload_insns): Likewise.
12582 (delete_output_reload): Likewise.
12583 * sel-sched.c (choose_best_reg_1): Likewise.
12584 (choose_best_pseudo_reg): Likewise.
12586 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12587 Alan Hayward <alan.hayward@arm.com>
12588 David Sherwood <david.sherwood@arm.com>
12590 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12591 * target.def (slow_unaligned_access): New hook.
12592 * targhooks.h (default_slow_unaligned_access): Declare.
12593 * targhooks.c (default_slow_unaligned_access): New function.
12594 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
12595 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
12596 * doc/tm.texi: Regenerate.
12597 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
12598 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
12599 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
12601 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
12602 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
12604 (rs6000_slow_unaligned_access): New function.
12605 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12606 (expand_block_compare): Likewise.
12607 (expand_strn_compare): Likewise.
12608 (rs6000_rtx_costs): Likewise.
12609 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
12610 (riscv_slow_unaligned_access): Likewise.
12611 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
12612 (riscv_slow_unaligned_access_p): ...this and make static.
12613 (riscv_option_override): Update accordingly.
12614 (riscv_slow_unaligned_access): New function.
12615 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12616 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
12617 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12618 (rs6000_slow_unaligned_access): New function.
12619 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12620 (rs6000_rtx_costs): Likewise.
12621 * config/rs6000/rs6000-string.c (expand_block_compare)
12622 (expand_strn_compare): Use targetm.slow_unaligned_access instead
12623 of SLOW_UNALIGNED_ACCESS.
12624 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
12625 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
12626 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
12627 of SLOW_UNALIGNED_ACCESS.
12628 * expmed.c (simple_mem_bitfield_p): Likewise.
12629 * expr.c (alignment_for_piecewise_move): Likewise.
12630 (emit_group_load_1): Likewise.
12631 (emit_group_store): Likewise.
12632 (copy_blkmode_from_reg): Likewise.
12633 (emit_push_insn): Likewise.
12634 (expand_assignment): Likewise.
12635 (store_field): Likewise.
12636 (expand_expr_real_1): Likewise.
12637 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12638 * lra-constraints.c (simplify_operand_subreg): Likewise.
12639 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12640 * gimple-ssa-store-merging.c: Likewise in block comment at start
12642 * tree-ssa-strlen.c: Include target.h.
12643 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
12644 of SLOW_UNALIGNED_ACCESS.
12645 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
12647 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12649 PR rtl-optimization/82185
12650 * expmed.c (emit_store_flag_int): Only test tem if it has been
12653 2017-09-12 Richard Biener <rguenther@suse.de>
12655 PR middle-end/82149
12656 * match.pd ((FTYPE) N CMP CST): Fix typo.
12658 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
12660 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
12662 (mips_near_type_p): Add 'short_call' attribute as a synonym
12664 * doc/extend.texi (short_call): Document new function attribute.
12666 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12669 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
12670 assertion check that in the condition.
12671 (get_atomic_generic_size): Likewise. Before testing if parameter
12672 has pointer type, if it has array type, call for C++
12673 default_conversion to perform array-to-pointer conversion.
12675 2017-09-12 Richard Biener <rguenther@suse.de>
12677 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
12678 for operations we cannot scalarize.
12680 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
12682 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
12683 vectors heap vectors. Clean up comments.
12684 Make visited_bbs a reference.
12685 (profitable_jump_thread_path): Make GC
12686 vectors heap vectors. Clean up comments.
12688 (convert_and_register_jump_thread_path): Make GC vectors heap
12690 (check_subpath_and_update_thread_path): Same. Clean up comments.
12691 Make visited_bbs a reference.
12692 (handle_phi): Abstract common code to to
12693 register_jump_thread_path_if_profitable.
12694 Rename VAR_BB to DEF_BB.
12696 Make GC vectors heap vectors.
12697 Make visited_bbs a reference.
12698 (handle_assignment): Same.
12699 (register_jump_thread_path_if_profitable): New.
12700 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
12702 Make GC vectors heap vectors. Clean up comments.
12703 Make visited_bbs a reference.
12704 (find_jump_threads_backwards): Make visited_bbs live in the stack.
12705 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
12708 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
12711 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
12712 words of E_DImode object are reachable by xtensa_uimm8x4 access.
12714 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
12716 Revert r251800 and r251799.
12718 2017-09-11 Martin Jambor <mjambor@suse.cz>
12721 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
12722 arguments in advance.
12723 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
12724 use it to find predecessor edges.
12725 (naive_outof_ssa): Collect vector of predecessors.
12727 2017-09-08 Jason Merrill <jason@redhat.com>
12729 PR c++/70029 - ICE with ref-qualifier and -flto
12730 * langhooks.h (struct lang_hooks_for_types): Add
12731 copy_lang_qualifiers.
12732 * attribs.c (build_type_attribute_qual_variant): Use it.
12733 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
12735 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
12736 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
12738 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
12741 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
12742 (*mulsi3_sp64): New instruction.
12743 (mulsi3): New expander.
12745 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
12747 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
12749 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12751 * sancov.c: Include memmodel.h.
12753 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
12756 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
12759 2017-09-07 Carl Love <cel@us.ibm.com>
12761 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
12762 the sldi instruction.
12764 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
12766 * sancov.c: Include tm_p.h.
12768 2017-09-07 Jakub Jelinek <jakub@redhat.com>
12771 * output.h (switch_to_other_text_partition): New declaration.
12772 * varasm.c (switch_to_other_text_partition): New function.
12773 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
12774 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
12775 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
12776 to the other text partition before emitting LCL label and switch back
12777 after emitting the word after it.
12779 2017-09-07 Richard Biener <rguenther@suse.de>
12781 * passes.def (pass_split_crit_edges): Remove instance before PRE.
12782 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
12783 critical edges here, after loop init.
12784 (pass_data_pre): Remove PROP_no_crit_edges flags.
12785 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
12786 for valueization of call args to avoid leaking VN_TOP.
12787 (visit_use): Assert we do not visit default defs.
12788 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
12789 Use error_mark_node to more easily detect leaking VN_TOP.
12790 All default-defs are varying, not VN_TOP. Mark them visited.
12791 (run_scc_vn): Make code match comment.
12793 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12795 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
12796 OPTION_MASK_FLOAT128_KEYWORD.
12797 (POWERPC_MASKS): Likewise.
12798 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
12799 support for the -mfloat128-type option, and make -mfloat128
12800 default on PowerPC Linux systems. Define or undefine
12801 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
12802 Define __float128 to be __ieee128 if IEEE 128-bit support is
12803 enabled, or undefine it.
12804 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
12805 Delete defining __FLOAT128_TYPE__.
12806 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
12807 -mfloat128-type option and make -mfloat128 default on PowerPC
12809 (TARGET_FLOAT128_TYPE): Likewise.
12810 (-mfloat128-type): Likewise.
12811 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12812 Delete the -mfloat128-type option and make -mfloat128 default on
12813 PowerPC Linux systems. Always use __ieee128 to be the keyword for
12814 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
12815 128-bit floating point is enabled. Change tests from using
12816 -mfloat128-type to -mfloat128.
12817 (rs6000_mangle_type): Use the correct mangling for the __float128
12818 type even if normal long double is restricted to 64-bits.
12819 (floatn_mode): Enable the _Float128 type by default on VSX Linux
12821 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
12822 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
12823 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
12825 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
12826 documentation for -mfloat128.
12828 2017-09-06 Olivier Hainque <hainque@adacore.com>
12830 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
12832 2017-09-06 Wish Wu <wishwu007@gmail.com>
12833 Jakub Jelinek <jakub@redhat.com>
12835 * asan.c (initialize_sanitizer_builtins): Add
12836 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
12837 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
12838 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
12839 BT_FN_VOID_UINT64_PTR variables.
12840 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
12841 (BT_FN_VOID_UINT16_UINT16): Likewise.
12842 (BT_FN_VOID_UINT32_UINT32): Likewise.
12843 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
12844 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
12845 (BT_FN_VOID_UINT64_PTR): Likewise.
12846 * common.opt (flag_sanitize_coverage): New variable.
12847 (fsanitize-coverage=trace-pc): Remove.
12848 (fsanitize-coverage=): Add.
12849 * flag-types.h (enum sanitize_coverage_code): New enum.
12850 * fold-const.c (fold_range_test): Disable non-short-circuit
12851 optimization if flag_sanitize_coverage.
12852 (fold_truth_andor): Likewise.
12853 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12854 * opts.c (COVERAGE_SANITIZER_OPT): Define.
12855 (coverage_sanitizer_opts): New array.
12856 (get_closest_sanitizer_option): Add OPTS argument, handle also
12857 OPT_fsanitize_coverage_.
12858 (parse_sanitizer_options): Adjusted to also handle
12859 OPT_fsanitize_coverage_.
12860 (common_handle_option): Add OPT_fsanitize_coverage_.
12861 * sancov.c (instrument_comparison, instrument_switch): New function.
12862 (sancov_pass): Add trace-cmp support.
12863 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
12864 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
12865 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
12866 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
12867 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
12868 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
12869 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
12870 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
12871 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
12872 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
12874 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12876 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12877 error. Only quit immediately if parsing is complete.
12878 (BEGIN): Initialize fatal_err and parse_done.
12879 (begin fpu, end fpu): Check number of arguments.
12880 (begin arch, end arch): Likewise.
12881 (begin cpu, end cpu): Likewise.
12882 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12883 (optalias): Likewise.
12885 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12887 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12888 * config/arm/arm-isa.h: Delete. Move definitions to ...
12889 * arm-cpus.in: ... here. Use new feature and fgroup values.
12890 * config/arm/arm.c (arm_option_override): Use lower case for feature
12892 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12893 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12894 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12896 (print_isa_bits_for): New function.
12897 (gen_isa): New function.
12898 (gen_comm_data): Use print_isa_bits_for.
12899 (define feature): New keyword.
12900 (define fgroup): New keyword.
12901 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
12902 (arm-isa.h): Add rule to generate file.
12903 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12904 case for feature bit names.
12906 2017-09-06 Richard Biener <rguenther@suse.de>
12908 * tree-ssa-pre.c (NECESSARY): Remove.
12909 (create_expression_by_pieces): Do not touch pass-local flags.
12910 (insert_into_preds_of_block): Likewise.
12911 (do_pre_regular_insertion): Likewise.
12912 (eliminate_insert): Likewise.
12913 (eliminate_dom_walker::before_dom_children): Likewise.
12914 (fini_eliminate): Do not look at inserted_exprs.
12915 (mark_operand_necessary): Remove.
12916 (remove_dead_inserted_code): Replace with simple work-list
12917 algorithm based on inserted_exprs and SSA uses.
12918 (pass_pre::execute): Re-order fini_eliminate and
12919 remove_dead_inserted_code.
12921 2017-09-06 Olivier Hainque <hainque@adacore.com>
12923 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
12924 for VxWorks 7. Adjust surrounding comments.
12926 2017-09-06 Richard Biener <rguenther@suse.de>
12928 * gimple-ssa-strength-reduction.c
12929 (find_candidates_dom_walker::before_dom_children): Also allow
12932 2017-09-06 Richard Biener <rguenther@suse.de>
12934 PR tree-optimization/82108
12935 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
12936 for gap in the non-permutation SLP case.
12938 2017-09-06 Martin Jambor <mjambor@suse.cz>
12940 PR tree-optimization/82078
12941 * tree-sra.c (sort_and_splice_var_accesses): Move call to
12942 add_access_to_work_queue...
12943 (build_accesses_from_assign): ...here.
12944 (propagate_all_subaccesses): Make sure racc is the group
12945 representative, if there is one.
12947 2017-09-06 Jakub Jelinek <jakub@redhat.com>
12949 PR middle-end/82095
12950 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
12953 2017-09-06 Richard Biener <rguenther@suse.de>
12955 * gimple-ssa-strength-reduction.c
12956 (find_candidates_dom_walker::before_doom_children): Use a
12957 type and not a mode check.
12959 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
12962 * config/arm/predicates.md (arm_general_adddi_operand): Create new
12964 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
12966 2017-09-05 Jeff Law <law@redhat.com>
12968 PR tree-optimization/64910
12969 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
12970 cases where we have 3 or more operands.
12972 2017-09-05 Jakub Jelinek <jakub@redhat.com>
12974 PR middle-end/81768
12975 * omp-low.c (lower_omp_for): Recompute tree invariant if
12976 gimple_omp_for_initial/final is ADDR_EXPR.
12978 PR middle-end/81768
12979 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
12980 into gimple val before gimplification fo the COND_EXPR.
12982 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
12984 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
12985 REGION_COPY argument.
12986 (thread_through_all_blocks): Remove unused argument to
12987 duplicate_thread_path.
12989 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12990 Alan Hayward <alan.hayward@arm.com>
12991 David Sherwood <david.sherwood@arm.com>
12993 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
12994 Take a scalar_mode rather than a machine_mode.
12995 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12996 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
12997 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12998 (aarch64_gen_adjusted_ldpstp): Likewise.
12999 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13001 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13002 Alan Hayward <alan.hayward@arm.com>
13003 David Sherwood <david.sherwood@arm.com>
13005 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13006 Take a scalar_int_mode instead of a machine_mode.
13007 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13008 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13009 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13010 (aarch64_simd_attr_length_rglist): Delete.
13011 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13012 a scalar_int_mode instead of a machine_mode.
13013 (aarch64_add_offset): Likewise.
13014 (aarch64_internal_mov_immediate): Likewise
13015 (aarch64_add_constant_internal): Likewise.
13016 (aarch64_add_constant): Likewise.
13017 (aarch64_movw_imm): Likewise.
13018 (aarch64_rtx_arith_op_extract_p): Likewise.
13019 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13020 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13021 Remove assert that the mode isn't a vector.
13022 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13023 (aarch64_expand_mov_immediate): Update calls after above changes.
13024 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13025 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13026 (aarch64_move_imm): Likewise.
13027 (aarch64_can_const_movi_rtx_p): Likewise.
13028 (aarch64_strip_extend): Likewise.
13029 (aarch64_extr_rtx_p): Likewise.
13030 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13031 a CONST_INT when the mode parameter is VOIDmode.
13032 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13034 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13036 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13037 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13038 (bitwise_type_for_mode): Update accordingly.
13040 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13042 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13043 * stor-layout.c (mode_for_size_tree): Likewise.
13044 (mode_for_array): Update accordingly.
13045 (layout_decl): Likewise.
13046 (compute_record_mode): Likewise. Only set the mode once.
13048 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13050 * target.def (get_mask_mode): Change return type to opt_mode.
13052 * doc/tm.texi: Regenerate.
13053 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13054 * targhooks.c (default_get_mask_mode): Likewise.
13055 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13056 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13057 targetm.get_mask_mode.
13058 * tree.c (build_truth_vector_type): Likewise.
13060 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13062 * machmode.h (mode_for_vector): Return an opt_mode.
13063 * stor-layout.c (mode_for_vector): Likewise.
13064 (mode_for_int_vector): Update accordingly.
13065 (layout_type): Likewise.
13066 * config/i386/i386.c (emit_memmov): Likewise.
13067 (ix86_expand_set_or_movmem): Likewise.
13068 (ix86_expand_vector_init): Likewise.
13069 (ix86_get_mask_mode): Likewise.
13070 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13072 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13073 * expmed.c (extract_bit_field_1): Likewise.
13074 * expr.c (expand_expr_real_2): Likewise.
13075 * optabs-query.c (can_vec_perm_p): Likewise.
13076 (can_vec_mask_load_store_p): Likewise.
13077 * optabs.c (expand_vec_perm): Likewise.
13078 * targhooks.c (default_get_mask_mode): Likewise.
13079 * tree-vect-stmts.c (vectorizable_store): Likewise.
13080 (vectorizable_load): Likewise.
13081 (get_vectype_for_scalar_type_and_size): Likewise.
13083 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13085 * machmode.h (mode_for_int_vector): New function.
13086 * stor-layout.c (mode_for_int_vector): Likewise.
13087 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13088 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13089 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13090 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13091 (s390_expand_vcond): Likewise.
13093 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13095 * machmode.h (opt_machine_mode): New type.
13096 (opt_mode<T>): Allow construction from anything that can be
13098 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13099 (mode_for_size): Return an opt_machine_mode.
13100 * stor-layout.c (mode_for_size): Likewise.
13101 (mode_for_size_tree): Update call accordingly.
13102 (bitwise_mode_for_mode): Likewise.
13103 (make_fract_type): Likewise.
13104 (make_accum_type): Likewise.
13105 * caller-save.c (replace_reg_with_saved_mem): Update call
13107 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13108 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13109 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13110 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13111 * expmed.c (extract_bit_field_1): Likewise.
13112 * reload.c (get_secondary_mem): Likewise.
13113 * varasm.c (assemble_integer): Likewise.
13114 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13117 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13119 * machmode.h (decimal_float_mode_for_size): New function.
13120 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13121 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13122 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13123 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13124 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13125 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13127 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13129 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13130 (get_builtin_sync_mode): Likewise.
13131 (expand_ifn_atomic_compare_exchange): Likewise.
13132 (expand_builtin_atomic_clear): Likewise.
13133 (expand_builtin_atomic_test_and_set): Likewise.
13134 (fold_builtin_atomic_always_lock_free): Likewise.
13135 * calls.c (compute_argument_addresses): Likewise.
13136 (emit_library_call_value_1): Likewise.
13137 (store_one_arg): Likewise.
13138 * combine.c (combine_instructions): Likewise.
13139 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13140 * config/arm/arm.c (arm_function_value): Likewise.
13141 (aapcs_allocate_return_reg): Likewise.
13142 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13143 * config/i386/i386.c (construct_container): Likewise.
13144 (ix86_gimplify_va_arg): Likewise.
13145 (ix86_expand_sse_cmp): Likewise.
13146 (emit_memmov): Likewise.
13147 (emit_memset): Likewise.
13148 (expand_small_movmem_or_setmem): Likewise.
13149 (ix86_expand_pextr): Likewise.
13150 (ix86_expand_pinsr): Likewise.
13151 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13152 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13154 * config/mips/mips.c (mips_function_value_1) Likewise.
13155 (mips_block_move_straight): Likewise.
13156 (mips_expand_ins_as_unaligned_store): Likewise.
13157 * config/powerpcspe/powerpcspe.c
13158 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13159 (rs6000_darwin64_record_arg_flush): Likewise.
13160 * config/rs6000/rs6000.c
13161 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13162 (rs6000_darwin64_record_arg_flush): Likewise.
13163 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13164 (sparc_function_value_1): Likewise.
13165 * config/spu/spu.c (adjust_operand): Likewise.
13166 (spu_emit_branch_or_set): Likewise.
13167 (arith_immediate_p): Likewise.
13168 * emit-rtl.c (gen_lowpart_common): Likewise.
13169 * expr.c (expand_expr_real_1): Likewise.
13170 * function.c (assign_parm_setup_block): Likewise.
13171 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13172 * reload1.c (alter_reg): Likewise.
13173 * stor-layout.c (mode_for_vector): Likewise.
13174 (layout_type): Likewise.
13176 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13178 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13179 (spu_convert_move): Likewise.
13180 * lower-subreg.c (resolve_simple_move): Likewise.
13182 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13185 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13186 define_insn to a define_expand.
13187 (altivec_vsum2sws_direct): New define_insn.
13188 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13190 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13192 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13195 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13200 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13201 realign stack if stack alignment needed is less than incoming
13204 2017-09-05 Marek Polacek <polacek@redhat.com>
13207 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13210 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13212 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13214 2017-09-05 Richard Biener <rguenther@suse.de>
13216 PR tree-optimization/82084
13217 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13219 2017-09-05 Richard Biener <rguenther@suse.de>
13221 PR tree-optimization/82102
13222 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13224 2017-09-05 Martin Liska <mliska@suse.cz>
13226 PR tree-optimization/82032
13227 * tree-cfg.c (generate_range_test): New function.
13228 * tree-cfg.h (generate_range_test): Declared here.
13229 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13230 (cleanup_control_expr_graph): Use it.
13231 * tree-switch-conversion.c (try_switch_expansion): Remove
13233 (emit_case_nodes): Use generate_range_test.
13235 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13238 * config/i386/i386.md (*<btsc><mode>_mask): Add
13239 TARGET_USE_BT to insn constraint.
13240 (*btr<mode>_mask): Ditto.
13242 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13244 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13245 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13247 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13250 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13251 TARGET_NEON and TARGET_IWMMXT.
13252 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13253 TARGET_NEON and TARGET_IWMMXT.
13254 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13256 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13258 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13259 (ix86_rewrite_tls_address): Ditto.
13260 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13261 (ix86_rewrite_tls_address_1): Ditto.
13262 (ix86_rewrite_tls_address): Ditto.
13263 * config/i386/predicates.md (tls_address_pattern): New predicate.
13264 * config/i386/i386.md (TLS address splitter): New splitter.
13266 2017-09-04 Richard Biener <rguenther@suse.de>
13268 PR tree-optimization/82084
13269 * fold-const.h (can_native_encode_string_p): Declare.
13270 * fold-const.c (can_native_encode_string_p): Factor out from ...
13271 (native_encode_string): ... here.
13272 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13273 vectorizing stores from constants we later cannot handle.
13275 2017-09-04 Marek Polacek <polacek@redhat.com>
13278 * doc/invoke.texi: Update -Wtautological-compare documentation.
13280 2017-09-04 Jeff Law <law@redhat.com>
13282 PR tree-optimization/64910
13283 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13284 swap the first and last operand if the last is a constant.
13286 2017-09-04 Marek Polacek <polacek@redhat.com>
13289 * convert.c (do_narrow): When sanitizing signed integer overflows,
13290 bail out for signed types.
13291 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13293 2017-09-04 Richard Biener <rguenther@suse.de>
13295 PR tree-optimization/82060
13296 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13297 Move devirtualization after stmt folding and before EH/AB/noreturn
13298 cleanup to get the stmt refs canonicalized. Use a bool instead
13299 of gimple_modified_p since that doesn't work for NOPs. Schedule
13300 NOPs generated by folding for removal.
13302 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13303 Alan Hayward <alan.hayward@arm.com>
13304 David Sherwood <david.sherwood@arm.com>
13306 * coretypes.h (pad_direction): New enum.
13307 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13308 (FUNCTION_ARG_PADDING): Likewise.
13309 * target.def (function_arg_padding): New hook.
13310 * targhooks.h (default_function_arg_padding): Declare.
13311 * targhooks.c (default_function_arg_padding): New function.
13312 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13313 (TARGET_FUNCTION_ARG_PADDING): ...this.
13314 * doc/tm.texi: Regenerate.
13315 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13316 instead of direction.
13317 (compute_argument_addresses): Likewise.
13318 (load_register_parameters): Likewise.
13319 (emit_library_call_value_1): Likewise.
13320 (store_one_arg): Use targetm.calls.function_arg_padding instead
13321 of FUNCTION_ARG_PADDING.
13322 (must_pass_in_stack_var_size_or_pad): Likewise.
13323 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13324 (emit_group_store): Likewise.
13325 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13326 instead of FUNCTION_ARG_PADDING.
13327 (emit_push_insn): Likewise, and propagate enum change throughout
13329 * function.h (direction): Delete.
13330 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13332 * function.c (assign_parm_find_stack_rtl): Likewise.
13333 (assign_parm_setup_block_p): Likewise.
13334 (assign_parm_setup_block): Likewise.
13335 (gimplify_parameters): Likewise.
13336 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13337 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13339 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13340 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13341 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13342 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13343 (aarch64_function_arg_padding): ...this new function.
13344 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13345 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13346 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13347 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13348 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13349 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13350 (arm_pad_arg_upward): Replace with...
13351 (arm_function_arg_padding): ...this new function.
13352 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13354 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13355 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13356 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13357 (ia64_hpux_function_arg_padding): Replace with...
13358 (ia64_function_arg_padding): ...this new function. Use pad_direction
13359 instead of direction. Check for TARGET_HPUX.
13360 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13361 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13362 (iq2000_function_arg_padding): New function.
13363 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13364 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13365 (mips_function_arg_padding): ...this new function.
13366 (mips_pad_reg_upward): Update accordingly.
13367 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13368 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13369 targetm.calls.function_arg_padding.
13370 (FUNCTION_ARG_PADDING): Delete.
13371 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13372 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13373 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13374 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13375 (nios2_block_reg_padding): Return pad_direction instead of direction.
13376 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13377 instead of direction.
13378 (nios2_function_arg_padding): Likewise. Make static.
13379 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13380 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13381 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13382 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13383 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13384 (pa_function_arg_padding): Make static. Return pad_direction instead
13386 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13387 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13388 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13389 instead of direction. Use targetm.calls.function_arg_padding.
13390 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13391 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13392 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13393 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13394 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13396 (function_arg_padding): Rename to...
13397 (rs6000_function_arg_padding): ...this. Make static. Return
13398 pad_direction instead of direction.
13399 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13400 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13401 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13402 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13403 instead of direction. Use targetm.calls.function_arg_padding.
13404 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13405 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13406 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13407 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13408 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13409 (function_arg_padding): Rename to...
13410 (rs6000_function_arg_padding): ...this. Make static. Return
13411 pad_direction instead of direction.
13412 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13413 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13414 * config/s390/s390.c (s390_function_arg_padding): New function.
13415 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13416 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13417 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13418 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13419 (function_arg_padding): Rename to...
13420 (sparc_function_arg_padding): ...this. Make static. Return
13421 pad_direction instead of direction.
13422 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13423 * config/spu/spu.c (spu_function_arg_padding): New function.
13424 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13425 * system.h (FUNCTION_ARG_PADDING): Poison.
13427 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13428 Alan Hayward <alan.hayward@arm.com>
13429 David Sherwood <david.sherwood@arm.com>
13431 * target.def (modes_tieable_p): New hook.
13432 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13433 (TARGET_MODES_TIEABLE_P): ...this.
13434 * doc/tm.texi.in: Regenerate.
13435 * hooks.h (hook_bool_mode_mode_true): Declare.
13436 * hooks.c (hook_bool_mode_mode_true): New function.
13437 * combine.c (subst): Use targetm.modes_tieable_p instead of
13439 * dse.c (find_shift_sequence): Likewise.
13440 * expmed.c (extract_low_bits): Likewise.
13441 * lower-subreg.c: Include target.h.
13442 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13444 * rtlanal.c (rtx_cost): Likewise.
13445 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13446 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13447 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13448 (TARGET_MODES_TIEABLE_P): Redefine.
13449 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13450 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13451 (TARGET_MODES_TIEABLE_P): Redefine.
13452 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13453 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13454 (arc_modes_tieable_p): New function.
13455 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13456 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13457 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13458 (arm_modes_tieable_p): Make static.
13459 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13460 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13461 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13462 (TARGET_MODES_TIEABLE_P): Redefine.
13463 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13464 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13465 (TARGET_MODES_TIEABLE_P): Redefine.
13466 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13467 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13468 (cr16_modes_tieable_p): New function.
13469 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13470 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13471 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13472 (TRULY_NOOP_TRUNCATION): Update comment.
13473 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13474 (TRULY_NOOP_TRUNCATION): Update comment.
13475 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13476 (frv_modes_tieable_p): New function.
13477 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13478 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13479 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13480 (TARGET_MODES_TIEABLE_P): Redefine.
13481 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13482 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13483 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13484 (TARGET_MODES_TIEABLE_P): Redefine.
13485 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13486 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13487 (ia64_modes_tieable_p): New function.
13488 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13489 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13490 (iq2000_modes_tieable_p): New function.
13491 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13492 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13493 (lm32_modes_tieable_p): New function.
13494 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13495 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13496 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13497 (TARGET_MODES_TIEABLE_P): Redefine.
13498 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13499 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13500 (m32r_modes_tieable_p): New function.
13501 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13502 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13503 (m68k_modes_tieable_p): New function.
13504 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13505 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13506 (mcore_modes_tieable_p): New function.
13507 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13508 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13510 (TARGET_MODES_TIEABLE_P): Redefine.
13511 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13512 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13513 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13514 (TARGET_MODES_TIEABLE_P): Redefine.
13515 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13516 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13517 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13518 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13519 (mn10300_modes_tieable_p): ...this and make static.
13520 (TARGET_MODES_TIEABLE_P): Redefine.
13521 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13522 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13523 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13524 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13525 (msp430_modes_tieable_p): Make static.
13526 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13527 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13528 (TARGET_MODES_TIEABLE_P): Redefine.
13529 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13530 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13531 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13532 (TARGET_MODES_TIEABLE_P): Redefine.
13533 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13534 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13535 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13536 (TARGET_MODES_TIEABLE_P): Redefine.
13537 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13538 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13539 (pdp11_modes_tieable_p): New function.
13540 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13541 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13542 (rs6000_modes_tieable_p): New function.
13543 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13544 * config/powerpcspe/powerpcspe.md: Update comment.
13545 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13546 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13547 (TARGET_MODES_TIEABLE_P): Redefine.
13548 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13549 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13550 (rl78_modes_tieable_p): New function.
13551 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13552 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13553 (rs6000_modes_tieable_p): New function.
13554 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13555 * config/rs6000/rs6000.md: Update comment.
13556 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13557 * config/rx/rx.c (rx_modes_tieable_p): New function.
13558 (TARGET_MODES_TIEABLE_P): Redefine.
13559 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13560 * config/s390/s390.c (s390_modes_tieable_p): New function.
13561 (TARGET_MODES_TIEABLE_P): Redefine.
13562 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13563 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13564 (sh_modes_tieable_p): New function.
13565 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13566 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13567 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13568 (sparc_modes_tieable_p): Make static.
13569 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13570 * config/spu/spu.c (spu_modes_tieable_p): New function.
13571 (TARGET_MODES_TIEABLE_P): Redefine.
13572 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13573 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13574 (TARGET_MODES_TIEABLE_P): Redefine.
13575 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13576 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13577 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13578 * config/v850/v850.c (v850_modes_tieable_p): New function.
13579 (TARGET_MODES_TIEABLE_P): Redefine.
13580 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13581 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13582 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13583 (visium_modes_tieable_p): New function.
13584 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13585 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13586 (xtensa_modes_tieable_p): New function.
13587 * system.h (MODES_TIEABLE_P): Poison.
13589 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13590 Alan Hayward <alan.hayward@arm.com>
13591 David Sherwood <david.sherwood@arm.com>
13593 * target.def (hard_regno_mode_ok): New hook.
13594 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
13595 (TARGET_HARD_REGNO_MODE_OK): ...this.
13596 * doc/tm.texi.in: Regenerate.
13597 * hooks.h (hook_bool_uint_mode_true): Declare.
13598 * hooks.c (hook_bool_uint_mode_true): New function.
13599 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
13600 HARD_REGNO_MODE_OK.
13601 * genpreds.c (write_insn_preds_c): Add an include of target.h.
13602 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
13603 instead of HARD_REGNO_MODE_OK.
13604 * caller-save.c: Include target.h.
13605 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
13606 HARD_REGNO_MODE_OK.
13607 * combine.c (can_combine_p): Likewise.
13608 (combinable_i3pat): Likewise.
13609 (can_change_dest_mode): Likewise.
13610 * expr.c (init_expr_target): Likewise.
13611 (convert_move): Likewise.
13612 (convert_modes): Likewise.
13613 * ira.c (setup_prohibited_class_mode_regs): Likewise.
13614 (setup_prohibited_mode_move_regs): Likewise.
13615 * ira.h (target_ira): Likewise.
13616 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13617 * lra-constraints.c (process_alt_operands): Likewise.
13618 (split_reg): Likewise.
13619 * recog.c (peep2_find_free_register): Likewise.
13620 * ree.c (combine_reaching_defs): Likewise.
13621 * regcprop.c (maybe_mode_change): Likewise.
13622 * reginfo.c (init_reg_sets_1): Likewise.
13623 (choose_hard_reg_mode): Likewise.
13624 (simplifiable_subregs): Likewise.
13625 * regrename.c (check_new_reg_p): Likewise.
13626 * reload.c (find_valid_class): Likewise.
13627 (find_valid_class_1): Likewise.
13628 (reload_inner_reg_of_subreg): Likewise.
13629 (push_reload): Likewise.
13630 (combine_reloads): Likewise.
13631 (find_dummy_reload): Likewise.
13632 (find_reloads): Likewise.
13633 * reload1.c (find_reg): Likewise.
13634 (set_reload_reg): Likewise.
13635 (allocate_reload_reg): Likewise.
13636 (choose_reload_regs): Likewise.
13637 (reload_adjust_reg_for_temp): Likewise.
13638 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
13639 (simplify_subreg_regno): Likewise.
13640 * sel-sched.c (init_regs_for_mode): Likewise.
13641 * varasm.c (make_decl_rtl): Likewise.
13642 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
13643 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
13644 HARD_REGNO_MODE_OK.
13645 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
13646 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
13647 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13648 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
13649 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
13650 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13651 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
13652 (arc_mode_class): Delete.
13653 (HARD_REGNO_MODE_OK): Delete.
13654 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13655 (arc_hard_regno_mode_ok): Rename old array to...
13656 (arc_hard_regno_mode_ok_modes): ...this.
13657 (arc_conditional_register_usage): Update accordingly.
13658 (arc_mode_class): Make static.
13659 (arc_hard_regno_mode_ok): New function.
13660 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
13661 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
13662 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13663 (arm_hard_regno_mode_ok): Make static.
13664 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
13665 HARD_REGNO_MODE_OK.
13666 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
13667 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
13668 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
13670 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13671 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
13672 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
13673 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
13674 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
13675 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13676 * config/bfin/predicates.md (valid_reg_operand): Use
13677 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
13678 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
13679 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
13680 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13681 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
13682 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
13683 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13684 (cr16_hard_regno_mode_ok): Make static and return a bool.
13685 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
13686 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13687 (cris_hard_regno_mode_ok): New function.
13688 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
13689 (epiphany_mode_class): Delete.
13690 (HARD_REGNO_MODE_OK): Delete.
13691 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
13692 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13693 (hard_regno_mode_ok): Rename to...
13694 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
13695 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
13696 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
13697 HARD_REGNO_MODE_OK.
13698 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
13699 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
13700 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13701 (frv_hard_regno_mode_ok): Make static and return a bool.
13702 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
13703 HARD_REGNO_MODE_OK.
13704 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
13705 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
13706 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
13707 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
13709 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13710 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
13711 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
13712 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
13714 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13715 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
13716 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13717 (ia64_hard_regno_mode_ok): New function.
13718 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
13719 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13720 (iq2000_hard_regno_mode_ok): New function.
13721 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
13722 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13723 (lm32_hard_regno_mode_ok): New function.
13724 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
13725 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
13726 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
13727 instead of HARD_REGNO_MODE_OK.
13728 (m32c_hard_regno_ok): Rename to...
13729 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
13730 (m32c_cannot_change_mode_class): Update accordingly.
13731 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13732 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
13733 (m32r_mode_class): Delete.
13734 (HARD_REGNO_MODE_OK): Delete.
13735 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13736 (m32r_hard_regno_mode_ok): Rename to...
13737 (m32r_hard_regno_modes): ...this.
13738 (m32r_mode_class): Make static.
13739 (m32r_hard_regno_mode_ok): New function.
13740 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
13741 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
13742 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13743 (m68k_hard_regno_mode_ok): Make static.
13744 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
13745 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13746 (mcore_hard_regno_mode_ok): New function.
13747 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
13748 (HARD_REGNO_MODE_OK): Delete.
13749 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
13751 (microblaze_hard_regno_mode_ok_p): ...this and make static.
13752 (microblaze_hard_regno_mode_ok): New function.
13753 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13754 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
13755 (mips_hard_regno_mode_ok): Delete.
13756 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
13757 (mips_hard_regno_mode_ok_p): ...this and make static.
13758 (mips_hard_regno_mode_ok_p): Rename to...
13759 (mips_hard_regno_mode_ok_uncached): ...this.
13760 (mips_hard_regno_mode_ok): New function.
13761 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
13762 of HARD_REGNO_MODE_OK.
13763 (mips_option_override): Update after above name changes.
13764 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13765 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
13766 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
13767 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
13768 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
13769 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13770 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
13771 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
13772 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
13773 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13774 (msp430_hard_regno_mode_ok): Make static and return a bool.
13775 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
13776 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
13777 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
13779 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13780 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
13781 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
13782 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
13783 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
13784 (PA_HARD_REGNO_MODE_OK): ...this
13785 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
13786 (PA_HARD_REGNO_MODE_OK): ...this.
13787 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13788 (pa_hard_regno_mode_ok): New function.
13789 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
13790 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13791 (pdp11_hard_regno_mode_ok): New function.
13792 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
13793 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
13795 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
13797 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13798 (rs6000_hard_regno_mode_ok): Rename to...
13799 (rs6000_hard_regno_mode_ok_uncached): ...this.
13800 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13801 (rs6000_hard_regno_mode_ok): New function.
13802 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
13803 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
13804 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
13805 (riscv_hard_regno_mode_ok): ...this and make static.
13806 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13807 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
13808 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
13809 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13810 (rl78_hard_regno_mode_ok): Make static and return bool.
13811 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
13812 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
13814 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
13815 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13816 (rs6000_hard_regno_mode_ok): Rename to...
13817 (rs6000_hard_regno_mode_ok_uncached): ...this.
13818 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13819 (rs6000_hard_regno_mode_ok): New function.
13820 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
13821 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
13822 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13823 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
13824 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
13825 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
13826 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13827 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
13828 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
13829 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13830 (sh_hard_regno_mode_ok): Make static.
13831 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
13832 instead of HARD_REGNO_MODE_OK.
13833 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
13834 (sparc_mode_class): Delete.
13835 (HARD_REGNO_MODE_OK): Delete.
13836 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13837 (hard_regno_mode_classes): Make static.
13838 (sparc_mode_class): Likewise.
13839 (sparc_hard_regno_mode_ok): New function.
13840 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
13841 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
13842 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
13844 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13845 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
13846 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
13847 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
13848 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
13849 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13850 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
13851 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
13852 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13853 (visium_hard_regno_mode_ok): New function.
13854 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
13855 instead of HARD_REGNO_MODE_OK.
13856 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
13857 (HARD_REGNO_MODE_OK): Delete.
13858 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
13859 (xtensa_hard_regno_mode_ok_p): ...this and make static.
13860 (xtensa_option_override): Update accordingly.
13861 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13862 (xtensa_hard_regno_mode_ok): New function.
13863 * system.h (HARD_REGNO_MODE_OK): Poison.
13865 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13866 Alan Hayward <alan.hayward@arm.com>
13867 David Sherwood <david.sherwood@arm.com>
13869 * target.def (hard_regno_call_part_clobbered): New hook.
13870 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
13871 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
13872 * doc/tm.texi: Regenerate.
13873 * hooks.h (hook_bool_uint_mode_false): Declare.
13874 * hooks.c (hook_bool_uint_mode_false): New function.
13875 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13876 * cselib.c (cselib_process_insn): Use
13877 targetm.hard_regno_call_part_clobbered instead of
13878 HARD_REGNO_CALL_PART_CLOBBERED.
13879 * ira-conflicts.c (ira_build_conflicts): Likewise.
13880 * ira-costs.c (ira_tune_allocno_costs): Likewise.
13881 * lra-constraints.c (need_for_call_save_p): Likewise.
13882 * lra-lives.c: Include target.h.
13883 (check_pseudos_live_through_calls): Use
13884 targetm.hard_regno_call_part_clobbered instead of
13885 HARD_REGNO_CALL_PART_CLOBBERED.
13886 * regcprop.c: Include target.h.
13887 (copyprop_hardreg_forward_1): Use
13888 targetm.hard_regno_call_part_clobbered instead of
13889 HARD_REGNO_CALL_PART_CLOBBERED.
13890 * reginfo.c (choose_hard_reg_mode): Likewise.
13891 * regrename.c (check_new_reg_p): Likewise.
13892 * reload.c (find_equiv_reg): Likewise.
13893 * reload1.c (emit_reload_insns): Likewise.
13894 * sched-deps.c (deps_analyze_insn): Likewise.
13895 * sel-sched.c (init_regs_for_mode): Likewise.
13896 (mark_unavailable_hard_regs): Likewise.
13897 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
13898 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13899 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
13901 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13902 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13903 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
13905 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
13907 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13908 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13909 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
13911 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13912 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13913 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
13915 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13916 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
13918 * config/powerpcspe/powerpcspe.c
13919 (rs6000_hard_regno_call_part_clobbered): New function.
13920 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13921 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13922 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
13924 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13925 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13926 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
13928 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13929 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13930 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
13932 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13933 Alan Hayward <alan.hayward@arm.com>
13934 David Sherwood <david.sherwood@arm.com>
13936 * rtl.h (subreg_memory_offset): Declare.
13937 * emit-rtl.c (subreg_memory_offset): New function.
13938 * expmed.c (store_bit_field_1): Use it.
13939 * expr.c (undefined_operand_subword_p): Likewise.
13940 * simplify-rtx.c (simplify_subreg): Likewise.
13942 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
13944 PR rtl-optimization/57448
13947 * optabs.c (expand_atomic_load): Place compiler memory barriers if
13948 using atomic_load pattern.
13949 (expand_atomic_store): Likewise.
13951 2017-09-04 Jakub Jelinek <jakub@redhat.com>
13954 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
13955 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
13956 handling. Use replace_call_with_value with NULL instead of
13957 gsi_replace, unlink_stmt_vdef and release_defs.
13959 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
13962 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
13964 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13967 * rtl.h (emit_library_call_value_1): Declare.
13968 (emit_library_call): Replace declaration with a series of overloads.
13969 Remove the parameter count argument.
13970 (emit_library_call_value): Likewise.
13971 * calls.c (emit_library_call_value_1): Make global. Replace varargs
13972 with an "rtx_mode_t *".
13973 (emit_library_call_value): Delete.
13974 (emit_library_call): Likewise.
13975 * asan.c (asan_emit_stack_protection): Update calls accordingly.
13976 (asan_emit_allocas_unpoison): Likewise.
13977 * builtins.c (expand_builtin_powi): Likewise.
13978 (expand_asan_emit_allocas_unpoison): Likewise.
13979 * cfgexpand.c (expand_main_function): Likewise.
13980 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
13981 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
13982 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
13983 * config/arm/arm.c (arm_trampoline_init): Likewise.
13984 (arm_call_tls_get_addr): Likewise.
13985 (arm_expand_divmod_libfunc): Likewise.
13986 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
13987 (smulsi3_highpart): Likewise.
13988 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
13989 (c6x_expand_compare): Likewise.
13990 (c6x_expand_movmem): Likewise.
13991 * config/frv/frv.c (frv_trampoline_init): Likewise.
13992 * config/i386/i386.c (ix86_trampoline_init): Likewise.
13993 (ix86_expand_divmod_libfunc): Likewise.
13994 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
13995 (ia64_expand_compare): Likewise.
13996 (ia64_profile_hook): Likewise.
13997 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
13998 (nonlocal_goto): Likewise.
13999 (restore_stack_nonlocal): Likewise.
14000 * config/m32r/m32r.c (block_move_call): Likewise.
14001 (m32r_trampoline_init): Likewise.
14002 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14003 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14004 (m68k_call_m68k_read_tp): Likewise.
14005 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14006 (microblaze_expand_divide): Likewise.
14007 * config/mips/mips.h (mips_args): Likewise.
14008 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14009 (MIPS_ICACHE_SYNC): Likewise.
14010 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14011 (nios2_trampoline_init): Likewise.
14012 * config/pa/pa.c (hppa_tls_call): Likewise.
14013 (pa_trampoline_init): Likewise.
14014 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14015 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14016 (expand_strn_compare): Likewise.
14017 (rs6000_generate_compare): Likewise.
14018 (rs6000_expand_float128_convert): Likewise.
14019 (output_profile_hook): Likewise.
14020 (rs6000_trampoline_init): Likewise.
14021 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14022 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14023 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14024 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14025 (rs6000_generate_compare): Likewise.
14026 (rs6000_expand_float128_convert): Likewise.
14027 (output_profile_hook): Likewise.
14028 (rs6000_trampoline_init): Likewise.
14029 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14030 * config/sh/sh.c (sh_trampoline_init): Likewise.
14031 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14032 (sparc_emit_float_lib_cmp): Likewise.
14033 (sparc32_initialize_trampoline): Likewise.
14034 (sparc64_initialize_trampoline): Likewise.
14035 (sparc_profile_hook): Likewise.
14036 * config/spu/spu.c (ea_load_store): Likewise.
14037 * config/spu/spu.md (floatunssidf2): Likewise.
14038 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14039 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14040 * config/visium/visium.c (expand_block_move_4): Likewise.
14041 (expand_block_move_2): Likewise.
14042 (expand_block_move_1): Likewise.
14043 (expand_block_set_4): Likewise.
14044 (expand_block_set_2): Likewise.
14045 (expand_block_set_1): Likewise.
14046 (visium_trampoline_init): Likewise.
14047 (visium_profile_hook): Likewise.
14048 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14049 (xtensa_setup_frame_addresses): Likewise.
14050 (xtensa_trampoline_init): Likewise.
14051 * except.c (sjlj_emit_function_enter): Likewise.
14052 (sjlj_emit_function_exit): Likewise.
14053 * explow.c (allocate_dynamic_stack_space): Likewise.
14054 (probe_stack_range): Likewise.
14055 * expr.c (convert_mode_scalar): Likewise.
14056 * optabs.c (expand_binop): Likewise.
14057 (expand_twoval_binop_libfunc): Likewise.
14058 (expand_unop): Likewise.
14059 (prepare_cmp_insn): Likewise.
14060 (prepare_float_lib_cmp): Likewise.
14061 (expand_float): Likewise.
14062 (expand_fix): Likewise.
14063 (expand_fixed_convert): Likewise.
14064 (maybe_emit_sync_lock_test_and_set): Likewise.
14065 (expand_atomic_compare_and_swap): Likewise.
14066 (expand_mem_thread_fence): Likewise.
14067 (expand_atomic_fetch_op): Likewise.
14069 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14071 * doc/generic.texi (OpenACC): Adjust URL.
14072 * doc/invoke.texi (C Dialect Options): Ditto.
14074 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14076 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14077 predicate for operand 1. Add (m,<S>) constraint.
14078 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14079 Prevent memory operand 1 with register operand 2.
14081 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14083 PR rtl-optimization/82024
14084 * combine.c (try_combine): If the combination result is a PARALLEL,
14085 and we only need to retain the SET in there that would be placed
14086 at I2, check that we can place that at I3 instead, before doing so.
14088 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14091 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14093 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14094 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14097 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14098 Jeff Law <law@redhat.com>
14100 * varasm.c (bss_initializer_p): Do not put constants into .bss
14101 (categorize_decl_for_section): Handle bss_initializer_p returning
14102 false when DECL_INITIAL is NULL.
14104 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14107 * config/s390/s390.c (s390_can_inline_p): New function.
14109 2017-09-01 Jeff Law <law@redhat.com>
14111 PR tree-optimization/82052
14112 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14113 Always initialize the returned slot after a hash table miss
14114 when INSERT is true.
14116 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14118 * config/s390/s390.md (mem_signal_fence): Remove.
14119 * doc/md.texi (mem_signal_fence): Remove.
14120 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14122 * target-insns.def (mem_signal_fence): Remove.
14124 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14127 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14130 * asan.c (create_odr_indicator): Strip name encoding from assembler
14131 name before appending it after __odr_asan_.
14133 2017-09-01 Martin Liska <mliska@suse.cz>
14135 PR tree-optimization/82059
14136 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14137 frequency only when an edge is redirected.
14139 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14141 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14142 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14143 (arc_conditional_register_usage): Remove ARC600 lp_count
14145 (arc_file_start): Emit Tag_ARC_CPU_variation.
14146 (arc_can_use_doloop_p): New conditions to use ZOLs.
14147 (hwloop_fail): New function.
14148 (hwloop_optimize): Likewise.
14149 (hwloop_pattern_reg): Likewise.
14150 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14151 (arc_reorg_loops): New function, calls reorg_loops.
14152 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14153 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14155 (arc_loop_hazard): Remove function, functionality moved into
14157 (arc_hazard): Remove arc_loop_hazard call.
14158 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14159 into hwloop_optimize.
14160 (arc_label_align): Remove ZOL handling.
14161 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14162 * config/arc/arc.md (doloop_begin): Remove pattern.
14163 (doloop_begin_i): Likewise.
14164 (doloop_end_i): Likewise.
14165 (doloop_fallback): Likewise.
14166 (doloop_fallback_m): Likewise.
14167 (doloop_end): Reimplement expand.
14168 (arc_lp): New pattern for LP instruction.
14169 (loop_end): New pattern.
14170 (loop_fail): Likewise.
14171 (decrement_and_branch_until_zero): Likewise.
14172 * config/arc/arc.opt (mlpc-width): New option.
14173 * doc/invoke.texi (mlpc-width): Document option.
14175 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14177 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14178 (arc_ccfsm_advance): Fix checking for delay slots.
14179 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14181 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14183 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14185 (movhi_insn): Update store instruction constraint which are saving
14186 6-bit short immediates.
14187 (movsi_insn): Consider also short scaled load operations.
14188 (zero_extendhisi2_i): Use Usd constraint instead of T.
14189 (extendhisi2_i): Add q constraint.
14190 (arc_clzsi2): Add type and length attributes.
14191 (arc_ctzsi2): Likewise.
14192 * config/arc/constraints.md (Usc): Update constraint, the
14193 assembler can parse two relocations for a single instruction.
14195 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14197 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14198 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14200 2017-08-31 Olivier Hainque <hainque@adacore.com>
14202 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14203 match as powerpc-wrs-vxworks*.
14205 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14207 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14208 register constraint for by-element operand.
14209 (aarch64_mls_elt_merge<mode>): Likewise.
14211 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14213 * config/arc/arc.c (arc_can_follow_jump): Check for short
14216 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14218 * config.gcc: Use g.opt for arc.
14219 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14220 functionality moved to ...
14221 (legitimate_scaled_address_p): New function, ...here.
14222 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14223 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14224 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14226 (arc_override_options): Handle G option.
14227 (arc_output_pic_addr_const): Correct function definition.
14228 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14229 (arc_decl_anon_ns_mem_p): Delete.
14230 (arc_in_small_data_p): Overhaul this function to take into
14231 consideration the value given via G option.
14232 (arc_rewrite_small_data_1): Renamed and corrected old
14233 arc_rewrite_small_data function.
14234 (arc_rewrite_small_data): New function.
14235 (small_data_pattern): Don't use pic_offset_table_rtx.
14236 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14237 * config/arc/simdext.md (movmisalignv2hi): Use
14238 prepare_move_operands function.
14240 (movmisalign*): Likewise.
14241 * doc/invoke.texi (ARC options): Document -G option.
14243 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14245 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14247 * config/arc/arc.c (arc_print_operand): Output scalled address for
14248 sdata whenever is possible.
14249 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14250 load/stores are available.
14251 (compact_sda_memory_operand): Check for the alignment required by
14252 code density instructions.
14253 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14255 * config/arc/constraints.md (Usd): Update constraint.
14256 (Us0): New constraint.
14257 (Usc): Update constraint.
14259 2017-08-31 Richard Biener <rguenther@suse.de>
14261 PR middle-end/82054
14262 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14263 function only once.
14265 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14267 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14268 Resize type_signature.
14270 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14271 Alan Hayward <alan.hayward@arm.com>
14272 David Sherwood <david.sherwood@arm.com>
14274 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14275 subregs whose inner modes can be stored in GPRs.
14276 (aarch64_classify_index): Likewise.
14278 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14279 Alan Hayward <alan.hayward@arm.com>
14280 David Sherwood <david.sherwood@arm.com>
14282 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14283 (V_INT_EQUIV): ...this.
14284 (v_cmp_result): Rename to...
14285 (v_int_equiv): ...this.
14286 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14287 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14288 (copysign<mode>3): Likewise.
14289 (aarch64_simd_bsl<mode>_internal): Likewise.
14290 (aarch64_simd_bsl<mode>): Likewise.
14291 (vec_cmp<mode><mode>): Likewise.
14292 (vcond<mode><mode>): Likewise.
14293 (vcond<v_cmp_mixed><mode>): Likewise.
14294 (vcondu<mode><v_cmp_mixed>): Likewise.
14295 (aarch64_cm<optab><mode>): Likewise.
14296 (aarch64_cmtst<mode>): Likewise.
14297 (aarch64_fac<optab><mode>): Likewise.
14298 (vec_perm_const<mode>): Likewise.
14299 (vcond_mask_<mode><v_cmp_result>): Rename to...
14300 (vcond_mask_<mode><v_int_equiv>): ...this.
14301 (vec_cmp<mode><v_cmp_result>): Rename to...
14302 (vec_cmp<mode><v_int_equiv>): ...this.
14304 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14305 Alan Hayward <alan.hayward@arm.com>
14306 David Sherwood <david.sherwood@arm.com>
14308 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14310 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14311 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14312 (UNSPEC_LD4_DREG): New unspecs.
14313 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14314 (aarch64_ld2<mode>_dreg_be): Replace with...
14315 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14317 (aarch64_ld3<mode>_dreg_le)
14318 (aarch64_ld3<mode>_dreg_be): Replace with...
14319 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14321 (aarch64_ld4<mode>_dreg_le)
14322 (aarch64_ld4<mode>_dreg_be): Replace with...
14323 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14326 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14328 PR tree-optimization/81987
14329 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14330 insert an initializer in a location not dominated by the stride
14333 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14335 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14336 on the entire header of the finally block in the fallthru case.
14338 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14340 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14342 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14344 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14345 rs6000_emit_move_from_cr and call renamed function.
14346 (rs6000_emit_prologue): Call renamed functions.
14347 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14348 movesi_from_cr, remove volatile CRs.
14350 2017-08-30 Jon Beniston <jon@beniston.com>
14351 Richard Biener <rguenther@suse.de>
14353 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14354 of VECTOR_MODE_P check.
14355 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14356 element vector types.
14358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14360 * df.h (df_read_modify_subreg_p): Remove in favor of...
14361 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14362 const_rtx instead of an rtx.
14363 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14364 * df-problems.c (df_word_lr_mark_ref): Likewise.
14365 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14366 (mark_pseudo_reg_dead): Likewise.
14367 (mark_ref_dead): Likewise.
14368 * reginfo.c (init_subregs_of_mode): Likewise.
14369 * sched-deps.c (sched_analyze_1): Likewise.
14370 * df-scan.c (df_def_record_1): Likewise.
14371 (df_uses_record): Likewise.
14372 (df_read_modify_subreg_p): Remove in favor of...
14373 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14374 const_rtx instead of an rtx.
14376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14377 Alan Hayward <alan.hayward@arm.com>
14378 David Sherwood <david.sherwood@arm.com>
14380 * rtl.h (partial_subreg_p): New function.
14381 * caller-save.c (save_call_clobbered_regs): Use it.
14382 * calls.c (expand_call): Likewise.
14383 * combine.c (combinable_i3pat): Likewise.
14384 (simplify_set): Likewise.
14385 (make_extraction): Likewise.
14386 (make_compound_operation_int): Likewise.
14387 (gen_lowpart_or_truncate): Likewise.
14388 (force_to_mode): Likewise.
14389 (make_field_assignment): Likewise.
14390 (reg_truncated_to_mode): Likewise.
14391 (record_truncated_value): Likewise.
14392 (move_deaths): Likewise.
14393 * cse.c (record_jump_cond): Likewise.
14394 (cse_insn): Likewise.
14395 * cselib.c (cselib_lookup_1): Likewise.
14396 * expmed.c (extract_bit_field_using_extv): Likewise.
14397 * function.c (assign_parm_setup_reg): Likewise.
14398 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14399 * ira-build.c (create_insn_allocnos): Likewise.
14400 * lra-coalesce.c (merge_pseudos): Likewise.
14401 * lra-constraints.c (match_reload): Likewise.
14402 (simplify_operand_subreg): Likewise.
14403 (curr_insn_transform): Likewise.
14404 * lra-lives.c (process_bb_lives): Likewise.
14405 * lra.c (new_insn_reg): Likewise.
14406 (lra_substitute_pseudo): Likewise.
14407 * regcprop.c (mode_change_ok): Likewise.
14408 (maybe_mode_change): Likewise.
14409 (copyprop_hardreg_forward_1): Likewise.
14410 * reload.c (push_reload): Likewise.
14411 (find_reloads): Likewise.
14412 (find_reloads_subreg_address): Likewise.
14413 * reload1.c (alter_reg): Likewise.
14414 (eliminate_regs_1): Likewise.
14415 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14417 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14419 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14420 back to if statements, including unpack.
14422 2017-08-30 Martin Liska <mliska@suse.cz>
14424 PR inline-asm/82001
14425 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14427 (func_checker::compare_asm_inputs_outputs): ... this function.
14428 (func_checker::compare_gimple_asm): Use the function to compare
14429 also ASM constrains.
14430 * ipa-icf-gimple.h: Rename the function.
14432 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14433 Alan Hayward <alan.hayward@arm.com>
14434 David Sherwood <david.sherwood@arm.com>
14436 * coretypes.h (complex_mode): New type.
14437 * gdbhooks.py (build_pretty_printer): Handle it.
14438 * machmode.h (complex_mode): New class.
14439 (complex_mode::includes_p): New function.
14440 (is_complex_int_mode): Likewise.
14441 (is_complex_float_mode): Likewise.
14442 * genmodes.c (get_mode_class): Handle complex mode classes.
14443 * function.c (expand_function_end): Use is_complex_int_mode.
14445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14446 Alan Hayward <alan.hayward@arm.com>
14447 David Sherwood <david.sherwood@arm.com>
14449 * coretypes.h (scalar_mode_pod): New typedef.
14450 * gdbhooks.py (build_pretty_printer): Handle it.
14451 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14452 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14453 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14454 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14455 as_a <scalar_mode>.
14457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14458 Alan Hayward <alan.hayward@arm.com>
14459 David Sherwood <david.sherwood@arm.com>
14461 * machmode.h (mode_for_vector): Take a scalar_mode instead
14463 * stor-layout.c (mode_for_vector): Likewise.
14464 * explow.c (promote_mode): Use as_a <scalar_mode>.
14465 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14468 Alan Hayward <alan.hayward@arm.com>
14469 David Sherwood <david.sherwood@arm.com>
14471 * target.def (preferred_simd_mode): Take a scalar_mode
14472 instead of a machine_mode.
14473 * targhooks.h (default_preferred_simd_mode): Likewise.
14474 * targhooks.c (default_preferred_simd_mode): Likewise.
14475 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14476 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14477 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14478 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14479 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14480 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14481 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14482 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14484 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14485 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14486 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14487 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14488 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14489 * doc/tm.texi: Regenerate.
14490 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14494 Alan Hayward <alan.hayward@arm.com>
14495 David Sherwood <david.sherwood@arm.com>
14497 * target.def (scalar_mode_supported_p): Take a scalar_mode
14498 instead of a machine_mode.
14499 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14500 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14501 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14502 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14503 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14504 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14505 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14506 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14507 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14508 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14509 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14510 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14511 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14512 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14514 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14515 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14516 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14517 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14518 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14520 * doc/tm.texi: Regenerate.
14522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14523 Alan Hayward <alan.hayward@arm.com>
14524 David Sherwood <david.sherwood@arm.com>
14526 * coretypes.h (opt_scalar_mode): New typedef.
14527 * gdbhooks.py (build_pretty_printers): Handle it.
14528 * machmode.h (mode_iterator::get_2xwider): Add overload for
14530 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14532 * expr.c (convert_mode_scalar): Likewise.
14533 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14534 * optabs.c (expand_float): Likewise.
14535 (expand_fix): Likewise.
14536 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14539 Alan Hayward <alan.hayward@arm.com>
14540 David Sherwood <david.sherwood@arm.com>
14542 * optabs.c (expand_float): Explicitly check for scalars before
14543 using a branching expansion.
14544 (expand_fix): Likewise.
14546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14547 Alan Hayward <alan.hayward@arm.com>
14548 David Sherwood <david.sherwood@arm.com>
14550 * expr.c (convert_mode): Split scalar handling out into...
14551 (convert_mode_scalar): ...this new function. Treat the modes
14554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14555 Alan Hayward <alan.hayward@arm.com>
14556 David Sherwood <david.sherwood@arm.com>
14558 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14560 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14563 Alan Hayward <alan.hayward@arm.com>
14564 David Sherwood <david.sherwood@arm.com>
14566 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14567 rather than a machine_mode.
14568 (fixed_from_string): Likewise.
14569 (fixed_convert): Likewise.
14570 (fixed_convert_from_int): Likewise.
14571 (fixed_convert_from_real): Likewise.
14572 (real_convert_from_fixed): Likewise.
14573 * fixed-value.c (fixed_from_double_int): Likewise.
14574 (fixed_from_string): Likewise.
14575 (fixed_convert): Likewise.
14576 (fixed_convert_from_int): Likewise.
14577 (fixed_convert_from_real): Likewise.
14578 (real_convert_from_fixed): Likewise.
14579 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14581 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14582 Alan Hayward <alan.hayward@arm.com>
14583 David Sherwood <david.sherwood@arm.com>
14585 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14586 of separate mode class checks. Do not allow vector modes here.
14587 (immed_wide_int_const): Use as_a <scalar_mode>.
14588 * explow.c (trunc_int_for_mode): Likewise.
14589 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14590 (wi::shwi): Likewise.
14591 (wi::min_value): Likewise.
14592 (wi::max_value): Likewise.
14593 * dwarf2out.c (loc_descriptor): Likewise.
14594 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
14595 for CONST_WIDE_INT.
14597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14598 Alan Hayward <alan.hayward@arm.com>
14599 David Sherwood <david.sherwood@arm.com>
14601 * tree.h (SCALAR_TYPE_MODE): New macro.
14602 * expr.c (expand_expr_addr_expr_1): Use it.
14603 (expand_expr_real_2): Likewise.
14604 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
14605 (fold_convert_const_fixed_from_int): Likewise.
14606 (fold_convert_const_fixed_from_real): Likewise.
14607 (native_encode_fixed): Likewise
14608 (native_encode_complex): Likewise
14609 (native_encode_vector): Likewise.
14610 (native_interpret_fixed): Likewise.
14611 (native_interpret_real): Likewise.
14612 (native_interpret_complex): Likewise.
14613 (native_interpret_vector): Likewise.
14614 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
14615 (simd_clone_adjust_argument_types): Likewise.
14616 (simd_clone_init_simd_arrays): Likewise.
14617 (simd_clone_adjust): Likewise.
14618 * stor-layout.c (layout_type): Likewise.
14619 * tree.c (build_minus_one_cst): Likewise.
14620 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14621 * tree-inline.c (estimate_move_cost): Likewise.
14622 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
14623 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
14624 (vectorizable_reduction): Likewise.
14625 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
14626 (vect_recog_mixed_size_cond_pattern): Likewise.
14627 (check_bool_pattern): Likewise.
14628 (adjust_bool_pattern): Likewise.
14629 (search_type_for_mask_1): Likewise.
14630 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
14631 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14632 (vectorizable_load): Likewise.
14633 (vectorizable_store): Likewise.
14634 * ubsan.c (ubsan_encode_value): Likewise.
14635 * varasm.c (output_constant): Likewise.
14637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14638 Alan Hayward <alan.hayward@arm.com>
14639 David Sherwood <david.sherwood@arm.com>
14641 * coretypes.h (scalar_mode): New class.
14642 * machmode.h (scalar_mode): Likewise.
14643 (scalar_mode::includes_p): New function.
14644 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
14645 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
14646 * genmodes.c (get_mode_class): Handle remaining scalar modes.
14647 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
14648 * expmed.c (store_bit_field_1): Likewise.
14649 (extract_bit_field_1): Likewise.
14650 * expr.c (write_complex_part): Likewise.
14651 (read_complex_part): Likewise.
14652 (emit_move_complex_push): Likewise.
14653 (expand_expr_real_2): Likewise.
14654 * function.c (assign_parm_setup_reg): Likewise.
14655 (assign_parms_unsplit_complex): Likewise.
14656 * optabs.c (expand_binop): Likewise.
14657 * rtlanal.c (subreg_get_info): Likewise.
14658 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14659 * varasm.c (output_constant_pool_2): Likewise.
14661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14662 Alan Hayward <alan.hayward@arm.com>
14663 David Sherwood <david.sherwood@arm.com>
14665 * expmed.c (extract_high_half): Use scalar_int_mode and remove
14667 (expmed_mult_highpart_optab): Likewise.
14668 (expmed_mult_highpart): Likewise.
14670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14671 Alan Hayward <alan.hayward@arm.com>
14672 David Sherwood <david.sherwood@arm.com>
14674 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
14675 instead of a machine_mode.
14676 (builtin_memset_read_str): Likewise.
14677 * builtins.c (c_readstr): Likewise.
14678 (builtin_memcpy_read_str): Likewise.
14679 (builtin_strncpy_read_str): Likewise.
14680 (builtin_memset_read_str): Likewise.
14681 (builtin_memset_gen_str): Likewise.
14682 (expand_builtin_signbit): Use scalar_int_mode for local variables.
14683 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
14684 instead of a machine_mode.
14685 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
14687 (make_extraction): Likewise.
14688 (try_widen_shift_mode): Take and return scalar_int_modes instead
14690 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
14691 a scalar_int_mode instead of a machine_mode.
14692 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
14693 (avr_addr_space_pointer_mode): Likewise.
14694 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
14695 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
14696 (msp430_unwind_word_mode): Likewise.
14697 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
14698 (spu_addr_space_pointer_mode): Likewise.
14699 (spu_addr_space_address_mode): Likewise.
14700 (spu_libgcc_cmp_return_mode): Likewise.
14701 (spu_libgcc_shift_count_mode): Likewise.
14702 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
14703 (rl78_addr_space_pointer_mode): Likewise.
14704 (fl78_unwind_word_mode): Likewise.
14705 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
14707 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
14708 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
14709 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
14710 (mips_valid_pointer_mode): Likewise.
14711 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
14712 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
14713 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
14715 (ft32_addr_space_address_mode): Likewise.
14716 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
14717 scalar_int_mode instead of a machine_mode.
14718 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
14720 (m32c_addr_space_address_mode): Likewise.
14721 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
14722 (rs6000_eh_return_filter_mode): Likewise.
14723 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
14724 (rs6000_eh_return_filter_mode): Likewise.
14725 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
14726 (s390_libgcc_shift_count_mode): Likewise.
14727 (s390_unwind_word_mode): Likewise.
14728 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
14730 * target.def (mode_rep_extended): Likewise.
14731 (valid_pointer_mode): Likewise.
14732 (addr_space.valid_pointer_mode): Likewise.
14733 (eh_return_filter_mode): Return a scalar_int_mode rather than
14735 (libgcc_cmp_return_mode): Likewise.
14736 (libgcc_shift_count_mode): Likewise.
14737 (unwind_word_mode): Likewise.
14738 (addr_space.pointer_mode): Likewise.
14739 (addr_space.address_mode): Likewise.
14740 * doc/tm.texi: Regenerate.
14741 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
14743 (do_jump): Use scalar_int_mode for local variables.
14744 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
14745 rather than a machine_mode.
14746 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
14747 (scompare_loc_descriptor_wide): Likewise.
14748 (scompare_loc_descriptor_narrow): Likewise.
14749 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
14751 * except.c (sjlj_emit_dispatch_table): Likewise.
14752 (expand_builtin_eh_copy_values): Likewise.
14753 * explow.c (convert_memory_address_addr_space_1): Likewise.
14754 Take a scalar_int_mode rather than a machine_mode.
14755 (convert_memory_address_addr_space): Take a scalar_int_mode rather
14756 than a machine_mode.
14757 (memory_address_addr_space): Use scalar_int_mode for local variables.
14758 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
14759 rather than a machine_mode.
14760 * expmed.c (mask_rtx): Likewise.
14761 (init_expmed_one_conv): Likewise.
14762 (expand_mult_highpart_adjust): Likewise.
14763 (extract_high_half): Likewise.
14764 (expmed_mult_highpart_optab): Likewise.
14765 (expmed_mult_highpart): Likewise.
14766 (expand_smod_pow2): Likewise.
14767 (expand_sdiv_pow2): Likewise.
14768 (emit_store_flag_int): Likewise.
14769 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
14771 (extract_low_bits): Likewise.
14772 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
14774 * expr.c (pieces_addr::adjust): Likewise.
14775 (can_store_by_pieces): Likewise.
14776 (store_by_pieces): Likewise.
14777 (clear_by_pieces_1): Likewise.
14778 (expand_expr_addr_expr_1): Likewise.
14779 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
14780 (expand_expr_real_1): Likewise.
14781 (try_casesi): Likewise.
14782 * final.c (shorten_branches): Likewise.
14783 * fold-const.c (fold_convert_const_int_from_fixed): Change the
14784 type of "mode" to machine_mode.
14785 * internal-fn.c (expand_arith_overflow_result_store): Take a
14786 scalar_int_mode rather than a machine_mode.
14787 (expand_mul_overflow): Use scalar_int_mode for local variables.
14788 * loop-doloop.c (doloop_modify): Likewise.
14789 (doloop_optimize): Likewise.
14790 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
14791 than a machine_mode.
14792 (expand_doubleword_shift_condmove): Likewise.
14793 (expand_doubleword_shift): Likewise.
14794 (expand_doubleword_clz): Likewise.
14795 (expand_doubleword_popcount): Likewise.
14796 (expand_doubleword_parity): Likewise.
14797 (expand_absneg_bit): Use scalar_int_mode for local variables.
14798 (prepare_float_lib_cmp): Likewise.
14799 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
14800 rather than a machine_mode.
14801 (convert_memory_address_addr_space): Likewise.
14802 (get_mode_bounds): Likewise.
14803 (get_address_mode): Return a scalar_int_mode rather than a
14805 * rtlanal.c (get_address_mode): Likewise.
14806 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
14807 than a machine_mode.
14808 * targhooks.c (default_mode_rep_extended): Likewise.
14809 (default_valid_pointer_mode): Likewise.
14810 (default_addr_space_valid_pointer_mode): Likewise.
14811 (default_eh_return_filter_mode): Return a scalar_int_mode rather
14812 than a machine_mode.
14813 (default_libgcc_cmp_return_mode): Likewise.
14814 (default_libgcc_shift_count_mode): Likewise.
14815 (default_unwind_word_mode): Likewise.
14816 (default_addr_space_pointer_mode): Likewise.
14817 (default_addr_space_address_mode): Likewise.
14818 * targhooks.h (default_eh_return_filter_mode): Likewise.
14819 (default_libgcc_cmp_return_mode): Likewise.
14820 (default_libgcc_shift_count_mode): Likewise.
14821 (default_unwind_word_mode): Likewise.
14822 (default_addr_space_pointer_mode): Likewise.
14823 (default_addr_space_address_mode): Likewise.
14824 (default_mode_rep_extended): Take a scalar_int_mode rather than
14826 (default_valid_pointer_mode): Likewise.
14827 (default_addr_space_valid_pointer_mode): Likewise.
14828 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
14830 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
14831 rather than a machine_mode.
14832 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
14833 for local variables.
14834 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14835 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
14836 than a machine_mode.
14838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14839 Alan Hayward <alan.hayward@arm.com>
14840 David Sherwood <david.sherwood@arm.com>
14842 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
14843 the mode argument to scalar_int_mode.
14844 (do_jump_by_parts_zero_rtx): Likewise.
14845 (do_jump_by_parts_equality_rtx): Likewise.
14846 (do_jump_by_parts_greater): Take a mode argument.
14847 (do_jump_by_parts_equality): Likewise.
14848 (do_jump_1): Update calls accordingly.
14850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14851 Alan Hayward <alan.hayward@arm.com>
14852 David Sherwood <david.sherwood@arm.com>
14854 * is-a.h (safe_dyn_cast): New function.
14855 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
14856 (jump_table_for_label): Likewise.
14857 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
14858 instead of an rtx_insn *.
14859 (shorten_branches): Use dyn_cast instead of LABEL_P and
14860 JUMP_TABLE_DATA_P. Use jump_table_for_label and
14861 rtx_jump_table_data::get_data_mode.
14862 (final_scan_insn): Likewise.
14864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14865 Alan Hayward <alan.hayward@arm.com>
14866 David Sherwood <david.sherwood@arm.com>
14868 * combine.c (try_combine): Use is_a <scalar_int_mode> when
14869 trying to combine a full-register integer set with a subreg
14872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14873 Alan Hayward <alan.hayward@arm.com>
14874 David Sherwood <david.sherwood@arm.com>
14876 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
14877 that is always either address_mode or pointer_mode.
14879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14880 Alan Hayward <alan.hayward@arm.com>
14881 David Sherwood <david.sherwood@arm.com>
14883 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
14884 when the two are known to be equal.
14886 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14887 Alan Hayward <alan.hayward@arm.com>
14888 David Sherwood <david.sherwood@arm.com>
14890 * simplify-rtx.c (simplify_const_unary_operation): Use
14891 is_a <scalar_int_mode> instead of checking for a nonzero
14892 precision. Forcibly convert op_mode to a scalar_int_mode
14893 in that case. More clearly differentiate the operand and
14894 result modes and use the former when deciding what the value
14895 of a count-bits operation should be. Use is_int_mode instead
14896 of checking for a MODE_INT. Remove redundant check for whether
14897 this mode has a zero precision.
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 * optabs.c (widen_leading): Change the type of the mode argument
14904 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
14905 (widen_bswap): Likewise.
14906 (expand_parity): Likewise.
14907 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
14908 (expand_ffs): Likewise.
14909 (epand_unop): Check for scalar integer modes before calling the
14912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14913 Alan Hayward <alan.hayward@arm.com>
14914 David Sherwood <david.sherwood@arm.com>
14916 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
14918 (expand_expr_real_1): Update call accordingly.
14920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14921 Alan Hayward <alan.hayward@arm.com>
14922 David Sherwood <david.sherwood@arm.com>
14924 * expmed.c (store_bit_field_using_insv): Add op0_mode and
14925 value_mode arguments. Use scalar_int_mode internally.
14926 (store_bit_field_1): Rename the new integer mode from imode
14927 to op0_mode and use it instead of GET_MODE (op0). Update calls
14928 to store_split_bit_field, store_bit_field_using_insv and
14929 store_fixed_bit_field.
14930 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
14931 Use scalar_int_mode internally. Use a bit count rather than a mode
14932 when calculating the largest bit size for get_best_mode.
14933 Update calls to store_split_bit_field and store_fixed_bit_field_1.
14934 (store_fixed_bit_field_1): Add mode and value_mode arguments.
14935 Remove assertion that OP0 has a scalar integer mode.
14936 (store_split_bit_field): Add op0_mode and value_mode arguments.
14937 Update calls to extract_fixed_bit_field.
14938 (extract_bit_field_using_extv): Add an op0_mode argument.
14939 Use scalar_int_mode internally.
14940 (extract_bit_field_1): Rename the new integer mode from imode to
14941 op0_mode and use it instead of GET_MODE (op0). Update calls to
14942 extract_split_bit_field, extract_bit_field_using_extv and
14943 extract_fixed_bit_field.
14944 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
14945 to extract_split_bit_field and extract_fixed_bit_field_1.
14946 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
14947 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
14948 on the target mode.
14949 (extract_split_bit_field): Add an op0_mode argument. Update call
14950 to extract_fixed_bit_field.
14952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14953 Alan Hayward <alan.hayward@arm.com>
14954 David Sherwood <david.sherwood@arm.com>
14956 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
14957 * explow.c (hard_function_value): Likewise.
14958 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
14959 convert_to_mode call outside the loop.
14960 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
14961 for the mode iterator. Require the mode specified by max_pieces
14963 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
14965 (copy_blkmode_to_reg): Likewise.
14966 (set_storage_via_setmem): Likewise.
14967 * optabs.c (prepare_cmp_insn): Likewise.
14968 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14969 * stor-layout.c (finish_bitfield_representative): Likewise.
14971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14972 Alan Hayward <alan.hayward@arm.com>
14973 David Sherwood <david.sherwood@arm.com>
14975 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
14976 * expr.c (convert_move): Use them.
14977 (convert_modes): Likewise.
14978 (store_expr_with_bounds): Likewise.
14980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14981 Alan Hayward <alan.hayward@arm.com>
14982 David Sherwood <david.sherwood@arm.com>
14984 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
14985 parameter for the mode of "x". Remove the "known_x", "known_mode"
14986 and "known_ret" arguments. Change the type of the mode argument
14987 to scalar_int_mode.
14988 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
14989 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
14990 (reg_num_sign_bit_copies_for_combine): Likewise.
14991 * rtlanal.c (nonzero_bits1): Likewise.
14992 (num_sign_bit_copies1): Likewise.
14993 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
14994 (reg_num_sign_bit_copies_general): Likewise.
14995 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
14996 (reg_nonzero_bits_general): Likewise.
14998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14999 Alan Hayward <alan.hayward@arm.com>
15000 David Sherwood <david.sherwood@arm.com>
15002 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15003 than in subroutines. Return 1 for non-integer modes.
15004 (cached_num_sign_bit_copies): Change the type of the mode parameter
15005 to scalar_int_mode.
15006 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15007 classes. Handle CONST_INT_P first and then check whether X also
15008 has a scalar integer mode. Check the same thing for inner registers
15009 of a SUBREG and for values that are being extended or truncated.
15011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15012 Alan Hayward <alan.hayward@arm.com>
15013 David Sherwood <david.sherwood@arm.com>
15015 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15016 in subroutines. Return the mode mask for non-integer modes.
15017 (cached_nonzero_bits): Change the type of the mode parameter
15018 to scalar_int_mode.
15019 (nonzero_bits1): Likewise. Remove early exit for other mode
15020 classes. Handle CONST_INT_P first and then check whether X
15021 also has a scalar integer mode.
15023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15024 Alan Hayward <alan.hayward@arm.com>
15025 David Sherwood <david.sherwood@arm.com>
15027 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15028 Return a scalar_int_mode and assert that the size is greater than
15030 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15031 (op_by_pieces_d::op_by_pieces_d): Likewise.
15032 (op_by_pieces_d::run): Likewise.
15033 (can_store_by_pieces): Likewise.
15035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15036 Alan Hayward <alan.hayward@arm.com>
15037 David Sherwood <david.sherwood@arm.com>
15039 * combine.c (extract_left_shift): Add a mode argument and update
15041 (make_compound_operation_int): Change the type of the mode parameter
15042 to scalar_int_mode and update the call to extract_left_shift.
15044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15045 Alan Hayward <alan.hayward@arm.com>
15046 David Sherwood <david.sherwood@arm.com>
15048 * combine.c (simplify_and_const_int): Change the type of the mode
15049 parameter to scalar_int_mode.
15050 (simplify_and_const_int_1): Likewise. Update recursive call.
15052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15053 Alan Hayward <alan.hayward@arm.com>
15054 David Sherwood <david.sherwood@arm.com>
15056 * combine.c (simplify_compare_const): Check that the mode is a
15057 scalar_int_mode (rather than VOIDmode) before testing its
15059 (simplify_comparison): Move COMPARISON_P handling out of the
15060 loop and restrict the latter part of the loop to scalar_int_modes.
15061 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15062 and when considering SUBREG_REGs. Use is_int_mode instead of
15063 checking GET_MODE_CLASS against MODE_INT.
15065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15066 Alan Hayward <alan.hayward@arm.com>
15067 David Sherwood <david.sherwood@arm.com>
15069 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15070 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15071 shift_unit_mode and for modes involved in scalar shifts.
15073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15074 Alan Hayward <alan.hayward@arm.com>
15075 David Sherwood <david.sherwood@arm.com>
15077 * combine.c (force_int_to_mode): New function, split out from...
15078 (force_to_mode): ...here. Keep xmode up-to-date and use it
15079 instead of GET_MODE (x).
15081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15082 Alan Hayward <alan.hayward@arm.com>
15083 David Sherwood <david.sherwood@arm.com>
15085 * optabs-query.h (extraction_insn::struct_mode): Change type to
15086 opt_scalar_int_mode and update comment.
15087 (extraction_insn::field_mode): Change type to scalar_int_mode.
15088 (extraction_insn::pos_mode): Likewise.
15089 * combine.c (make_extraction): Update accordingly.
15090 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15091 (get_optab_extraction_insn): Likewise.
15092 * recog.c (simplify_while_replacing): Likewise.
15093 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15094 parameter to opt_scalar_int_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 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15101 to a scalar_int_mode instead of a machine_mode.
15102 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15103 (get_best_mode): Return a boolean and use a pointer argument to store
15104 the selected mode. Replace the limit mode parameter with a bit limit.
15105 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15106 for the values returned by bit_field_mode_iterator::next_mode.
15107 (store_bit_field): Update call to get_best_mode.
15108 (store_fixed_bit_field): Likewise.
15109 (extract_fixed_bit_field): Likewise.
15110 * expr.c (optimize_bitfield_assignment_op): Likewise.
15111 * fold-const.c (optimize_bit_field_compare): Likewise.
15112 (fold_truth_andor_1): Likewise.
15113 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15114 Update for new type of m_mode.
15115 (get_best_mode): As above.
15117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15118 Alan Hayward <alan.hayward@arm.com>
15119 David Sherwood <david.sherwood@arm.com>
15121 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15122 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15123 (store_bit_field): Check is_a <scalar_int_mode> before calling
15124 strict_volatile_bitfield_p.
15125 (extract_bit_field): Likewise.
15127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15128 Alan Hayward <alan.hayward@arm.com>
15129 David Sherwood <david.sherwood@arm.com>
15131 * target.def (cstore_mode): Return a scalar_int_mode.
15132 * doc/tm.texi: Regenerate.
15133 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15134 * targhooks.h (default_cstore_mode): Likewise.
15135 * targhooks.c (default_cstore_mode): Likewise, using a forced
15137 * expmed.c (emit_cstore): Expect the target of the cstore to be
15140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15141 Alan Hayward <alan.hayward@arm.com>
15142 David Sherwood <david.sherwood@arm.com>
15144 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15146 (niter_desc): Likewise mode.
15147 (iv_analyze): Add a mode parameter.
15149 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15150 and change its type to scalar_int_mode.
15151 * loop-iv.c: Update commentary at head of file.
15152 (iv_constant): Pass the mode paraeter before the rtx it describes
15153 and change its type to scalar_int_mode. Remove VOIDmode handling.
15154 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15155 (iv_extend): Likewise.
15156 (shorten_into_mode): Likewise.
15157 (iv_add): Use scalar_int_mode.
15158 (iv_mult): Likewise.
15159 (iv_shift): Likewise.
15160 (canonicalize_iv_subregs): Likewise.
15161 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15162 it describes and change its mode to scalar_int_mode. Also change
15163 the type of the returned inner_mode to scalar_int_mode.
15164 (get_biv_step): Likewise, turning outer_mode from a pointer
15165 into a direct parameter. Update call to get_biv_step_1.
15166 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15167 iv_constant and get_biv_step.
15168 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15169 and change its type to scalar_int_mode. Don't initialise iv->mode
15170 to VOIDmode and remove later checks for its still being VOIDmode.
15171 Update calls to iv_analyze_op and iv_analyze_expr. Check
15172 is_a <scalar_int_mode> when changing the mode under consideration.
15173 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15174 Update call to iv_analyze_expr.
15175 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15176 inner register is not also a scalar_int_mode. Update call to
15178 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15179 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15180 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15181 separate mode class checks. Update calls to iv_analyze. Remove
15182 fix-up of VOIDmodes after iv_analyze_biv.
15183 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15184 don't have a scalar_int_mode. Update call to biv_p.
15186 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15187 Alan Hayward <alan.hayward@arm.com>
15188 David Sherwood <david.sherwood@arm.com>
15190 * cfgexpand.c (convert_debug_memory_address): Use
15191 as_a <scalar_int_mode>.
15192 * combine.c (expand_compound_operation): Likewise.
15193 (make_extraction): Likewise.
15194 (change_zero_ext): Likewise.
15195 (simplify_comparison): Likewise.
15196 * cse.c (cse_insn): Likewise.
15197 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15198 (mem_loc_descriptor): Likewise.
15199 (loc_descriptor): Likewise.
15200 * expmed.c (init_expmed_one_mode): Likewise.
15201 (synth_mult): Likewise.
15202 (emit_store_flag_1): Likewise.
15203 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15204 of a comparison with size.
15205 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15206 (reduce_to_bit_field_precision): Likewise.
15207 * function.c (expand_function_end): Likewise.
15208 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15209 * loop-doloop.c (doloop_modify): Likewise.
15210 * optabs.c (expand_binop): Likewise.
15211 (expand_unop): Likewise.
15212 (expand_copysign_absneg): Likewise.
15213 (prepare_cmp_insn): Likewise.
15214 (maybe_legitimize_operand): Likewise.
15215 * recog.c (const_scalar_int_operand): Likewise.
15216 * rtlanal.c (get_address_mode): Likewise.
15217 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15218 (simplify_cond_clz_ctz): Likewise.
15219 * tree-nested.c (get_nl_goto_field): Likewise.
15220 * tree.c (build_vector_type_for_mode): Likewise.
15221 * var-tracking.c (use_narrower_mode): Likewise.
15223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15224 Alan Hayward <alan.hayward@arm.com>
15225 David Sherwood <david.sherwood@arm.com>
15227 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15228 * builtins.c (expand_builtin_signbit): Use it.
15229 * cfgexpand.c (expand_debug_expr): Likewise.
15230 * dojump.c (do_jump): Likewise.
15231 (do_compare_and_jump): Likewise.
15232 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15233 * expmed.c (make_tree): Likewise.
15234 * expr.c (expand_expr_real_2): Likewise.
15235 (expand_expr_real_1): Likewise.
15236 (try_casesi): Likewise.
15237 * fold-const-call.c (fold_const_call_ss): Likewise.
15238 * fold-const.c (unextend): Likewise.
15239 (extract_muldiv_1): Likewise.
15240 (fold_single_bit_test): Likewise.
15241 (native_encode_int): Likewise.
15242 (native_encode_string): Likewise.
15243 (native_interpret_int): Likewise.
15244 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15245 * internal-fn.c (expand_addsub_overflow): Likewise.
15246 (expand_neg_overflow): Likewise.
15247 (expand_mul_overflow): Likewise.
15248 (expand_arith_overflow): Likewise.
15249 * match.pd: Likewise.
15250 * stor-layout.c (layout_type): Likewise.
15251 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15252 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15253 * tree-ssanames.c (get_range_info): Likewise.
15254 * tree-switch-conversion.c (array_value_type) Likewise.
15255 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15256 (vect_recog_divmod_pattern): Likewise.
15257 (vect_recog_mixed_size_cond_pattern): Likewise.
15258 * tree-vrp.c (extract_range_basic): Likewise.
15259 (simplify_float_conversion_using_ranges): Likewise.
15260 * tree.c (int_fits_type_p): Likewise.
15261 * ubsan.c (instrument_bool_enum_load): Likewise.
15262 * varasm.c (mergeable_string_section): Likewise.
15263 (narrowing_initializer_constant_valid_p): Likewise.
15264 (output_constant): Likewise.
15266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15267 Alan Hayward <alan.hayward@arm.com>
15268 David Sherwood <david.sherwood@arm.com>
15270 * machmode.h (NARROWEST_INT_MODE): New macro.
15271 * expr.c (alignment_for_piecewise_move): Use it instead of
15272 GET_CLASS_NARROWEST_MODE (MODE_INT).
15273 (push_block): Likewise.
15274 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15276 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15279 Alan Hayward <alan.hayward@arm.com>
15280 David Sherwood <david.sherwood@arm.com>
15282 * postreload.c (move2add_valid_value_p): Change the type of the
15283 mode parameter to scalar_int_mode.
15284 (move2add_use_add2_insn): Add a mode parameter and use it instead
15286 (move2add_use_add3_insn): Likewise.
15287 (reload_cse_move2add): Update accordingly.
15289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15290 Alan Hayward <alan.hayward@arm.com>
15291 David Sherwood <david.sherwood@arm.com>
15293 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15295 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15296 * optabs.c (expand_unop): Likewise.
15298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15299 Alan Hayward <alan.hayward@arm.com>
15300 David Sherwood <david.sherwood@arm.com>
15302 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15303 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15304 (popcount_loc_descriptor): Likewise.
15305 (bswap_loc_descriptor): Likewise.
15306 (rotate_loc_descriptor): Likewise.
15307 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15308 calling the functions above.
15310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15311 Alan Hayward <alan.hayward@arm.com>
15312 David Sherwood <david.sherwood@arm.com>
15314 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15316 (try_combine): Likewise.
15317 (simplify_if_then_else): Likewise.
15318 * cse.c (cse_insn): Likewise.
15319 * dwarf2out.c (mem_loc_descriptor): Likewise.
15320 * emit-rtl.c (gen_lowpart_common): Likewise.
15321 * simplify-rtx.c (simplify_truncation): Likewise.
15322 (simplify_binary_operation_1): Likewise.
15323 (simplify_const_relational_operation): Likewise.
15324 (simplify_ternary_operation): Likewise.
15325 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15328 Alan Hayward <alan.hayward@arm.com>
15329 David Sherwood <david.sherwood@arm.com>
15331 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15332 * reload.c (push_reload): Likewise.
15333 (find_reloads): Likewise.
15335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15336 Alan Hayward <alan.hayward@arm.com>
15337 David Sherwood <david.sherwood@arm.com>
15339 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15340 (make_compound_operation_int): Likewise.
15341 (change_zero_ext): Likewise.
15342 * expr.c (convert_move): Likewise.
15343 (convert_modes): Likewise.
15344 * fwprop.c (forward_propagate_subreg): Likewise.
15345 * loop-iv.c (get_biv_step_1): Likewise.
15346 * optabs.c (widen_operand): Likewise.
15347 * postreload.c (move2add_valid_value_p): Likewise.
15348 * recog.c (simplify_while_replacing): Likewise.
15349 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15350 (simplify_binary_operation_1): Likewise. Remove redundant
15351 mode equality check.
15353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15354 Alan Hayward <alan.hayward@arm.com>
15355 David Sherwood <david.sherwood@arm.com>
15357 * combine.c (combine_simplify_rtx): Add checks for
15358 is_a <scalar_int_mode>.
15359 (simplify_if_then_else): Likewise.
15360 (make_field_assignment): Likewise.
15361 (simplify_comparison): Likewise.
15362 * ifcvt.c (noce_try_bitop): Likewise.
15363 * loop-invariant.c (canonicalize_address_mult): Likewise.
15364 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15367 Alan Hayward <alan.hayward@arm.com>
15368 David Sherwood <david.sherwood@arm.com>
15370 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15371 is_a <scalar_int_mode> instead of != BLKmode.
15373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15374 Alan Hayward <alan.hayward@arm.com>
15375 David Sherwood <david.sherwood@arm.com>
15377 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15378 instead of != VOIDmode.
15379 * combine.c (if_then_else_cond): Likewise.
15380 (change_zero_ext): Likewise.
15381 * dwarf2out.c (mem_loc_descriptor): Likewise.
15382 (loc_descriptor): Likewise.
15383 * rtlanal.c (canonicalize_condition): Likewise.
15384 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15387 Alan Hayward <alan.hayward@arm.com>
15388 David Sherwood <david.sherwood@arm.com>
15390 * simplify-rtx.c (simplify_binary_operation_1): Use
15391 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15394 Alan Hayward <alan.hayward@arm.com>
15395 David Sherwood <david.sherwood@arm.com>
15397 * wide-int.h (int_traits<unsigned char>) New class.
15398 (int_traits<unsigned short>) Likewise.
15399 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15400 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15402 * combine.c (set_nonzero_bits_and_sign_copies): Use
15403 is_a <scalar_int_mode>.
15404 (find_split_point): Likewise.
15405 (combine_simplify_rtx): Likewise.
15406 (simplify_logical): Likewise.
15407 (expand_compound_operation): Likewise.
15408 (expand_field_assignment): Likewise.
15409 (make_compound_operation): Likewise.
15410 (extended_count): Likewise.
15411 (change_zero_ext): Likewise.
15412 (simplify_comparison): Likewise.
15413 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15414 (ucompare_loc_descriptor): Likewise.
15415 (minmax_loc_descriptor): Likewise.
15416 (mem_loc_descriptor): Likewise.
15417 (loc_descriptor): Likewise.
15418 * expmed.c (init_expmed_one_mode): Likewise.
15419 * lra-constraints.c (lra_constraint_offset): Likewise.
15420 * optabs.c (prepare_libcall_arg): Likewise.
15421 * postreload.c (move2add_note_store): Likewise.
15422 * reload.c (operands_match_p): Likewise.
15423 * rtl.h (load_extend_op): Likewise.
15424 * rtlhooks.c (gen_lowpart_general): Likewise.
15425 * simplify-rtx.c (simplify_truncation): Likewise.
15426 (simplify_unary_operation_1): Likewise.
15427 (simplify_binary_operation_1): Likewise.
15428 (simplify_const_binary_operation): Likewise.
15429 (simplify_const_relational_operation): Likewise.
15430 (simplify_subreg): Likewise.
15431 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15432 * var-tracking.c (adjust_mems): Likewise.
15433 (prepare_call_arguments): Likewise.
15435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15436 Alan Hayward <alan.hayward@arm.com>
15437 David Sherwood <david.sherwood@arm.com>
15439 * machmode.h (is_int_mode): New fuction.
15440 * combine.c (find_split_point): Use it.
15441 (combine_simplify_rtx): Likewise.
15442 (simplify_if_then_else): Likewise.
15443 (simplify_set): Likewise.
15444 (simplify_shift_const_1): Likewise.
15445 (simplify_comparison): Likewise.
15446 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15447 * cse.c (notreg_cost): Likewise.
15448 (cse_insn): Likewise.
15449 * cselib.c (cselib_lookup_1): Likewise.
15450 * dojump.c (do_jump_1): Likewise.
15451 (do_compare_rtx_and_jump): Likewise.
15452 * dse.c (get_call_args): Likewise.
15453 * dwarf2out.c (rtl_for_decl_init): Likewise.
15454 (native_encode_initializer): Likewise.
15455 * expmed.c (emit_store_flag_1): Likewise.
15456 (emit_store_flag): Likewise.
15457 * expr.c (convert_modes): Likewise.
15458 (store_field): Likewise.
15459 (expand_expr_real_1): Likewise.
15460 * fold-const.c (fold_read_from_constant_string): Likewise.
15461 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15462 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15463 * optabs.c (expand_binop): Likewise.
15464 (expand_unop): Likewise.
15465 (expand_abs_nojump): Likewise.
15466 (expand_one_cmpl_abs_nojump): Likewise.
15467 * simplify-rtx.c (mode_signbit_p): Likewise.
15468 (val_signbit_p): Likewise.
15469 (val_signbit_known_set_p): Likewise.
15470 (val_signbit_known_clear_p): Likewise.
15471 (simplify_relational_operation_1): Likewise.
15472 * tree.c (vector_type_mode): Likewise.
15474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15475 Alan Hayward <alan.hayward@arm.com>
15476 David Sherwood <david.sherwood@arm.com>
15478 * machmode.h (smallest_mode_for_size): Fix formatting.
15479 (smallest_int_mode_for_size): New function.
15480 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15481 instead of smallest_mode_for_size.
15482 * combine.c (make_extraction): Likewise.
15483 * config/arc/arc.c (arc_expand_movmem): Likewise.
15484 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15485 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15486 * config/s390/s390.c (s390_expand_insv): Likewise.
15487 * config/sparc/sparc.c (assign_int_registers): Likewise.
15488 * config/spu/spu.c (spu_function_value): Likewise.
15489 (spu_function_arg): Likewise.
15490 * coverage.c (get_gcov_type): Likewise.
15491 (get_gcov_unsigned_t): Likewise.
15492 * dse.c (find_shift_sequence): Likewise.
15493 * expmed.c (store_bit_field_1): Likewise.
15494 * expr.c (convert_move): Likewise.
15495 (store_field): Likewise.
15496 * internal-fn.c (expand_arith_overflow): Likewise.
15497 * optabs-query.c (get_best_extraction_insn): Likewise.
15498 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15499 * stor-layout.c (layout_type): Likewise.
15500 (initialize_sizetypes): Likewise.
15501 * targhooks.c (default_get_mask_mode): Likewise.
15502 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15505 Alan Hayward <alan.hayward@arm.com>
15506 David Sherwood <david.sherwood@arm.com>
15508 * machmode.h (opt_mode::else_blk): New function.
15509 (int_mode_for_mode): Declare.
15510 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15511 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15513 * cfgexpand.c (expand_debug_expr): Likewise.
15514 * combine.c (gen_lowpart_or_truncate): Likewise.
15515 (gen_lowpart_for_combine): Likewise.
15516 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15517 * config/avr/avr.c (avr_to_int_mode): Likewise.
15518 (avr_out_plus_1): Likewise.
15519 (avr_out_plus): Likewise.
15520 (avr_out_round): Likewise.
15521 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15522 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15523 (s390_expand_vcond): Likewise.
15524 * config/spu/spu.c (spu_split_immediate): Likewise.
15525 (spu_expand_mov): Likewise.
15526 * dse.c (get_stored_val): Likewise.
15527 * expmed.c (store_bit_field_1): Likewise.
15528 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15530 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15531 (extract_low_bits): Likewise.
15532 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15533 handling rather than repeating the check.
15534 (emit_group_store): Likewise.
15535 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15536 * optabs.c (expand_absneg_bit): Likewise.
15537 (expand_copysign_absneg): Likewise.
15538 (expand_copysign_bit): Likewise.
15539 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15540 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15541 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15542 * var-tracking.c (prepare_call_arguments): Likewise.
15543 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15544 int_mode_for_mode instead of mode_for_size.
15545 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15548 Alan Hayward <alan.hayward@arm.com>
15549 David Sherwood <david.sherwood@arm.com>
15551 * machmode.h (int_mode_for_size): New function.
15552 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15553 instead of mode_for_size.
15554 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15556 * combine.c (expand_field_assignment): Use int_mode_for_size
15557 instead of mode_for_size.
15558 (make_extraction): Likewise.
15559 (simplify_shift_const_1): Likewise.
15560 (simplify_comparison): Likewise.
15561 * dojump.c (do_jump): Likewise.
15562 * dwarf2out.c (mem_loc_descriptor): Likewise.
15563 * emit-rtl.c (init_derived_machine_modes): Likewise.
15564 * expmed.c (flip_storage_order): Likewise.
15565 (convert_extracted_bit_field): Likewise.
15566 * expr.c (copy_blkmode_from_reg): Likewise.
15567 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15568 * internal-fn.c (expand_mul_overflow): Likewise.
15569 * lower-subreg.c (simple_move): Likewise.
15570 * optabs-libfuncs.c (init_optabs): Likewise.
15571 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15572 * tree.c (vector_type_mode): Likewise.
15573 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15574 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15575 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15576 * tree-vect-stmts.c (vectorizable_load): Likewise.
15577 (vectorizable_store): Likewise.
15579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15580 Alan Hayward <alan.hayward@arm.com>
15581 David Sherwood <david.sherwood@arm.com>
15583 * coretypes.h (pod_mode): New type.
15584 (scalar_int_mode_pod): New typedef.
15585 * machmode.h (pod_mode): New class.
15586 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15587 * genmodes.c (emit_mode_int_n): Update accordingly.
15588 * lower-subreg.h (target_lower_subreg): Change type to
15589 scalar_int_mode_pod.
15590 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15591 scalar_int_mode_pod.
15593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15594 Alan Hayward <alan.hayward@arm.com>
15595 David Sherwood <david.sherwood@arm.com>
15597 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
15598 machine_mode to scalar_int_mode.
15599 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
15600 (rs6000_option_override_internal): Remove cast to int.
15601 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
15602 machine_mode to scalar_int_mode.
15603 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
15604 (rs6000_option_override_internal): Remove cast to int.
15605 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
15606 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
15608 * config/s390/s390.c (s390_expand_builtin): Likewise.
15609 * coretypes.h (scalar_int_mode): New type.
15610 (opt_scalar_int_mode): New typedef.
15611 * machmode.h (scalar_int_mode): New class.
15612 (scalar_int_mode::includes_p): New function.
15613 (byte_mode): Change type to scalar_int_mode.
15614 (word_mode): Likewise.
15615 (ptr_mode): Likewise.
15616 * emit-rtl.c (byte_mode): Likewise.
15617 (word_mode): Likewise.
15618 (ptr_mode): Likewise.
15619 (init_derived_machine_modes): Update accordingly.
15620 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
15621 and MODE_PARTIAL_INT.
15622 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
15623 opt_scalar_int_mode.
15625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15626 Alan Hayward <alan.hayward@arm.com>
15627 David Sherwood <david.sherwood@arm.com>
15629 * target.def (libgcc_floating_mode_supported_p): Take a
15631 * doc/tm.texi: Regenerate.
15632 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
15634 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15635 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
15638 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15639 Alan Hayward <alan.hayward@arm.com>
15640 David Sherwood <david.sherwood@arm.com>
15642 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
15643 * doc/tm.texi: Regenerate.
15644 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
15645 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
15646 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
15647 * targhooks.h (default_floatn_mode): Likewise.
15648 * targhooks.c (default_floatn_mode): Likewise.
15649 * tree.c (build_common_tree_nodes): Update accordingly.
15651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15652 Alan Hayward <alan.hayward@arm.com>
15653 David Sherwood <david.sherwood@arm.com>
15655 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
15656 (mode_iterator::iterate_p): Likewise.
15657 (mode_iterator::get_wider): Likewise.
15658 * expr.c (init_expr_target): Use opt_scalar_float_mode.
15660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15661 Alan Hayward <alan.hayward@arm.com>
15662 David Sherwood <david.sherwood@arm.com>
15664 * coretypes.h (opt_scalar_float_mode): New typedef.
15665 * machmode.h (float_mode_for_size): New function.
15666 * emit-rtl.c (double_mode): Delete.
15667 (init_emit_once): Use float_mode_for_size.
15668 * stor-layout.c (layout_type): Likewise.
15669 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
15671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15672 Alan Hayward <alan.hayward@arm.com>
15673 David Sherwood <david.sherwood@arm.com>
15675 * output.h (assemble_real): Take a scalar_float_mode.
15676 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
15677 * config/arm/arm.md (consttable_4): Likewise.
15678 (consttable_8): Likewise.
15679 (consttable_16): Likewise.
15680 * config/mips/mips.md (consttable_float): Likewise.
15681 * config/s390/s390.c (s390_output_pool_entry): Likewise.
15682 * varasm.c (assemble_real): Take a scalar_float_mode.
15683 (output_constant_pool_2): Update accordingly.
15684 (output_constant): Likewise.
15686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15687 Alan Hayward <alan.hayward@arm.com>
15688 David Sherwood <david.sherwood@arm.com>
15690 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
15691 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
15692 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
15693 (native_encode_real): Likewise.
15694 (native_interpret_real): Likewise.
15695 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
15696 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15699 Alan Hayward <alan.hayward@arm.com>
15700 David Sherwood <david.sherwood@arm.com>
15702 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
15703 <scalar_float_mode>. Simplify.
15704 (gen_extend_conv_libfunc): Likewise.
15706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15707 Alan Hayward <alan.hayward@arm.com>
15708 David Sherwood <david.sherwood@arm.com>
15710 * coretypes.h (scalar_float_mode): New type.
15711 * machmode.h (mode_traits::from_int): Use machine_mode if
15712 USE_ENUM_MODES is defined.
15713 (is_a): New function.
15715 (dyn_cast): Likewise.
15716 (scalar_float_mode): New class.
15717 (scalar_float_mode::includes_p): New function.
15718 (is_float_mode): Likewise.
15719 * gdbhooks.py (MachineModePrinter): New class.
15720 (build_pretty_printer): Use it for scalar_float_mode.
15721 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
15722 (format_helper::format_helper): Turn into a template.
15723 * genmodes.c (get_mode_class): New function.
15724 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
15725 or machine_mode if none.
15726 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
15727 as_a <scalar_float_mode>.
15728 * dwarf2out.c (mem_loc_descriptor): Likewise.
15729 (insert_float): Likewise.
15730 (add_const_value_attribute): Likewise.
15731 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15732 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
15733 (expand_unop): Update accordingly.
15734 (expand_abs_nojump): Likewise.
15735 (expand_copysign_absneg): Take a scalar_float_mode.
15736 (expand_copysign_bit): Likewise.
15737 (expand_copysign): Update accordingly.
15739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15740 Alan Hayward <alan.hayward@arm.com>
15741 David Sherwood <david.sherwood@arm.com>
15743 * coretypes.h (opt_mode): New class.
15744 * machmode.h (opt_mode): Likewise.
15745 (opt_mode::else_void): New function.
15746 (opt_mode::require): Likewise.
15747 (opt_mode::exists): Likewise.
15748 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
15749 (GET_MODE_2XWIDER_MODE): Likewise.
15750 (mode_iterator::get_wider): Update accordingly.
15751 (mode_iterator::get_2xwider): Likewise.
15752 (mode_iterator::get_known_wider): Likewise, turning into a template.
15753 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
15754 forcing a wider mode to exist.
15755 * config/cr16/cr16.h (LONG_REG_P): Likewise.
15756 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15757 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
15758 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
15759 * lower-subreg.c (init_lower_subreg): Likewise.
15760 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
15761 on the final iteration.
15762 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
15763 a wider mode exists before asking for a move pattern.
15764 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
15765 forcing a wider mode to exist.
15766 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
15767 returning false if no such mode exists.
15768 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
15769 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
15770 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
15771 Avoid checking for a MODE_INT if we already know the mode is not a
15773 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
15774 forcing a wider mode to exist.
15775 (expmed_mult_highpart_optab): Likewise.
15776 (expmed_mult_highpart): Likewise.
15777 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
15779 * lto-streamer-in.c (lto_input_mode_table): Likewise.
15780 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
15781 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15782 * internal-fn.c (expand_mul_overflow): Update use of
15783 GET_MODE_2XWIDER_MODE.
15784 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15785 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
15786 GET_MODE_WIDER_MODE.
15787 (convert_plusminus_to_widen): Likewise.
15788 * tree-switch-conversion.c (array_value_type): Likewise.
15789 * var-tracking.c (emit_note_insn_var_location): Likewise.
15790 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15791 Return false inside rather than outside the loop if no wider mode
15793 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
15794 and GET_MODE_2XWIDER_MODE
15795 (can_compare_p): Use else_void.
15796 * gdbhooks.py (OptMachineModePrinter): New class.
15797 (build_pretty_printer): Use it for opt_mode.
15799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15800 Alan Hayward <alan.hayward@arm.com>
15801 David Sherwood <david.sherwood@arm.com>
15803 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
15804 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
15806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15807 Alan Hayward <alan.hayward@arm.com>
15808 David Sherwood <david.sherwood@arm.com>
15810 * machmode.h (mode_traits): New structure.
15811 (get_narrowest_mode): New function.
15812 (mode_iterator::start): Likewise.
15813 (mode_iterator::iterate_p): Likewise.
15814 (mode_iterator::get_wider): Likewise.
15815 (mode_iterator::get_known_wider): Likewise.
15816 (mode_iterator::get_2xwider): Likewise.
15817 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
15818 (FOR_EACH_MODE): Likewise.
15819 (FOR_EACH_MODE_FROM): Likewise.
15820 (FOR_EACH_MODE_UNTIL): Likewise.
15821 (FOR_EACH_WIDER_MODE): Likewise.
15822 (FOR_EACH_2XWIDER_MODE): Likewise.
15823 * builtins.c (expand_builtin_strlen): Use new mode iterators.
15824 * combine.c (simplify_comparison): Likewise
15825 * config/i386/i386.c (type_natural_mode): Likewise.
15826 * cse.c (cse_insn): Likewise.
15827 * dse.c (find_shift_sequence): Likewise.
15828 * emit-rtl.c (init_derived_machine_modes): Likewise.
15829 (init_emit_once): Likewise.
15830 * explow.c (hard_function_value): Likewise.
15831 * expmed.c (extract_fixed_bit_field_1): Likewise.
15832 (extract_bit_field_1): Likewise.
15833 (expand_divmod): Likewise.
15834 (emit_store_flag_1): Likewise.
15835 * expr.c (init_expr_target): Likewise.
15836 (convert_move): Likewise.
15837 (alignment_for_piecewise_move): Likewise.
15838 (widest_int_mode_for_size): Likewise.
15839 (emit_block_move_via_movmem): Likewise.
15840 (copy_blkmode_to_reg): Likewise.
15841 (set_storage_via_setmem): Likewise.
15842 (compress_float_constant): Likewise.
15843 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15844 * optabs-query.c (get_best_extraction_insn): Likewise.
15845 * optabs.c (expand_binop): Likewise.
15846 (expand_twoval_unop): Likewise.
15847 (expand_twoval_binop): Likewise.
15848 (widen_leading): Likewise.
15849 (widen_bswap): Likewise.
15850 (expand_parity): Likewise.
15851 (expand_unop): Likewise.
15852 (prepare_cmp_insn): Likewise.
15853 (prepare_float_lib_cmp): Likewise.
15854 (expand_float): Likewise.
15855 (expand_fix): Likewise.
15856 (expand_sfix_optab): Likewise.
15857 * postreload.c (move2add_use_add2_insn): Likewise.
15858 * reg-stack.c (reg_to_stack): Likewise.
15859 * reginfo.c (choose_hard_reg_mode): Likewise.
15860 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15861 * stor-layout.c (mode_for_size): Likewise.
15862 (smallest_mode_for_size): Likewise.
15863 (mode_for_vector): Likewise.
15864 (finish_bitfield_representative): Likewise.
15865 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
15866 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
15867 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15868 * var-tracking.c (prepare_call_arguments): Likewise.
15870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15871 Alan Hayward <alan.hayward@arm.com>
15872 David Sherwood <david.sherwood@arm.com>
15874 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
15875 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
15876 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
15877 * machmode.h (mode_size): Move earlier in file.
15878 (mode_precision): Likewise.
15879 (mode_inner): Likewise.
15880 (mode_nunits): Likewise.
15881 (mode_unit_size): Likewise.
15882 (unit_unit_precision): Likewise.
15883 (mode_wider): Likewise.
15884 (mode_2xwider): Likewise.
15885 (machine_mode): New class.
15886 (mode_to_bytes): New function.
15887 (mode_to_bits): Likewise.
15888 (mode_to_precision): Likewise.
15889 (mode_to_inner): Likewise.
15890 (mode_to_unit_size): Likewise.
15891 (mode_to_unit_precision): Likewise.
15892 (mode_to_nunits): Likewise.
15893 (GET_MODE_SIZE): Use mode_to_bytes.
15894 (GET_MODE_BITSIZE): Use mode_to_bits.
15895 (GET_MODE_PRECISION): Use mode_to_precision.
15896 (GET_MODE_INNER): Use mode_to_inner.
15897 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
15898 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
15899 (GET_MODE_NUNITS): Use mode_to_nunits.
15900 * system.h (ALWAYS_INLINE): New macro.
15901 * config/powerpcspe/powerpcspe-c.c
15902 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
15903 int for arg1_mode and arg2_mode.
15905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15906 Alan Hayward <alan.hayward@arm.com>
15907 David Sherwood <david.sherwood@arm.com>
15909 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15910 Prefix mode names with E_ in case statements.
15911 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15912 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
15913 (aarch64_split_simd_move): Likewise.
15914 (aarch64_gen_storewb_pair): Likewise.
15915 (aarch64_gen_loadwb_pair): Likewise.
15916 (aarch64_gen_store_pair): Likewise.
15917 (aarch64_gen_load_pair): Likewise.
15918 (aarch64_get_condition_code_1): Likewise.
15919 (aarch64_constant_pool_reload_icode): Likewise.
15920 (get_rsqrte_type): Likewise.
15921 (get_rsqrts_type): Likewise.
15922 (get_recpe_type): Likewise.
15923 (get_recps_type): Likewise.
15924 (aarch64_gimplify_va_arg_expr): Likewise.
15925 (aarch64_simd_container_mode): Likewise.
15926 (aarch64_emit_load_exclusive): Likewise.
15927 (aarch64_emit_store_exclusive): Likewise.
15928 (aarch64_expand_compare_and_swap): Likewise.
15929 (aarch64_gen_atomic_cas): Likewise.
15930 (aarch64_emit_bic): Likewise.
15931 (aarch64_emit_atomic_swap): Likewise.
15932 (aarch64_emit_atomic_load_op): Likewise.
15933 (aarch64_evpc_trn): Likewise.
15934 (aarch64_evpc_uzp): Likewise.
15935 (aarch64_evpc_zip): Likewise.
15936 (aarch64_evpc_ext): Likewise.
15937 (aarch64_evpc_rev): Likewise.
15938 (aarch64_evpc_dup): Likewise.
15939 (aarch64_gen_ccmp_first): Likewise.
15940 (aarch64_gen_ccmp_next): Likewise.
15941 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15942 (alpha_emit_xfloating_libcall): Likewise.
15943 (emit_insxl): Likewise.
15944 (alpha_arg_type): Likewise.
15945 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
15946 (arc_preferred_simd_mode): Likewise.
15947 (arc_secondary_reload): Likewise.
15948 (get_arc_condition_code): Likewise.
15949 (arc_print_operand): Likewise.
15950 (arc_legitimate_constant_p): Likewise.
15951 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15952 * config/arc/arc.md (casesi_load): Likewise.
15953 (casesi_compact_jump): Likewise.
15954 * config/arc/predicates.md (proper_comparison_operator): Likewise.
15955 (cc_use_register): Likewise.
15956 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15957 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15958 (arm_init_iwmmxt_builtins): Likewise.
15959 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
15960 (neon_expand_vector_init): Likewise.
15961 (arm_attr_length_move_neon): Likewise.
15962 (maybe_get_arm_condition_code): Likewise.
15963 (arm_emit_vector_const): Likewise.
15964 (arm_preferred_simd_mode): Likewise.
15965 (arm_output_iwmmxt_tinsr): Likewise.
15966 (thumb1_output_casesi): Likewise.
15967 (thumb2_output_casesi): Likewise.
15968 (arm_emit_load_exclusive): Likewise.
15969 (arm_emit_store_exclusive): Likewise.
15970 (arm_expand_compare_and_swap): Likewise.
15971 (arm_evpc_neon_vuzp): Likewise.
15972 (arm_evpc_neon_vzip): Likewise.
15973 (arm_evpc_neon_vrev): Likewise.
15974 (arm_evpc_neon_vtrn): Likewise.
15975 (arm_evpc_neon_vext): Likewise.
15976 (arm_validize_comparison): Likewise.
15977 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
15978 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
15979 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
15980 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
15981 (c6x_preferred_simd_mode): Likewise.
15982 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
15983 (epiphany_rtx_costs): Likewise.
15984 * config/epiphany/predicates.md (proper_comparison_operator):
15986 * config/frv/frv.c (condexec_memory_operand): Likewise.
15987 (frv_emit_move): Likewise.
15988 (output_move_single): Likewise.
15989 (output_condmove_single): Likewise.
15990 (frv_hard_regno_mode_ok): Likewise.
15991 (frv_matching_accg_mode): Likewise.
15992 * config/h8300/h8300.c (split_adds_subs): Likewise.
15993 (h8300_rtx_costs): Likewise.
15994 (h8300_print_operand): Likewise.
15995 (compute_mov_length): Likewise.
15996 (output_logical_op): Likewise.
15997 (compute_logical_op_length): Likewise.
15998 (compute_logical_op_cc): Likewise.
15999 (h8300_shift_needs_scratch_p): Likewise.
16000 (output_a_shift): Likewise.
16001 (compute_a_shift_length): Likewise.
16002 (compute_a_shift_cc): Likewise.
16003 (expand_a_rotate): Likewise.
16004 (output_a_rotate): Likewise.
16005 * config/i386/i386.c (classify_argument): Likewise.
16006 (function_arg_advance_32): Likewise.
16007 (function_arg_32): Likewise.
16008 (function_arg_64): Likewise.
16009 (function_value_64): Likewise.
16010 (ix86_gimplify_va_arg): Likewise.
16011 (ix86_legitimate_constant_p): Likewise.
16012 (put_condition_code): Likewise.
16013 (split_double_mode): Likewise.
16014 (ix86_avx256_split_vector_move_misalign): Likewise.
16015 (ix86_expand_vector_logical_operator): Likewise.
16016 (ix86_split_idivmod): Likewise.
16017 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16018 (ix86_build_const_vector): Likewise.
16019 (ix86_build_signbit_mask): Likewise.
16020 (ix86_match_ccmode): Likewise.
16021 (ix86_cc_modes_compatible): Likewise.
16022 (ix86_expand_branch): Likewise.
16023 (ix86_expand_sse_cmp): Likewise.
16024 (ix86_expand_sse_movcc): Likewise.
16025 (ix86_expand_int_sse_cmp): Likewise.
16026 (ix86_expand_vec_perm_vpermi2): Likewise.
16027 (ix86_expand_vec_perm): Likewise.
16028 (ix86_expand_sse_unpack): Likewise.
16029 (ix86_expand_int_addcc): Likewise.
16030 (ix86_split_to_parts): Likewise.
16031 (ix86_vectorize_builtin_gather): Likewise.
16032 (ix86_vectorize_builtin_scatter): Likewise.
16033 (avx_vpermilp_parallel): Likewise.
16034 (inline_memory_move_cost): Likewise.
16035 (ix86_tieable_integer_mode_p): Likewise.
16036 (x86_maybe_negate_const_int): Likewise.
16037 (ix86_expand_vector_init_duplicate): Likewise.
16038 (ix86_expand_vector_init_one_nonzero): Likewise.
16039 (ix86_expand_vector_init_one_var): Likewise.
16040 (ix86_expand_vector_init_concat): Likewise.
16041 (ix86_expand_vector_init_interleave): Likewise.
16042 (ix86_expand_vector_init_general): Likewise.
16043 (ix86_expand_vector_set): Likewise.
16044 (ix86_expand_vector_extract): Likewise.
16045 (emit_reduc_half): Likewise.
16046 (ix86_emit_i387_round): Likewise.
16047 (ix86_mangle_type): Likewise.
16048 (ix86_expand_round_sse4): Likewise.
16049 (expand_vec_perm_blend): Likewise.
16050 (canonicalize_vector_int_perm): Likewise.
16051 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16052 (expand_vec_perm_1): Likewise.
16053 (expand_vec_perm_interleave3): Likewise.
16054 (expand_vec_perm_even_odd_pack): Likewise.
16055 (expand_vec_perm_even_odd_1): Likewise.
16056 (expand_vec_perm_broadcast_1): Likewise.
16057 (ix86_vectorize_vec_perm_const_ok): Likewise.
16058 (ix86_expand_vecop_qihi): Likewise.
16059 (ix86_expand_mul_widen_hilo): Likewise.
16060 (ix86_expand_sse2_abs): Likewise.
16061 (ix86_expand_pextr): Likewise.
16062 (ix86_expand_pinsr): Likewise.
16063 (ix86_preferred_simd_mode): Likewise.
16064 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16065 * config/i386/sse.md (*andnot<mode>3): Likewise.
16066 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16067 (*<code><mode>3): Likewise.
16068 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16069 (ia64_expand_atomic_op): Likewise.
16070 (ia64_arg_type): Likewise.
16071 (ia64_mode_to_int): Likewise.
16072 (ia64_scalar_mode_supported_p): Likewise.
16073 (ia64_vector_mode_supported_p): Likewise.
16074 (expand_vec_perm_broadcast): Likewise.
16075 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16076 (iq2000_function_arg_advance): Likewise.
16077 (iq2000_function_arg): Likewise.
16078 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16079 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16080 (m68k_libcall_value): Likewise.
16081 (m68k_function_value): Likewise.
16082 (sched_attr_op_type): Likewise.
16083 * config/mcore/mcore.c (mcore_output_move): Likewise.
16084 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16086 (microblaze_function_arg): Likewise.
16087 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16088 (mips_print_operand): Likewise.
16089 (mips_mode_ok_for_mov_fmt_p): Likewise.
16090 (mips_vector_mode_supported_p): Likewise.
16091 (mips_preferred_simd_mode): Likewise.
16092 (mips_expand_vpc_loongson_even_odd): Likewise.
16093 (mips_expand_vec_unpack): Likewise.
16094 (mips_expand_vi_broadcast): Likewise.
16095 (mips_expand_vector_init): Likewise.
16096 (mips_expand_vec_reduc): Likewise.
16097 (mips_expand_msa_cmp): Likewise.
16098 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16099 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16100 (cc_flags_for_mode): Likewise.
16101 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16102 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16103 (nds32_output_casesi_pc_relative): Likewise.
16104 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16105 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16106 (nvptx_gen_unpack): Likewise.
16107 (nvptx_gen_pack): Likewise.
16108 (nvptx_gen_shuffle): Likewise.
16109 (nvptx_gen_wcast): Likewise.
16110 (nvptx_preferred_simd_mode): Likewise.
16111 * config/pa/pa.c (pa_secondary_reload): Likewise.
16112 * config/pa/predicates.md (base14_operand): Likewise.
16113 * config/powerpcspe/powerpcspe-c.c
16114 (altivec_resolve_overloaded_builtin): Likewise.
16115 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16117 (rs6000_preferred_simd_mode): Likewise.
16118 (output_vec_const_move): Likewise.
16119 (rs6000_expand_vector_extract): Likewise.
16120 (rs6000_split_vec_extract_var): Likewise.
16121 (reg_offset_addressing_ok_p): Likewise.
16122 (rs6000_legitimate_offset_address_p): Likewise.
16123 (rs6000_legitimize_address): Likewise.
16124 (rs6000_emit_set_const): Likewise.
16125 (rs6000_const_vec): Likewise.
16126 (rs6000_emit_move): Likewise.
16127 (spe_build_register_parallel): Likewise.
16128 (rs6000_darwin64_record_arg_recurse): Likewise.
16129 (swap_selector_for_mode): Likewise.
16130 (spe_init_builtins): Likewise.
16131 (paired_init_builtins): Likewise.
16132 (altivec_init_builtins): Likewise.
16133 (do_load_for_compare): Likewise.
16134 (rs6000_generate_compare): Likewise.
16135 (rs6000_expand_float128_convert): Likewise.
16136 (emit_load_locked): Likewise.
16137 (emit_store_conditional): Likewise.
16138 (rs6000_output_function_epilogue): Likewise.
16139 (rs6000_handle_altivec_attribute): Likewise.
16140 (rs6000_function_value): Likewise.
16141 (emit_fusion_gpr_load): Likewise.
16142 (emit_fusion_p9_load): Likewise.
16143 (emit_fusion_p9_store): Likewise.
16144 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16145 (fusion_gpr_mem_load): Likewise.
16146 (fusion_addis_mem_combo_load): Likewise.
16147 (fusion_addis_mem_combo_store): Likewise.
16148 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16149 (fusion_gpr_mem_load): Likewise.
16150 (fusion_addis_mem_combo_load): Likewise.
16151 (fusion_addis_mem_combo_store): Likewise.
16152 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16154 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16155 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16156 (rs6000_preferred_simd_mode): Likewise.
16157 (output_vec_const_move): Likewise.
16158 (rs6000_expand_vector_extract): Likewise.
16159 (rs6000_split_vec_extract_var): Likewise.
16160 (reg_offset_addressing_ok_p): Likewise.
16161 (rs6000_legitimate_offset_address_p): Likewise.
16162 (rs6000_legitimize_address): Likewise.
16163 (rs6000_emit_set_const): Likewise.
16164 (rs6000_const_vec): Likewise.
16165 (rs6000_emit_move): Likewise.
16166 (rs6000_darwin64_record_arg_recurse): Likewise.
16167 (swap_selector_for_mode): Likewise.
16168 (paired_init_builtins): Likewise.
16169 (altivec_init_builtins): Likewise.
16170 (rs6000_expand_float128_convert): Likewise.
16171 (emit_load_locked): Likewise.
16172 (emit_store_conditional): Likewise.
16173 (rs6000_output_function_epilogue): Likewise.
16174 (rs6000_handle_altivec_attribute): Likewise.
16175 (rs6000_function_value): Likewise.
16176 (emit_fusion_gpr_load): Likewise.
16177 (emit_fusion_p9_load): Likewise.
16178 (emit_fusion_p9_store): Likewise.
16179 * config/rx/rx.c (rx_gen_move_template): Likewise.
16180 (flags_from_mode): Likewise.
16181 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16182 (s390_slb_comparison): Likewise.
16183 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16184 (s390_vector_mode_supported_p): Likewise.
16185 (s390_cc_modes_compatible): Likewise.
16186 (s390_match_ccmode_set): Likewise.
16187 (s390_canonicalize_comparison): Likewise.
16188 (s390_emit_compare_and_swap): Likewise.
16189 (s390_branch_condition_mask): Likewise.
16190 (s390_rtx_costs): Likewise.
16191 (s390_secondary_reload): Likewise.
16192 (__SECONDARY_RELOAD_CASE): Likewise.
16193 (s390_expand_cs): Likewise.
16194 (s390_preferred_simd_mode): Likewise.
16195 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16196 * config/sh/sh.c (sh_print_operand): Likewise.
16197 (dump_table): Likewise.
16198 (sh_secondary_reload): Likewise.
16199 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16200 * config/sh/sh.md (casesi_worker_1): Likewise.
16201 (casesi_worker_2): Likewise.
16202 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16203 (fcc_comparison_operator): Likewise.
16204 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16205 (emit_soft_tfmode_cvt): Likewise.
16206 (sparc_preferred_simd_mode): Likewise.
16207 (output_cbranch): Likewise.
16208 (sparc_print_operand): Likewise.
16209 (sparc_expand_vec_perm_bmask): Likewise.
16210 (vector_init_bshuffle): Likewise.
16211 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16212 (spu_vector_mode_supported_p): Likewise.
16213 (spu_expand_insv): Likewise.
16214 (spu_emit_branch_or_set): Likewise.
16215 (spu_handle_vector_attribute): Likewise.
16216 (spu_builtin_splats): Likewise.
16217 (spu_builtin_extract): Likewise.
16218 (spu_builtin_promote): Likewise.
16219 (spu_expand_sign_extend): Likewise.
16220 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16221 (tilegx_simd_int): Likewise.
16222 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16223 (tilepro_simd_int): Likewise.
16224 * config/v850/v850.c (const_double_split): Likewise.
16225 (v850_print_operand): Likewise.
16226 (ep_memory_offset): Likewise.
16227 * config/vax/vax.c (vax_rtx_costs): Likewise.
16228 (vax_output_int_move): Likewise.
16229 (vax_output_int_add): Likewise.
16230 (vax_output_int_subtract): Likewise.
16231 * config/visium/predicates.md (visium_branch_operator): Likewise.
16232 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16233 (visium_print_operand_address): Likewise.
16234 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16235 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16236 (xtensa_expand_conditional_branch): Likewise.
16237 (xtensa_copy_incoming_a7): Likewise.
16238 (xtensa_output_literal): Likewise.
16239 * dfp.c (decimal_real_maxval): Likewise.
16240 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16243 Alan Hayward <alan.hayward@arm.com>
16244 David Sherwood <david.sherwood@arm.com>
16246 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16247 (mode_nunits_inline): Likewise.
16248 (mode_inner_inline): Likewise.
16249 (mode_unit_size_inline): Likewise.
16250 (mode_unit_precision_inline): Likewise.
16251 (emit_insn_modes_h): Likewise. Also emit a #define of the
16253 (emit_mode_wider): Add an E_ prefix to mode names.
16254 (emit_mode_complex): Likewise.
16255 (emit_mode_inner): Likewise.
16256 (emit_mode_adjustments): Likewise.
16257 (emit_mode_int_n): Likewise.
16258 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16259 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16260 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16261 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16262 (CRC32_BUILTIN, ENTRY): Likewise.
16263 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16264 (aarch64_pop_regs): Likewise.
16265 (aarch64_process_components): Likewise.
16266 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16267 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16268 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16269 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16270 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16271 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16273 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16274 (pcmpistr): Likewise.
16275 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16276 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16277 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16279 * config/rl78/rl78.c (mduc_regs): Likewise.
16280 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16281 (htm_expand_builtin): Likewise.
16282 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16283 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16284 * config/xtensa/xtensa.c (print_operand): Likewise.
16285 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16286 (NUM_MODE_VECTOR_INT): Likewise.
16287 * genoutput.c (null_operand): Likewise.
16288 (output_operand_data): Likewise.
16289 * genrecog.c (print_parameter_value): Likewise.
16290 * lra.c (debug_operand_data): Likewise.
16292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16294 * dwarf2out.c (scompare_loc_descriptor_wide)
16295 (scompare_loc_descriptor_narrow): New functions, split out from...
16296 (scompare_loc_descriptor): ...here.
16297 * expmed.c (emit_store_flag_int): New function, split out from...
16298 (emit_store_flag): ...here.
16300 2017-08-30 Richard Biener <rguenther@suse.de>
16302 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16303 (dwarf2out_early_finish): Move setting of AT_pubnames from
16304 early debug output to early finish.
16306 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16308 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16309 and -mdata-region to the assembler.
16311 2017-08-30 Richard Biener <rguenther@suse.de>
16313 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16315 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16316 (dwarf2out_early_global_decl): Only generate a DIE for the
16317 abstract origin if it doesn't already exist or is a declaration DIE.
16318 (resolve_addr): Do not add the linkage name twice when
16319 generating a stub DIE for the DW_TAG_GNU_call_site target.
16321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16323 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16324 Use machine_mode rather than int for arg1_mode.
16326 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16329 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16330 that the second argument of the built-in functions to unpack
16331 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16332 switch statement instead a lot of if statements.
16333 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16334 Allow 64-bit values to be in Altivec registers as well as
16335 traditional floating point registers.
16336 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16338 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16340 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16341 MAX_REGS_PER_ADDRESS == 1.
16343 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16345 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16346 * config/i386/i386.c (ix86_option_override_internal): Simplify
16347 setting of opts->x_flag_entry.
16349 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16350 Jakub Jelinek <jakub@redhat.com>
16351 Richard Biener <rguenther@suse.de>
16353 PR tree-optimization/81503
16354 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16355 folded constant fits in the target type; reorder tests for clarity.
16357 2017-08-29 Martin Liska <mliska@suse.cz>
16359 * passes.def: Include pass_lower_switch.
16360 * stmt.c (dump_case_nodes): Remove and move to
16361 tree-switch-conversion.
16362 (case_values_threshold): Likewise.
16363 (expand_switch_as_decision_tree_p): Likewise.
16364 (emit_case_decision_tree): Likewise.
16365 (expand_case): Likewise.
16366 (balance_case_nodes): Likewise.
16367 (node_has_low_bound): Likewise.
16368 (node_has_high_bound): Likewise.
16369 (node_is_bounded): Likewise.
16370 (emit_case_nodes): Likewise.
16371 (struct simple_case_node): New struct.
16372 (add_case_node): Remove.
16373 (emit_case_dispatch_table): Use vector instead of case_list.
16374 (reset_out_edges_aux): Remove.
16375 (compute_cases_per_edge): Likewise.
16376 (expand_case): Build list of simple_case_node.
16377 (expand_sjlj_dispatch_table): Use it.
16378 * tree-switch-conversion.c (struct case_node): Moved from
16379 stmt.c and adjusted.
16380 (emit_case_nodes): Likewise.
16381 (node_has_low_bound): Likewise.
16382 (node_has_high_bound): Likewise.
16383 (node_is_bounded): Likewise.
16384 (case_values_threshold): Likewise.
16385 (reset_out_edges_aux): Likewise.
16386 (compute_cases_per_edge): Likewise.
16387 (add_case_node): Likewise.
16388 (dump_case_nodes): Likewise.
16389 (balance_case_nodes): Likewise.
16390 (expand_switch_as_decision_tree_p): Likewise.
16391 (emit_jump): Likewise.
16392 (emit_case_decision_tree): Likewise.
16393 (try_switch_expansion): Likewise.
16394 (do_jump_if_equal): Likewise.
16395 (emit_cmp_and_jump_insns): Likewise.
16396 (fix_phi_operands_for_edge): New function.
16397 (record_phi_operand_mapping): Likewise.
16398 (class pass_lower_switch): New pass.
16399 (pass_lower_switch::execute): New function.
16400 (make_pass_lower_switch): Likewise.
16401 (conditional_probability):
16402 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16403 * tree-pass.h: Add make_pass_lower_switch.
16405 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16408 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16411 2017-08-29 Richard Biener <rguenther@suse.de>
16413 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16414 we do not add a DW_AT_inline attribute twice.
16415 (gen_subprogram_die): Remove code setting DW_AT_inline on
16416 DECL_ABSTRACT_P nodes.
16418 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16420 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16421 calls to internal functions.
16422 (gimplify_modify_expr): Likewise.
16423 * tree-call-cdce.c (use_internal_fn): Likewise.
16424 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16425 (convert_to_divmod): Set the nothrow flag.
16426 * tree-if-conv.c (predicate_mem_writes): Likewise.
16427 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16428 (vectorizable_call): Likewise.
16429 (vectorizable_store): Likewise.
16430 (vectorizable_load): Likewise.
16431 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16432 (vect_recog_mask_conversion_pattern): Likewise.
16434 2017-08-29 Martin Liska <mliska@suse.cz>
16437 * gcc.c (driver_handle_option): Add new argument.
16438 * opts-common.c (handle_option): Pass
16439 target_option_override_hook.
16440 * opts-global.c (lang_handle_option): Add new option.
16441 (set_default_handlers): Add new argument.
16442 (decode_options): Likewise.
16443 * opts.c (target_handle_option): Likewise.
16444 (common_handle_option): Call target_option_override_hook.
16445 * opts.h (struct cl_option_handler_func): Add hook for
16446 target option override.
16447 (struct cl_option_handlers): Likewise.
16448 (set_default_handlers): Add new argument.
16449 (decode_options): Likewise.
16450 (common_handle_option): Likewise.
16451 (target_handle_option): Likewise.
16452 * toplev.c (toplev::main): Pass targetm.target_option.override
16455 2017-08-29 Richard Biener <rguenther@suse.de>
16456 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16458 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16459 life to the active subtree.
16461 2017-08-28 Jeff Law <law@redhat.com>
16463 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16464 derive_equivalences.
16465 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16466 Code moved into....
16467 (edge_info::derive_equivalences): New private member function
16469 * tree-ssa-dom.c (class edge_info): Changed from a struct
16470 to a class. Add ctor/dtor, methods and data members.
16471 (edge_info::edge_info): Renamed from allocate_edge_info.
16472 Initialize additional members.
16473 (edge_info::~edge_info): New.
16474 (free_dom_edge_info): Delete the edge info.
16475 (record_edge_info): Use new class & associated member functions.
16476 Tighten forms for testing for edge equivalences.
16477 (record_temporary_equivalences): Iterate over the simple
16478 equivalences rather than assuming there's only one per edge.
16479 (cprop_into_successor_phis): Iterate over the simple
16480 equivalences rather than assuming there's only one per edge.
16481 (optimize_stmt): Use operand_equal_p rather than pointer
16482 equality for mini-DSE code.
16484 2017-08-28 Nathan Sidwell <nathan@acm.org>
16486 * gcc.c (execute): Fold SIGPIPE handling into switch
16487 statement. Adjust internal error message.
16489 2017-08-28 Richard Biener <rguenther@suse.de>
16492 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16493 Do nothing for removed DIEs.
16495 2017-08-28 Richard Biener <rguenther@suse.de>
16497 PR tree-optimization/81977
16498 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16501 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16504 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16505 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16506 using targetm.gen_mem_thread_fence.
16508 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16511 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16512 predicate to register_operand. Reorder operands.
16513 (*btr<mode>): Ditto.
16514 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16515 (*btr<mode>_mask): Ditto.
16517 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16519 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16520 * config/rs6000/xmmintrin.h: New file.
16521 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16523 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16526 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16527 parameter and_insn and return it.
16528 (recombine_lvx_pattern): Insert a copy to ensure availability of
16529 the base register of the copied masking operation at the point of
16530 the instruction replacement.
16531 (recombine_stvx_pattern): Likewise.
16533 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16535 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16536 undocumented switches.
16537 (-mpower9-dform-vector): Likewise.
16538 (-mpower9-dform): Likewise.
16539 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16540 comments to delete references to -mpower9-dform* switches.
16541 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16542 Delete reference to -mpower9-dform* switches, test for
16543 -mpower9-vector instead.
16544 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16545 (OTHER_P9_VECTOR_MASKS): Likewise.
16546 (POWERPC_MASKS): Likewise.
16547 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16548 tests against -mpower9-dform* to -mpower9-vector. Delete code
16549 that checked for -mpower9-dform* consistancy with other options.
16550 Add test for -mpower9-misc to enable other power9 switches.
16551 (rs6000_init_hard_regno_mode_ok): Likewise.
16552 (rs6000_option_override_internal): Likewise.
16553 (rs6000_emit_prologue): Likewise.
16554 (rs6000_emit_epilogue): Likewise.
16555 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16556 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16557 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16559 (emit_fusion_p9_store): Likewise.
16560 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16561 resetting these macros if the assembler does not support ISA 3.0
16563 (TARGET_P9_DFORM_VECTOR): Likewise.
16564 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16565 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16567 2017-08-25 Alan Modra <amodra@gmail.com>
16569 PR rtl-optimization/81747
16570 * cse.c (cse_extended_basic_block): Don't attempt to record
16571 equivalences for degenerate conditional branches that branch
16572 to their fall-through.
16574 2017-08-24 Martin Sebor <msebor@redhat.com>
16576 PR middle-end/81908
16577 * gimple-fold.c (size_must_be_zero_p): New function.
16578 (gimple_fold_builtin_memory_op): Call it.
16580 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16582 * config/rs6000/mm_malloc.h: New file.
16584 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16586 PR tree-optimization/81913
16587 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16588 analysis when either IVs in condition can wrap.
16590 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16592 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16593 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
16595 2017-08-24 Richard Biener <rguenther@suse.de>
16598 * targhooks.c (default_target_can_inline_p): Properly
16599 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
16600 is present and always compare.
16601 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
16602 infer -mfpmath=sse from TARGET_SSE_P.
16603 (ix86_can_inline_p): Properly use target_option_default_node when
16604 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
16606 2017-08-24 Richard Biener <rguenther@suse.de>
16609 * dwarf2out.c (output_die): Handle flag_generate_offload like
16611 (output_comp_unit): Likewise.
16612 (gen_array_type_die): Likewise.
16613 (dwarf2out_early_finish): Likewise.
16614 (note_variable_value_in_expr): Likewise.
16615 (dwarf2out_finish): Likewise. Adjust assert.
16616 * cgraphunit.c (symbol_table::compile): Move setting of
16617 flag_generate_offload earlier ...
16618 (symbol_table::finalize_compilation_unit): ... here, before
16619 early debug finalization.
16621 2017-08-24 Richard Biener <rguenther@suse.de>
16623 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
16624 and ipa-fnsummary.h.
16625 (ix86_can_inline_p): When ix86_fpmath flags do not match
16626 check whether the callee uses FP math at all.
16628 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
16630 PR middle-end/81931
16631 * tree-ssanames.c (get_nonzero_bits): Use element_precision
16632 instead of TYPE_PRECISION.
16634 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
16635 Alan Hayward <alan.hayward@arm.com>
16636 David Sherwood <david.sherwood@arm.com>
16638 * combine.c (make_extraction): Use subreg_offset_from_lsb.
16640 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
16642 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
16644 (ix86_frame::stack_realign_allocate): New field.
16645 (struct machine_frame_state): Modify comments.
16646 (machine_frame_state::sp_realigned_fp_end): New field.
16647 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
16648 layout calculation.
16649 (sp_valid_at): Add assertion to assure no attempt to access invalid
16650 offset of a realigned stack.
16651 (fp_valid_at): Likewise.
16652 (choose_baseaddr): Modify comments.
16653 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
16654 ix86_expand_prologue.
16655 (ix86_expand_prologue): Modify stack realignment and allocation.
16656 (ix86_expand_epilogue): Modify comments.
16657 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
16658 avx2_runtime, avx512f, and avx512f_runtime.
16660 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
16662 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
16663 (mstackrealign): Do not init to -1.
16664 * config/i386/i386.c (ix86_option_override_internal):
16665 Check opts_set, not opts when setting default value of
16666 opts->x_ix86_force_align_arg_pointer.
16668 2017-08-23 Richard Biener <rguenther@suse.de>
16670 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
16671 lang_hooks.decl_printable_name.
16672 * print-rtl-function.c (print_rtx_function): Likewise.
16673 * tree-pretty-print.c (dump_function_header): Likewise.
16675 2017-08-23 Richard Biener <rguenther@suse.de>
16678 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
16679 -g0 at compile-time.
16681 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16683 PR middle-end/19706
16684 * doc/sourcebuild.texi (Other hardware attributes):
16687 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16689 PR middle-end/19706
16690 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
16691 Fix single-use check.
16693 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
16695 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
16697 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16699 * doc/install.texi: Modify to add more details on running selected
16702 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16704 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
16705 is combined with -mabi=ms.
16706 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
16709 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16711 PR tree-optimization/81488
16712 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
16713 and cached_basis fields.
16714 (MAX_SPREAD): New constant.
16715 (alloc_cand_and_find_basis): Initialize new fields.
16716 (clear_visited): New function.
16717 (create_phi_basis_1): Rename from create_phi_basis, set visited
16718 and cached_basis fields.
16719 (create_phi_basis): New wrapper function.
16720 (phi_add_costs_1): Rename from phi_add_costs, add spread
16721 parameter, set visited field, short-circuit when limits reached.
16722 (phi_add_costs): New wrapper function.
16723 (record_phi_increments_1): Rename from record_phi_increments, set
16725 (record_phi_increments): New wrapper function.
16726 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
16727 (phi_incr_cost): New wrapper function.
16728 (all_phi_incrs_profitable_1): Rename from
16729 all_phi_incrs_profitable, set visited field.
16730 (all_phi_incrs_profitable): New wrapper function.
16732 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16733 Alan Hayward <alan.hayward@arm.com>
16734 David Sherwood <david.sherwood@arm.com>
16736 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
16737 that takes the outer and inner modes.
16738 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
16739 comparison as the canonical test for a paradoxical subreg.
16740 * combine.c (simplify_set): Use paradoxical_subreg_p.
16741 (make_extraction): Likewise.
16742 (force_to_mode): Likewise.
16743 (rtx_equal_for_field_assignment_p): Likewise.
16744 (gen_lowpart_for_combine): Likewise.
16745 (simplify_comparison): Likewise.
16746 * cse.c (equiv_constant): Likewise.
16747 * expmed.c (store_bit_field_1): Likewise.
16748 * final.c (alter_subreg): Likewise.
16749 * fwprop.c (propagate_rtx): Likewise.
16750 (forward_propagate_subreg): Likewise.
16751 * ira-conflicts.c (ira_build_conflicts): Likewise.
16752 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
16753 * lra-constraints.c (curr_insn_transform): Likewise.
16754 (split_reg): Likewise.
16755 * lra-eliminations.c (move_plus_up): Likewise.
16756 (lra_eliminate_regs_1): Likewise.
16757 * recog.c (general_operand): Likewise.
16758 * ree.c (combine_reaching_defs): Likewise.
16759 * reload.c (push_reload): Likewise.
16760 (find_reloads): Likewise.
16761 * reload1.c (elimination_effects): Likewise.
16762 (compute_reload_subreg_offset): Likewise.
16763 (choose_reload_regs): Likewise.
16764 * rtlanal.c (subreg_lsb_1): Likewise.
16765 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16766 (simplify_subreg): Likewise.
16767 * var-tracking.c (track_loc_p): Likewise.
16768 * emit-rtl.c (byte_lowpart_offset): Likewise.
16769 (paradoxical_subreg_p): Delete out-of-line definition.
16771 2017-08-22 Jeff Law <law@redhat.com>
16773 PR tree-optimization/81741
16774 PR tree-optimization/71947
16775 * tree-ssa-dom.c: Include tree-inline.h.
16776 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
16777 equivalences if one is more expensive to compute than the other.
16778 * tree-ssa-scopedtables.h (class const_or_copies): Make
16779 record_const_or_copy_raw method private.
16780 (class avail_exprs_stack): New method simplify_binary_operation.
16781 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
16782 avail_exprs_stack::simplify_binary_operation as needed.
16783 (avail_exprs_stack::simplify_binary_operation): New function.
16785 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
16787 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
16788 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
16789 (DOT_SYMBOLS): Likewise.
16790 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
16791 (RELOCATABLE_NEEDS_FIXUP): Likewise.
16792 (RS6000_ABI_NAME): Likewise.
16793 (TARGET_CMODEL): Likewise.
16794 (TOC_SECTION_ASM_OP): Likewise.
16795 (SET_CMODEL): New macro.
16796 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
16798 2017-08-22 Richard Biener <rguenther@suse.de>
16800 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
16801 to allow for free-lang-data replacements similar to verify_type_variant.
16803 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16804 Alan Hayward <alan.hayward@arm.com>
16805 David Sherwood <david.sherwood@arm.com>
16807 * config/aarch64/aarch64.md (casesi): Use DImode rather than
16808 VOIDmode for the LABEL_REF.
16810 2017-08-22 Richard Biener <rguenther@suse.de>
16812 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
16814 2017-08-22 Richard Biener <rguenther@suse.de>
16816 * common.opt (feliminate-dwarf2-dups): Ignore.
16817 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
16818 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
16819 same_die_p_wrap, compute_section_prefix,
16820 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
16821 (comdat_symbol_id, comdat_symbol_number): Likewise.
16822 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
16824 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
16825 (output_die): Mark unreachable path unreachable.
16826 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
16827 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
16828 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
16829 (dwarf2out_early_finish): Likewise.
16831 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
16833 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
16835 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
16838 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
16839 not VAR_P. Filter attribute warnings with OPT_Wattributes.
16840 (avr_attribute_table) <io, io_low, address>: Initialize
16841 .decl_required with true.
16843 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
16845 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
16846 undocumented debugging options.
16847 (-mvsx-scalar-double): Likewise.
16848 (-mallow-df-permute): Likewise.
16849 (-mvectorize-builtins): Likewise.
16850 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16851 (rs6000_builtin_vectorized_function): Likewise.
16852 (rs6000_builtin_md_vectorized_function): Likewise.
16853 (rs6000_opt_vars): Likewise.
16855 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
16858 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
16859 (*btrq_imm): Rename from *btrq.
16860 (*btcq_imm): Rename from *btcq.
16861 (btsc): New code attribute.
16862 (*<btsc><mode>): New insn pattern.
16863 (*btr<mode>): Ditto.
16864 (*<btsc><mode>_mask): New insn_and_split pattern.
16865 (*btr<mode>_mask): Ditto.
16867 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16868 Alan Hayward <alan.hayward@arm.com>
16869 David Sherwood <david.sherwood@arm.com>
16871 * function.c (pad_below): Simplify padding calculation.
16873 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16874 Alan Hayward <alan.hayward@arm.com>
16875 David Sherwood <david.sherwood@arm.com>
16877 * target.def (function_prologue): Remove frame size argument.
16878 (function_epilogue): Likewise.
16879 * doc/tm.texi: Regenerate.
16880 * final.c (final_start_function): Update call to function_prologue.
16881 (final_end_function): Update call to function_epilogue.
16882 (default_function_pro_epilogue): Remove frame size argument.
16883 * output.h (default_function_pro_epilogue): Likewise.
16884 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
16885 (arm_output_function_prologue): Likewise.
16886 * config/frv/frv.c (frv_function_prologue): Likewise.
16887 (frv_function_epilogue): Likewise.
16888 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
16889 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
16890 (ia64_output_function_epilogue): Likewise.
16891 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
16892 (m32r_output_function_epilogue): Likewise.
16893 * config/microblaze/microblaze.c (microblaze_function_prologue)
16894 (microblaze_function_epilogue): Likewise.
16895 * config/mips/mips.c (mips_output_function_prologue): Likewise.
16896 (mips_output_function_epilogue): Likewise.
16897 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
16898 (mmix_target_asm_function_epilogue): Likewise.
16899 * config/msp430/msp430.c (msp430_start_function): Likewise.
16900 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
16901 (nds32_asm_function_epilogue): Likewise.
16902 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
16903 * config/pa/pa.c (pa_output_function_prologue): Likewise.
16904 (pa_output_function_epilogue): Likewise.
16905 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
16906 (rs6000_output_function_epilogue): Likewise.
16907 * config/rl78/rl78.c (rl78_start_function): Likewise.
16908 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
16909 (rs6000_output_function_epilogue): Likewise.
16910 * config/rx/rx.c (rx_output_function_prologue): Likewise.
16911 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
16912 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
16913 (sparc_asm_function_epilogue): Likewise.
16915 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16917 * tree.h (type_has_mode_precision_p): New function.
16918 * convert.c (convert_to_integer_1): Use it.
16919 * expr.c (expand_expr_real_2): Likewise.
16920 (expand_expr_real_1): Likewise.
16921 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
16922 * match.pd: Likewise.
16923 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
16924 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
16925 * tree-tailcall.c (process_assignment): Likewise.
16926 * tree-vect-loop.c (vectorizable_reduction): Likewise.
16927 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
16928 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
16929 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16930 (vectorizable_assignment): Likewise.
16931 (vectorizable_shift): Likewise.
16932 (vectorizable_operation): Likewise.
16933 * tree-vrp.c (register_edge_assert_for_2): Likewise.
16935 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
16937 * match.pd: Add pow (C, x) simplification.
16939 2017-08-21 Richard Biener <rguenther@suse.de>
16941 PR tree-optimization/81900
16942 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
16943 for blocks with abnormal predecessors.
16944 (compute_antic): Do not set visited flag prematurely.
16946 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
16949 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
16951 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16953 * stor-layout.h (vector_type_mode): Move to...
16954 * tree.h (vector_type_mode): ...here.
16955 * stor-layout.c (vector_type_mode): Move to...
16956 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
16958 2017-08-21 Richard Biener <rguenther@suse.de>
16960 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
16961 register_external_die hooks.
16962 (debug_false_tree_charstarstar_uhwistar): Declare.
16963 (debug_nothing_tree_charstar_uhwi): Likewise.
16964 * debug.c (do_nothing_debug_hooks): Adjust.
16965 (debug_false_tree_charstarstar_uhwistar): New do nothing.
16966 (debug_nothing_tree_charstar_uhwi): Likewise.
16967 * dbxout.c (dbx_debug_hooks): Adjust.
16968 (xcoff_debug_hooks): Likewise.
16969 * sdbout.c (sdb_debug_hooks): Likewise.
16970 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16971 * dwarf2out.c (macinfo_label_base): New global.
16972 (dwarf2out_register_external_die): New function for the
16973 register_external_die hook.
16974 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
16975 (dwarf2_debug_hooks): Use them.
16976 (dwarf2_lineno_debug_hooks): Adjust.
16977 (struct die_struct): Add with_offset flag.
16978 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
16979 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
16980 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
16981 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
16982 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
16983 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
16984 defining section names for the early LTO debug variants.
16985 (reset_indirect_string): New helper.
16986 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
16987 (print_dw_val): Add support for offsetted symbol references.
16988 (get_ultimate_context): Split out from is_cxx.
16989 (is_cxx): Use get_ultimate_context.
16990 (is_fortran): Add decl overload.
16991 (compute_comp_unit_symbol): Split out worker from
16992 compute_section_prefix.
16993 (compute_section_prefix): Call compute_comp_unit_symbol and
16994 set comdat_type_p here.
16995 (output_die): Skip DIE symbol output for the LTO added one.
16996 Handle DIE symbol references with offset.
16997 (output_comp_unit): Guard section name mangling properly.
16998 For LTO debug sections emit a symbol at the section beginning
16999 which we use to refer to its DIEs.
17000 (add_abstract_origin_attribute): For DIEs registered via
17001 dwarf2out_register_external_die directly refer to the early
17002 DIE rather than indirectly through the shadow one we created.
17003 Remove obsolete call to dwarf2out_abstract_function for
17004 non-function/block origins.
17005 (gen_array_type_die): When generating early LTO debug do
17006 not emit DW_AT_string_length.
17007 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17008 late when in LTO. As suggested place a gcc_unreachable for
17009 the DECL_ABSTRACT_P case.
17010 (gen_subprogram_die): Avoid another specification DIE
17011 for early built declarations/definitions for the late LTO case.
17012 (gen_variable_die): Add type references for late duplicated VLA dies
17014 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17015 we have the abstract instance already.
17016 (process_scope_var): Adjust decl DIE contexts in LTO which
17017 first puts them in limbo.
17018 (gen_decl_die): Do not generate type DIEs late apart from
17019 types for VLAs or for decls we do not yet have a DIE. Do not
17020 call dwarf2out_abstract_function late.
17021 (dwarf2out_early_global_decl): Make sure to create DIEs
17022 for abstract instances of a decl first.
17023 (dwarf2out_late_global_decl): Adjust comment.
17024 (output_macinfo_op): With multiple macro sections use
17025 macinfo_label_base to distinguish labels.
17026 (output_macinfo): Likewise. Update macinfo_label_base.
17027 Pass in the line info label.
17028 (note_variable_value_in_expr): When generating LTO resolve
17029 all variable values here by generating DIEs as needed.
17030 (init_sections_and_labels): Add early LTO debug flag parameter
17031 and generate different sections and names if set. Add generation
17032 counter for the labels so we can have multiple of them.
17033 (reset_dies): Helper to allow DIEs to be output multiple times.
17034 (dwarf2out_finish): When outputting DIEs to the fat part of an
17035 LTO object first reset DIEs.
17036 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17037 (modified_type_die): Check for decl_ultimate_origin being self
17039 (gen_type_die_with_usage): Likewise.
17040 (gen_typedef_die): Allow decl_ultimate_origin being self.
17041 (set_decl_abstract_flags): Remove.
17042 (set_block_abstract_flags): Likewise.
17043 (dwarf2out_abstract_function): Treat the early generated DIEs
17044 as the abstract copy and only add DW_AT_inline and
17045 DW_AT_artificial here and call set_decl_origin_self.
17046 If the DIE has an abstract origin don't do anything.
17047 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17048 if we have none yet (Go fails to build one, PR78628).
17049 (variably_modified_type_p): Prevent endless recursion for Ada
17050 cyclic pointer types.
17051 * lto-streamer-in.c: Include debug.h.
17052 (dref_queue): New global.
17053 (lto_read_tree_1): Stream in DIE references.
17054 (lto_input_tree): Register DIE references.
17055 (input_function): Stream DECL_DEBUG_ARGS.
17056 * lto-streamer-out.c: Include debug.h.
17057 (lto_write_tree_1): Output DIE references.
17058 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17059 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17060 (output_function): Stream DECL_DEBUG_ARGS.
17061 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17062 Stream DECL_ABSTRACT_ORIGIN.
17063 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17064 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17065 DECL_CONTEXT for file-scope decls.
17066 * lto-streamer.h (struct dref_entry): Declare.
17067 (dref_queue): Likewise.
17068 * cfgexpand.c (pass_expand::execute): Do not call the
17069 outlining_inline_function hook here.
17070 * lto-wrapper.c (debug_obj): New global.
17071 (tool_cleanup): Unlink it if required.
17072 (debug_objcopy): New function.
17073 (run_gcc): Handle early debug sections in the IL files by
17074 extracting them to separate files, partially linkin them and
17075 feeding the result back as result to the linker.
17076 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17077 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17078 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17079 sections into a separate segment.
17080 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17082 (darwin_asm_dwarf_section): Likewise.
17083 (darwin_asm_output_dwarf_offset): Likewise.
17084 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17086 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17087 Alan Hayward <alan.hayward@arm.com>
17088 David Sherwood <david.sherwood@arm.com>
17090 * read-md.h (md_reader::record_potential_iterator_use): Replace
17091 pointer argument with an rtx and an index.
17092 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17093 (apply_mode_iterator): Likewise.
17094 (apply_code_iterator): Likewise.
17095 (apply_int_iterator): Likewise.
17096 (apply_subst_iterator): Likewise.
17097 (record_iterator_use): Likewise.
17098 (record_attribute_use): Likewise.
17099 (md_reader::record_potential_iterator_use): Likewise. Update calls
17100 to record_iterator_use and apply_iterator.
17101 (iterator_use): Replace ptr with x and index.
17102 (attribute_use): Likewise.
17103 (apply_attribute_uses): Update calls to apply_iterator.
17104 (apply_iterators): Likewise. Update initialization of iterator_use.
17105 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17106 and record_potential_iterator_use.
17107 (rtx_reader::read_rtx_operand): Likewise.
17109 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17110 Alan Hayward <alan.hayward@arm.com>
17111 David Sherwood <david.sherwood@arm.com>
17113 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17116 2017-08-21 Richard Biener <rguenther@suse.de>
17118 PR middle-end/81884
17119 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17120 at struct end conservatively when comparing common bases.
17122 2017-08-21 Richard Biener <rguenther@suse.de>
17124 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17125 (mem_ref_in_stmt): Remove.
17126 (determine_max_movement): Use ref index to get at the reference.
17127 (invariantness_dom_walker::before_dom_children): Deal with
17128 lim data already initialized.
17129 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17131 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17133 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17134 (TARGET_ISA_ROUND): Ditto.
17135 (TARGET_ROUND): Ditto.
17136 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17137 * config/i386/i386.md: Ditto.
17138 * config/i386/sse.md: Ditto.
17139 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17140 with OPTION_MASK_ISA_SSE4_1.
17142 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17145 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17146 __builtin_ia32_lzcnt_u16.
17148 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17151 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17152 (rs6000_set_current_function): Rewrite function to use it.
17154 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17157 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17158 and TYPE_WARN_IF_NOT_ALIGN.
17159 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17160 (handle_warn_if_not_align): New.
17161 (place_union_field): Call handle_warn_if_not_align.
17162 (place_field): Call handle_warn_if_not_align.
17163 Copy TYPE_WARN_IF_NOT_ALIGN.
17164 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17165 (layout_type): Likewise.
17166 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17168 (tree_decl_common): Add warn_if_not_align.
17169 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17170 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17171 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17172 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17173 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17174 * doc/extend.texi: Document warn_if_not_aligned attribute.
17175 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17177 2017-08-17 Martin Liska <mliska@suse.cz>
17180 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17181 (get_data_dependence): Use it as pointer type.
17182 (distribute_loop): Likewise.
17184 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17186 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17187 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17188 (p8_vmrgow_<mode>_direct): New define_insn.
17189 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17190 handle endianness for vmrgew and vmrgow permute patterns.
17192 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17194 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17195 * config/rs6000/rs6000-cpus.def: Remove comment.
17196 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17197 (POWERPC_MASKS): Likewise.
17198 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17199 use of TARGET_VSX_TIMODE.
17200 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17201 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17202 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17203 (rs6000_option_override_internal): Remove dead code.
17204 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17205 (rs6000_legitimize_reload_address): Likewise.
17206 (rs6000_legitimate_address_p): Likewise.
17207 (rs6000_opt_masks): Delete "vsx-timode".
17208 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17209 from function comment.
17210 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17211 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17212 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17214 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17215 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17216 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17217 (VSX_TI): Likewise.
17219 (define_peephole2): Likewise.
17221 2017-08-17 Martin Sebor <msebor@redhat.com>
17224 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17225 past the end of an array.
17226 (test_pp_format): Add test cases.
17228 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17230 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17231 missing ECF_NOTHROW flags.
17233 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17236 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17237 operands residing in integer registers.
17238 (*vsx_le_perm_load_<mode>): Likewise.
17239 (*vsx_le_perm_store_<mode>): Likewise.
17240 (define_peephole2): Add peepholes to optimize the above.
17242 2017-08-17 Marek Polacek <polacek@redhat.com>
17244 PR middle-end/81814
17245 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17246 to mimic what shorten_compare did. Change the return type to bool.
17247 (fold_cond_expr_with_comparison): Update call to
17248 operand_equal_for_comparison_p.
17249 (fold_ternary_loc): Likewise.
17251 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17253 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17255 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17257 2017-08-17 Richard Biener <rguenther@suse.de>
17259 * tree-ssa-structalias.c (solve_graph): When propagating
17260 to successors update the graphs succ edges and avoid duplicate work.
17262 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17265 * config/i386/i386.c (ix86_option_override_internal): Save target
17266 specific options after ix86_stack_protector_guard_reg was changed.
17268 2017-08-17 Richard Biener <rguenther@suse.de>
17270 PR tree-optimization/81827
17271 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17272 (new_var_info): Initialize it conservatively.
17273 (get_call_vi): Mark register vars.
17274 (new_scalar_tmp_constraint_exp): Likewise.
17275 (handle_rhs_call): Likewise.
17276 (handle_const_call): Likewise.
17277 (create_function_info_for): Likewise.
17278 (solve_constraints): Sort varinfos to separate register from
17279 non-register vars to pack points-to solution bitmaps during
17282 2017-08-17 Marek Polacek <polacek@redhat.com>
17284 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17286 2017-08-17 Richard Biener <rguenther@suse.de>
17288 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17289 to true when overflow is undefined and we saturated the result.
17291 2017-08-17 Alan Modra <amodra@gmail.com>
17294 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17295 Don't use store multiple if only one reg needs saving.
17296 (interesting_frame_related_regno): New function.
17297 (rs6000_frame_related): Don't emit frame info for regs that
17299 (rs6000_emit_epilogue): Likewise.
17301 2017-08-16 Nathan Sidwell <nathan@acm.org>
17303 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17304 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17305 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17306 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17307 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17308 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17309 (verify_type): Adjust for TYPE_BINFO move.
17310 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17311 process TYPE_BINFO directly.
17312 (hash_tree): Likewise.
17313 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17315 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17318 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17320 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17322 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17325 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17326 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17327 Add ix86_binary_operator_ok to insn constraint.
17328 (*iordi_1_bts): Ditto.
17329 (*xordi_1_btc): Ditto.
17330 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17331 Update corresponding peephole2 pattern.
17335 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17337 PR tree-optimization/81832
17338 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17339 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17341 2017-08-16 Marek Polacek <polacek@redhat.com>
17344 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17345 regarding size_zero_node.
17347 2017-08-16 Martin Liska <mliska@suse.cz>
17350 * config.gcc: Respect previously set extra_objs in case
17353 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17355 PR tree-optimization/81835
17356 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17357 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17358 not depend on the phi.
17360 2017-08-16 Alan Modra <amodra@gmail.com>
17362 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17365 2017-08-16 Alan Modra <amodra@gmail.com>
17367 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17368 (save_reg_p): ..into this. Update all callers.
17369 (first_reg_to_save): Simplify.
17371 2017-08-16 Alan Modra <amodra@gmail.com>
17373 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17376 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17380 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17381 constant count if that count is less than 32.
17383 2017-08-15 Nathan Sidwell <nathan@acm.org>
17385 * gcc.c (execute): Emit friendlier message if inferior is killed
17386 by an external cause.
17388 2017-08-15 Richard Biener <rguenther@suse.de>
17390 PR tree-optimization/81790
17391 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17392 CONSTRUCTORs from simplifying and VN.
17394 2017-08-14 Martin Sebor <msebor@redhat.com>
17396 * builtin-attrs.def: Add comments.
17398 2017-08-14 Martin Sebor <msebor@redhat.com>
17401 * doc/extend.texi (attribute nonstring): Document new attribute.
17403 2017-08-14 Martin Sebor <msebor@redhat.com>
17406 * tree-diagnostic.c (default_tree_printer): Handle %G.
17407 * gimple-pretty-print.h (percent_G_format): Declare new function.
17408 * gimple-pretty-print.c (percent_G_format): Define.
17409 * tree-pretty-print.c (percent_K_format): Add argument.
17411 2017-08-14 Martin Sebor <msebor@redhat.com>
17413 PR translation/79998
17414 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17415 Remove a stray space.
17417 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17420 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17421 (*iordi_1_bts): Ditto.
17422 (*xordi_1_btc): Ditto.
17424 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17427 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17428 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17430 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17431 quoted strings, and make more translator-friendly.
17432 (darwin_rs6000_override_options): Likewise.
17433 (rs6000_option_override_internal): Likewise.
17434 (rs6000_return_in_memory): Fix overlong line.
17435 (init_cmulative_args): Use quoted strings, and make more
17436 translator-friendly.
17437 (rs6000_pass_by_reference): Fix overlong line.
17438 (def_builtin): Use quoted strings.
17439 (altivec_expand_predicate_builtin): Use quoted strings, and make
17440 more translator-friendly.
17441 (htm_expand_builtin): Use quoted strings.
17442 (cpu_expand_builtin): Use quoted strings, and make more
17443 translator-friendly.
17444 (altivec_expand_builtin): Likewise.
17445 (paired_expand_predicate_builtin): Likewise.
17446 (rs6000_invalid_builtin): Likewise.
17447 (builtin_function_type): Use quoted strings.
17448 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17449 more translator-friendly.
17450 (rs6000_trampoline_init): Likewise.
17451 (rs6000_handle_altivec_attribute): Likewise.
17452 (rs6000_inner_target_options): Use quoted strings.
17453 (rs6000_disable_incompatible_switches): Likewise.
17454 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17455 strings, and make more translator-friendly.
17456 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17458 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17460 PR tree-optimization/81799
17461 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17462 cond_expr to simple gimple operand.
17464 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17466 PR middle-end/46932
17467 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17469 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17473 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17474 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17475 TARGET_GASISR_PROLOGUES.
17476 * config/avr/avr.c (avr_option_override): Same.
17477 (avr_pass_pre_proep::execute): Same.
17479 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17482 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17483 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17485 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17487 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17488 define_insn_and_split. Split to a memory load from 0 in
17489 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17490 using PTR mode iterator.
17491 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17492 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17493 (*add_tp_<mode>): Redefine as define_insn_and_split.
17494 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17495 address space. Merge with *add_tp_x32 using PTR mode iterator.
17496 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17497 Split to an add with a memory load from 0 in
17498 DEFAULT_TLS_SEG_REG address space.
17500 2017-08-12 Andrew Pinski <apinski@cavium.com>
17502 * config/aarch64/aarch64-option-extensions.def (rdma):
17503 Fix feature string to what Linux prints out in /proc/cpuinfo.
17505 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17508 * dwarf2out.c (modified_type_die): For C typedef types that have
17509 an ultimate origin, process the ultimate origin instead of the
17511 (gen_typedef_die): Assert that input DECLs have no ultimate
17513 (gen_type_die_with_usage): For typedef variants that have an
17514 ultimate origin, just call gen_decl_die on the original DECL.
17515 (process_scope_var): Avoid creating DIEs for local typedefs and
17516 concrete static variables.
17518 2017-08-12 Alan Modra <amodra@gmail.com>
17522 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17523 match gnu-user.h startfile.
17524 (ENDFILE_LINUX_SPEC): Similarly.
17526 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17529 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17531 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17533 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17534 * config/aarch64/aarch64.md (mov<mode>): Change.
17535 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17536 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17537 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17539 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17541 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17542 for storage order barriers.
17544 2017-08-11 Martin Liska <mliska@suse.cz>
17547 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17548 variables of void type.
17550 2017-08-11 Martin Liska <mliska@suse.cz>
17552 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17553 TARGET_SUPPORTS_ALIASES.
17554 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17555 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17556 (optimize_weakref): Likewise.
17557 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17558 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17559 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17561 2017-08-11 Martin Liska <mliska@suse.cz>
17564 * config/i386/i386.c (make_resolver_func): Do complete
17565 refactoring of the function.
17567 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17570 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17571 * config/i386/i386.c (ix86_stack_protect_guard): Use
17572 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17573 * doc/invoke.texi (x86 Options): Document
17574 -mstack-protector-guard-symbol= option.
17576 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17578 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17579 * config/i386/i386.c (ix86_split_stack_guard): New function.
17580 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17581 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17582 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17583 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17584 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17585 (split_stack_space_check): Call ix86_split_stack_guard.
17587 2017-08-10 Martin Sebor <msebor@redhat.com>
17589 * print-tree.c (print_node): Print location using the established
17591 Replace spaces with colons.
17592 (debug_raw, debug): Ditto.
17594 2017-08-10 Martin Sebor <msebor@redhat.com>
17597 * pretty-print.c (pp_format): Correct the handling of %s precision.
17599 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
17602 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17604 (ix86_finalize_stack_frame_flags): This. Also clear
17605 frame_pointer_needed if -fno-omit-frame-pointer is used without
17607 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17608 with ix86_finalize_stack_frame_flags.
17609 (ix86_expand_epilogue): Likewise.
17610 (ix86_expand_split_stack_prologue): Likewise.
17611 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
17613 2017-08-10 Martin Liska <mliska@suse.cz>
17616 * c-attribs.c (handle_target_attribute):
17617 Report warning for an empty string argument of target attribute.
17619 2017-08-09 Jakub Jelinek <jakub@redhat.com>
17622 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
17624 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
17625 or DECL_NONLOCAL labels.
17626 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
17627 or DECL_NONLOCAL labels here.
17629 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
17631 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
17632 to indicate when early gimple folding has been disabled.
17633 (rs6000_gimple_fold_builtin): Add debug content.
17634 (rs6000_invalid_builtin): Fix whitespace.
17635 (rs6000_expand_builtin): Fix whitespace.
17636 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
17638 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
17641 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
17642 SAVE_MULTIPLE if not all the registers that saves, should be saved.
17644 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
17646 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
17647 (qdf24xx): Likewise.
17648 * config/aarch64/aarch64.md: Include falkor.md.
17649 * config/aarch64/falkor.md: New.
17651 2017-08-09 Marek Polacek <polacek@redhat.com>
17654 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
17655 * diagnostic.c (emit_diagnostic): Add a comment.
17656 (emit_diagnostic_valist): New function.
17658 2017-08-09 Marek Polacek <polacek@redhat.com>
17661 * input.c (make_location): New overload.
17662 * input.h (make_location): Declare.
17664 2017-08-08 Alan Modra <amodra@gmail.com>
17665 H.J. Lu <hongjiu.lu@intel.com>
17668 * gcc.c (NO_PIE_SPEC): Delete.
17669 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
17671 (LINK_PIE_SPEC): ..to here.
17672 (LINK_COMMAND_SPEC): Support -no-pie.
17673 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
17674 chain of crtbegin*.o selection, update for PIE_SPEC changes and
17676 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
17677 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
17678 (ENDFILE_CRTEND_SPEC): Similarly.
17680 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
17683 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
17684 (mstack-protector-guard-offset=): Ditto.
17685 * config/i386/i386.c (ix86_option_override): Handle
17686 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
17688 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
17689 ix86_stack_protect_guard_offset variables.
17690 (TARGET_STACK_PROTECT_GUARD): Always define.
17691 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
17692 and -mstack-protector-guard-offset= options.
17694 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17696 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
17697 boundary case for the last candidate.
17699 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17701 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
17702 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
17704 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17706 PR middle-end/19706
17707 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
17708 * config/aarch64/aarch64-builtins.c
17709 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
17710 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
17711 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
17713 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17714 Andrew Pinski <pinskia@gmail.com>
17716 PR middle-end/19706
17717 * internal-fn.def (XORSIGN): New.
17718 * optabs.def (xorsign_optab): New.
17719 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
17720 (convert_expand_mult_copysign): New.
17721 (pass_optimize_widening_mul::execute): Call
17722 convert_expand_mult_copysign.
17724 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17726 PR tree-optimization/81354
17727 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
17728 Insert on edges rather than explicitly creating landing pads.
17729 (analyze_candidates_and_replace): Commit edge inserts.
17731 2017-08-08 Richard Biener <rguenther@suse.de>
17733 PR middle-end/81719
17734 * tree-ssa-loop-niter.c: Include tree-dfa.h.
17735 (expand_simple_operations): Also look through ADDR_EXPRs with
17736 MEM_REF bases treating them as POINTER_PLUS_EXPR.
17738 2017-08-08 Richard Biener <rguenther@suse.de>
17740 PR tree-optimization/81723
17741 * tree-vect-slp.c (struct bst_traits): New hash traits.
17742 (bst_fail): New global.
17743 (vect_build_slp_tree_2): New worker, split out from ...
17744 (vect_build_slp_tree): ... this now wrapping it with using
17745 bst_fail set to cache SLP tree build fails. Properly handle
17747 (vect_analyze_slp_instance): Allocate and free bst_fail.
17749 2017-08-08 Martin Liska <mliska@suse.cz>
17752 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
17753 LABEL_DECLs that can be from a different function.
17755 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17757 PR tree-optimization/81744
17758 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
17759 loop's number of iterations.
17761 2017-08-08 Martin Liska <mliska@suse.cz>
17763 * asan.c: Include header files.
17764 * attribs.c (build_decl_attribute_variant): New function moved
17766 (build_type_attribute_qual_variant): Likewise.
17767 (cmp_attrib_identifiers): Likewise.
17768 (simple_cst_list_equal): Likewise.
17769 (omp_declare_simd_clauses_equal): Likewise.
17770 (attribute_value_equal): Likewise.
17771 (comp_type_attributes): Likewise.
17772 (build_type_attribute_variant): Likewise.
17773 (lookup_ident_attribute): Likewise.
17774 (remove_attribute): Likewise.
17775 (merge_attributes): Likewise.
17776 (merge_type_attributes): Likewise.
17777 (merge_decl_attributes): Likewise.
17778 (merge_dllimport_decl_attributes): Likewise.
17779 (handle_dll_attribute): Likewise.
17780 (attribute_list_equal): Likewise.
17781 (attribute_list_contained): Likewise.
17782 * attribs.h (lookup_attribute): New function moved from tree.[ch].
17783 (lookup_attribute_by_prefix): Likewise.
17784 * bb-reorder.c: Include header files.
17785 * builtins.c: Likewise.
17786 * calls.c: Likewise.
17787 * cfgexpand.c: Likewise.
17788 * cgraph.c: Likewise.
17789 * cgraphunit.c: Likewise.
17790 * convert.c: Likewise.
17791 * dwarf2out.c: Likewise.
17792 * final.c: Likewise.
17793 * fold-const.c: Likewise.
17794 * function.c: Likewise.
17795 * gimple-expr.c: Likewise.
17796 * gimple-fold.c: Likewise.
17797 * gimple-pretty-print.c: Likewise.
17798 * gimple.c: Likewise.
17799 * gimplify.c: Likewise.
17800 * hsa-common.c: Likewise.
17801 * hsa-gen.c: Likewise.
17802 * internal-fn.c: Likewise.
17803 * ipa-chkp.c: Likewise.
17804 * ipa-cp.c: Likewise.
17805 * ipa-devirt.c: Likewise.
17806 * ipa-fnsummary.c: Likewise.
17807 * ipa-inline.c: Likewise.
17808 * ipa-visibility.c: Likewise.
17810 * lto-cgraph.c: Likewise.
17811 * omp-expand.c: Likewise.
17812 * omp-general.c: Likewise.
17813 * omp-low.c: Likewise.
17814 * omp-offload.c: Likewise.
17815 * omp-simd-clone.c: Likewise.
17816 * opts-global.c: Likewise.
17817 * passes.c: Likewise.
17818 * predict.c: Likewise.
17819 * sancov.c: Likewise.
17820 * sanopt.c: Likewise.
17821 * symtab.c: Likewise.
17822 * toplev.c: Likewise.
17823 * trans-mem.c: Likewise.
17824 * tree-chkp.c: Likewise.
17825 * tree-eh.c: Likewise.
17826 * tree-into-ssa.c: Likewise.
17827 * tree-object-size.c: Likewise.
17828 * tree-parloops.c: Likewise.
17829 * tree-profile.c: Likewise.
17830 * tree-ssa-ccp.c: Likewise.
17831 * tree-ssa-live.c: Likewise.
17832 * tree-ssa-loop.c: Likewise.
17833 * tree-ssa-sccvn.c: Likewise.
17834 * tree-ssa-structalias.c: Likewise.
17835 * tree-ssa.c: Likewise.
17836 * tree-streamer-in.c: Likewise.
17837 * tree-vectorizer.c: Likewise.
17838 * tree-vrp.c: Likewise.
17839 * tsan.c: Likewise.
17840 * ubsan.c: Likewise.
17841 * varasm.c: Likewise.
17842 * varpool.c: Likewise.
17843 * tree.c: Remove functions moved to attribs.[ch].
17844 * tree.h: Likewise.
17845 * config/aarch64/aarch64.c: Add attrs.h header file.
17846 * config/alpha/alpha.c: Likewise.
17847 * config/arc/arc.c: Likewise.
17848 * config/arm/arm.c: Likewise.
17849 * config/avr/avr.c: Likewise.
17850 * config/bfin/bfin.c: Likewise.
17851 * config/c6x/c6x.c: Likewise.
17852 * config/cr16/cr16.c: Likewise.
17853 * config/cris/cris.c: Likewise.
17854 * config/darwin.c: Likewise.
17855 * config/epiphany/epiphany.c: Likewise.
17856 * config/fr30/fr30.c: Likewise.
17857 * config/frv/frv.c: Likewise.
17858 * config/ft32/ft32.c: Likewise.
17859 * config/h8300/h8300.c: Likewise.
17860 * config/i386/winnt.c: Likewise.
17861 * config/ia64/ia64.c: Likewise.
17862 * config/iq2000/iq2000.c: Likewise.
17863 * config/lm32/lm32.c: Likewise.
17864 * config/m32c/m32c.c: Likewise.
17865 * config/m32r/m32r.c: Likewise.
17866 * config/m68k/m68k.c: Likewise.
17867 * config/mcore/mcore.c: Likewise.
17868 * config/microblaze/microblaze.c: Likewise.
17869 * config/mips/mips.c: Likewise.
17870 * config/mmix/mmix.c: Likewise.
17871 * config/mn10300/mn10300.c: Likewise.
17872 * config/moxie/moxie.c: Likewise.
17873 * config/msp430/msp430.c: Likewise.
17874 * config/nds32/nds32-isr.c: Likewise.
17875 * config/nds32/nds32.c: Likewise.
17876 * config/nios2/nios2.c: Likewise.
17877 * config/nvptx/nvptx.c: Likewise.
17878 * config/pa/pa.c: Likewise.
17879 * config/pdp11/pdp11.c: Likewise.
17880 * config/powerpcspe/powerpcspe.c: Likewise.
17881 * config/riscv/riscv.c: Likewise.
17882 * config/rl78/rl78.c: Likewise.
17883 * config/rx/rx.c: Likewise.
17884 * config/s390/s390.c: Likewise.
17885 * config/sh/sh.c: Likewise.
17886 * config/sol2.c: Likewise.
17887 * config/sparc/sparc.c: Likewise.
17888 * config/spu/spu.c: Likewise.
17889 * config/stormy16/stormy16.c: Likewise.
17890 * config/tilegx/tilegx.c: Likewise.
17891 * config/tilepro/tilepro.c: Likewise.
17892 * config/v850/v850.c: Likewise.
17893 * config/vax/vax.c: Likewise.
17894 * config/visium/visium.c: Likewise.
17895 * config/xtensa/xtensa.c: Likewise.
17897 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17900 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
17901 constraints since the -mupper-regs-* switches have been
17903 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
17904 into a vector from a double word element that was extracted from
17905 another vector, and eliminate extra XXPERMDI instructions.
17906 (vsx_concat_<mode>_2): Likewise.
17907 (vsx_concat_<mode>_3): Likewise.
17908 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
17909 concat to allow optimizing inserts from previous extracts.
17911 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
17913 * config/i386/i386.c (ix86_stack_protect_guard): Generate
17914 memory reference to a SSP offset in TLS address space.
17915 (ix86_print_operand) <case '@'>: Remove.
17916 (ix86_print_operand_punct_valid_p): Remove '@' code.
17917 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
17918 UNSPEC_SP_TLS_TEST.
17919 (stack_tls_protect_set_<mode>): Remove.
17920 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
17921 (stack_tls_protect_test_<mode>): Remove.
17922 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
17924 2017-08-07 Olivier Hainque <hainque@adacore.com>
17927 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
17929 2017-08-07 Douglas Rupp <rupp@adacore.com>
17931 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
17932 variable was referenced as multidir in command.
17934 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17937 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
17940 2017-08-07 Martin Liska <mliska@suse.cz>
17942 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
17943 * config/rl78/rl78.c: Add include of attribs.h.
17944 * config/sh/sh.c: Likewise.
17945 * config/v850/v850.c: Likewise.
17947 2017-08-07 Tom de Vries <tom@codesourcery.com>
17949 PR middle-end/78266
17950 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
17952 2017-08-07 Martin Liska <mliska@suse.cz>
17954 * config/mips/mips.c: Include attribs.h.
17956 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
17959 * doc/invoke.texi: Document change in behvaior for -Ofast for
17962 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
17964 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
17966 (aarch64_pop_regs): Likewise.
17967 (aarch64_gen_load_pair): Likewise.
17968 (aarch64_save_callee_saves): Likewise.
17969 (aarch64_restore_callee_saves): Likewise.
17971 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17973 * config/i386/i386.c: Revert the last change.
17975 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17978 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17980 (ix86_finalize_stack_frame_flags): This. Also clear
17981 frame_pointer_needed if -fno-omit-frame-pointer is used without
17983 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17984 with ix86_finalize_stack_frame_flags.
17985 (ix86_expand_epilogue): Likewise.
17986 (ix86_expand_split_stack_prologue): Likewise.
17988 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17991 * config/i386/i386.c (get_builtin_code_for_version): Set priority
17992 to P_AES for Westmere.
17994 2017-08-07 Jonathan Yong <10walls@gmail.com>
17996 * config/i386/mingw.opt (fset-stack-executable): Removed.
17997 * config/i386/cygming.opt (fset-stack-executable): Moved
17999 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18001 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18003 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18005 2017-08-07 Marek Polacek <polacek@redhat.com>
18007 PR middle-end/81737
18008 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18010 2017-08-07 Martin Liska <mliska@suse.cz>
18012 * attribs.h (canonicalize_attr_name): New function.
18013 (cmp_attribs): Move from c-format.c and adjusted.
18014 (is_attribute_p): Moved from tree.h.
18015 * tree-inline.c: Add new includes.
18016 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18017 (private_is_attribute_p): Remove.
18018 (private_lookup_attribute): Likewise.
18019 (private_lookup_attribute_by_prefix): Simplify.
18020 (remove_attribute): Use is_attribute_p.
18021 * tree.h: Remove removed declarations.
18023 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18025 PR middle-end/81698
18026 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18027 instead of computing it in the function. Formatting fix.
18028 (expand_case): Don't rely on default_edge being the first edge,
18029 clear it if removing it, pass default_edge to
18030 emit_case_dispatch_table.
18031 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18034 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18036 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18037 insn in the function, emit NOP after the insn.
18039 2017-08-06 Tom de Vries <tom@codesourcery.com>
18041 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18044 2017-08-06 Tom de Vries <tom@codesourcery.com>
18046 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18049 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18051 PR tree-optimization/57371
18052 * match.pd: New pattern.
18054 2017-08-04 Marek Polacek <polacek@redhat.com>
18056 PR middle-end/81695
18057 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18058 perform the computation in offset_int.
18060 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18062 PR tree-optimization/81136
18063 * tree-vectorizer.h: Include tree-hash-traits.h.
18064 (vec_base_alignments): New typedef.
18065 (vec_info): Add a base_alignments field.
18066 (vect_record_base_alignments): Declare.
18067 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18069 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18070 (create_data_ref): Add an is_conditional_in_stmt argument.
18071 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18072 the is_conditional_in_stmt field.
18073 (data_ref_loc): Add an is_conditional_in_stmt field.
18074 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18075 (find_data_references_in_stmt): Update call to create_data_ref.
18076 (graphite_find_data_references_in_stmt): Likewise.
18077 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18078 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18079 (vect_record_base_alignment): New function.
18080 (vect_record_base_alignments): Likewise.
18081 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18082 for nested statements even if we fail to compute a misalignment.
18083 Use pooled base alignments for unconditional references.
18084 (vect_find_same_alignment_drs): Compare base addresses instead
18086 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18087 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18089 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18091 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18092 Add an explicit name for the enum. Use auto_vec for slp_instances
18093 and grouped_stores.
18094 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18096 (_bb_vec_info): Add a constructor and destructor.
18097 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18098 (destroy_loop_vec_info): Delete.
18099 (vect_destroy_datarefs): Likewise.
18100 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18101 (vec_info::vec_info): New function.
18102 (vec_info::~vec_info): Likewise.
18103 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18104 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18105 destroy_loop_vec_info.
18106 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18107 (_loop_vec_info::_loop_vec_info): ...this.
18108 (destroy_loop_vec_info): Replace with...
18109 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18110 the stmt_vec_infos. Leave handling of vec_info information to its
18111 destructor. Remove explicit vector releases.
18112 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18113 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18114 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18115 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18116 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18117 (destroy_bb_vec_info): Replace with...
18118 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18119 information to its destructor.
18120 (vect_slp_analyze_bb_1): Use new and delete instead of
18121 new_bb_vec_info and destroy_bb_vec_info.
18122 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18125 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18127 * tree-data-ref.h (subscript): Add access_fn field.
18128 (data_dependence_relation): Add could_be_independent_p.
18129 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18130 (same_access_functions): Move to tree-data-ref.c.
18131 * tree-data-ref.c (ref_contains_union_access_p): New function.
18132 (access_fn_component_p): Likewise.
18133 (access_fn_components_comparable_p): Likewise.
18134 (dr_analyze_indices): Add a reference to access_fn_component_p.
18135 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18137 (constant_access_functions): Likewise.
18138 (add_other_self_distances): Likewise.
18139 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18140 (initialize_data_dependence_relation): Use XCNEW and remove
18141 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18142 of access functions that have the same type. Allow the
18143 subsequence to end with different bases in some circumstances.
18144 Record the chosen access functions in SUB_ACCESS_FN.
18145 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18146 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18147 (subscript_dependence_tester_1): Likewise dra and drb.
18148 (build_classic_dist_vector): Update calls accordingly.
18149 (subscript_dependence_tester): Likewise.
18150 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18151 DDR_COULD_BE_INDEPENDENT_P.
18152 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18153 comp_alias_ddrs instead of may_alias_ddrs.
18154 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18156 (vect_analyze_data_ref_dependence): Use it if
18157 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18158 distance vectors if that fails.
18159 (dependence_distance_ge_vf): New function.
18160 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18161 LOOP_VINFO_MAY_ALIAS_DDRS.
18163 2017-08-04 Richard Biener <rguenther@suse.de>
18165 PR middle-end/81705
18166 * fold-const.c (fold_binary_loc): Properly restrict
18167 minus_var0 && minus_var1 case when associating undefined overflow
18170 2017-08-04 Tom de Vries <tom@codesourcery.com>
18172 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18174 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18176 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18177 Don't start diagnostic messages with a capital letter.
18178 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18180 (rs6000_invalid_builtin): Likewise.
18181 (rs6000_trampoline_init): Likewise.
18183 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18186 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18187 after setting changeable df flags.
18189 2017-08-03 Richard Biener <rguenther@suse.de>
18191 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18192 up if the use is in USE - X.
18194 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18196 * toplev.c (dumpfile.h): New include.
18197 (internal_error_reentered): New static function. Use it...
18198 (internal_error_function): ...here to handle reentered internal_error.
18200 2017-08-03 Richard Biener <rguenther@suse.de>
18202 PR middle-end/81148
18203 * fold-const.c (split_tree): Add minus_var and minus_con
18204 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18205 here but always use minus_*.
18206 (associate_trees): Assert we never associate with MINUS_EXPR
18207 and NULL first operand. Do not recurse for PLUS_EXPR operands
18208 when associating as MINUS_EXPR either.
18209 (fold_binary_loc): Track minus_var and minus_con.
18211 2017-08-03 Tom de Vries <tom@codesourcery.com>
18214 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18215 ACCEL_COMPILER, apply finish_options on
18216 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18218 2017-08-03 Tom de Vries <tom@codesourcery.com>
18221 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18222 function_entry_patch_area_size > 0.
18224 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18227 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18228 instead of 10??LU, perform unit multiplication in wide_int,
18229 don't change alloc_object_size_limit if the limit is larger
18232 PR tree-optimization/81655
18233 PR tree-optimization/81588
18234 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18235 the case when ranges[i].low and high are 1 for unsigned type with
18238 PR middle-end/81052
18239 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18240 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18242 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18244 * tree-vrp.h: Add include guard.
18246 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18249 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18250 (ud2): New insn pattern.
18251 * config/i386/i386.c (ix86_expand_epilogue):
18252 For naked functions, generate ud2 instead of trap insn.
18254 2017-08-02 Marek Polacek <polacek@redhat.com>
18257 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18259 2017-08-02 Tom de Vries <tom@codesourcery.com>
18260 Cesar Philippidis <cesar@codesourcery.com>
18262 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18263 Add missing edge probabilities.
18265 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18267 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18268 Correct endianness.
18270 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18272 PR middle-end/79499
18273 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18274 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18275 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18276 prologue_seq sequences - if any.
18278 2017-08-02 Richard Biener <rguenther@suse.de>
18280 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18281 via vectors if supported, integer extracts via punning if supported
18282 or otherwise vector extracts.
18284 2017-08-02 Richard Biener <rguenther@suse.de>
18286 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18288 (bitmap_insert_into_set): ... this.
18290 2017-08-02 Richard Biener <rguenther@suse.de>
18292 PR tree-optimization/81633
18294 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18296 PR tree-optimization/71752
18297 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18299 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18301 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18302 (machine_function::call_ms2sysv_pad_out): Remove field.
18303 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18304 (ix86_compute_frame_layout): Likewise.
18306 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18309 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18310 attribute with interrupt attribute.
18312 2017-08-01 Andrew Pinski <apinski@cavium.com>
18314 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18315 BIT_INSERT_EXPR's operand 1
18316 to see if the types precision matches.
18318 2017-08-01 Martin Liska <mliska@suse.cz>
18320 PR middle-end/70140
18321 * builtins.c (expand_builtin_memcpy_args): Remove.
18322 (expand_builtin_memcpy): Call newly added function
18323 expand_builtin_memory_copy_args.
18324 (expand_builtin_memcpy_with_bounds): Likewise.
18325 (expand_builtin_mempcpy): Remove last argument.
18326 (expand_builtin_mempcpy_with_bounds): Likewise.
18327 (expand_builtin_memory_copy_args): New function created from
18328 expand_builtin_mempcpy_args with small modifications.
18329 (expand_builtin_mempcpy_args): Remove.
18330 (expand_builtin_stpcpy): Remove unused argument.
18331 (expand_builtin): Likewise.
18332 (expand_builtin_with_bounds): Likewise.
18334 2017-08-01 Martin Liska <mliska@suse.cz>
18337 Make mempcpy more optimal (PR middle-end/70140).
18339 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18342 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18343 __builtin_vec_cmpne verify both arguments are compatible vectors
18344 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18345 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18346 move computation of aligned to after checking the argument types.
18350 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18351 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18354 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18355 Doug Rupp <rupp@adacore.com>
18356 Olivier Hainque <hainque@adacore.com>
18358 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18359 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18361 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18362 for TARGET_OS_CPP_BUILTIN.
18363 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18364 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18366 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18367 passing required abi options to the assembler for EABI configurations.
18368 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18369 of .text.hot and .text.unlikely sections for kernel modules when
18370 using ARM style exceptions.
18371 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18372 options. Add EXTRA_CC1_SPEC.
18373 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18375 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18377 (ARM_TARGET2_DWARF_FORMAT): Define.
18378 * config/arm/t-vxworks: Adjust multilib control to removal of the
18379 Diab command line options.
18381 2017-08-01 Martin Liska <mliska@suse.cz>
18383 PR gcov-profile/81561
18384 * gcov.c (unblock): Make unblocking safe as we need to preserve
18385 index correspondence of blocks and block_lists.
18387 2017-08-01 Richard Biener <rguenther@suse.de>
18389 PR tree-optimization/81181
18390 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18391 (compute_antic): ... end of iteration here.
18393 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18395 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18396 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18397 (ftree-slp-vectorize): Likewise.
18398 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18399 can no longer be set independent of flag_tree_loop_vectorize.
18400 * omp-general.c (emp_max_vf): Likewise.
18401 * opts.c (enable_fdo_optimizations): Remove references to
18402 flag_tree_vectorize, these are now implicit.
18403 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18404 and leave it for the options machinery.
18406 2017-08-01 Martin Liska <mliska@suse.cz>
18408 PR middle-end/70140
18409 * builtins.c (expand_builtin_memcpy_args): Remove.
18410 (expand_builtin_memcpy): Call newly added function
18411 expand_builtin_memory_copy_args.
18412 (expand_builtin_memcpy_with_bounds): Likewise.
18413 (expand_builtin_mempcpy): Remove last argument.
18414 (expand_builtin_mempcpy_with_bounds): Likewise.
18415 (expand_builtin_memory_copy_args): New function created from
18416 expand_builtin_mempcpy_args with small modifications.
18417 (expand_builtin_mempcpy_args): Remove.
18418 (expand_builtin_stpcpy): Remove unused argument.
18419 (expand_builtin): Likewise.
18420 (expand_builtin_with_bounds): Likewise.
18422 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18425 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18426 print "ds:" only for immediates in generic address space.
18428 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18431 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18432 (ix86_function_ok_for_sibcall): Return false for naked functions.
18434 2017-08-01 Richard Biener <rguenther@suse.de>
18436 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18437 (compute_antic): Seed worklist with exit block predecessors.
18438 * cfganal.c (dfs_find_deadend): For a cycle return the source
18439 of the edge closing it.
18441 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18443 * config/aarch64/aarch64.c
18444 (aarch64_can_const_movi_rtx_p): Move 0 check.
18446 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18448 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18449 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18451 * match.pd: Ditto in address comparison pattern.
18453 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18455 PR tree-optimization/81627
18456 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18457 closed ssa form for store-store chain.
18459 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18461 PR tree-optimization/81620
18462 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18463 for store-store chain.
18465 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18467 PR tree-optimization/81588
18468 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18469 ranges[i].in_p, invert comparison code ccode. For >/>=,
18470 swap rhs1 and rhs2 and comparison code unconditionally,
18471 for </<= don't do that. Don't swap rhs1/rhs2 again if
18472 ranges[i].in_p, instead invert comparison code ccode if
18473 opcode or oe->rank is BIT_IOR_EXPR.
18476 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18477 a direct optab to conversion optab.
18478 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18479 with GET_MODE_INNER as last argument instead of optab_handler.
18480 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18481 vector extraction if possible and optab is available.
18482 * expr.c (store_constructor): Use convert_optab_handler instead
18483 of optab_handler. Use vector initialization from smaller
18484 vectors if possible and optab is available.
18485 * tree-vect-stmts.c (vectorizable_load): Likewise.
18486 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18487 now have two modes.
18488 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18489 of vec_init from half-sized vectors with the same element mode.
18490 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18491 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18492 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18493 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18494 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18495 after mode in gen_vec_extract* calls.
18496 (vec_extract<mode>): Renamed to ...
18497 (vec_extract<mode><ssescalarmodelower>): ... this.
18498 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18499 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18500 element mode after mode in gen_vec_init* calls.
18501 (VEC_INIT_HALF_MODE): New mode iterator.
18502 (vec_init<mode>): Renamed to ...
18503 (vec_init<mode><ssescalarmodelower>): ... this.
18504 (vec_init<mode><ssehalfvecmodelower>): New expander.
18505 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18506 (vec_extractv2sfsf): ... this.
18507 (vec_initv2sf): Renamed to ...
18508 (vec_initv2sfsf): ... this.
18509 (vec_extractv2si): Renamed to ...
18510 (vec_extractv2sisi): ... this.
18511 (vec_initv2si): Renamed to ...
18512 (vec_initv2sisi): ... this.
18513 (vec_extractv4hi): Renamed to ...
18514 (vec_extractv4hihi): ... this.
18515 (vec_initv4hi): Renamed to ...
18516 (vec_initv4hihi): ... this.
18517 (vec_extractv8qi): Renamed to ...
18518 (vec_extractv8qiqi): ... this.
18519 (vec_initv8qi): Renamed to ...
18520 (vec_initv8qiqi): ... this.
18521 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18522 (vec_init<mode>): Renamed to ...
18523 (vec_init<mode><VEC_base_l>): ... this.
18524 (vec_extract<mode>): Renamed to ...
18525 (vec_extract<mode><VEC_base_l>): ... this.
18526 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18527 (vec_initv2sfsf): ... this.
18528 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18529 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18530 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18531 element mode after mode in gen_vec_init* calls.
18532 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18533 (vec_init<mode><Vel>): ... this.
18534 (vec_extract<mode>): Renamed to ...
18535 (vec_extract<mode><Vel>): ... this.
18536 * config/aarch64/iterators.md (Vel): New mode attribute.
18537 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18538 Add element mode after mode in gen_vec_extract* calls.
18539 * config/s390/vector.md (non_vec_l): New mode attribute.
18540 (vec_extract<mode>): Renamed to ...
18541 (vec_extract<mode><non_vec_l>): ... this.
18542 (vec_init<mode>): Renamed to ...
18543 (vec_init<mode><non_vec_l>): ... this.
18544 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18545 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18547 * config/arm/iterators.md (V_elem_l): New mode attribute.
18548 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18549 (vec_extract<mode><V_elem_l>): ... this.
18550 (vec_extractv2di): Renamed to ...
18551 (vec_extractv2didi): ... this.
18552 (vec_init<mode>): Renamed to ...
18553 (vec_init<mode><V_elem_l>): ... this.
18554 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18555 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18556 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18557 Add element mode after gen_vec_extract* calls.
18558 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18559 (vec_init<mode><unitmode>): ... this.
18560 (vec_extract<mode>): Renamed to ...
18561 (vec_extract<mode><unitmode>): ... this.
18562 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18563 (vec_init<mode><unitmode>): ... this.
18564 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18565 (vec_initv2sfsf): ... this.
18566 (vec_extractv2sf): Renamed to ...
18567 (vec_extractv2sfsf): ... this.
18568 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18569 Add element mode after gen_vec_extract* calls.
18570 * config/mips/mips.md (unitmode): New mode iterator.
18571 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18572 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18573 * config/spu/spu.md (inner_l): New mode attribute.
18574 (vec_init<mode>): Renamed to ...
18575 (vec_init<mode><inner_l>): ... this.
18576 (vec_extract<mode>): Renamed to ...
18577 (vec_extract<mode><inner_l>): ... this.
18578 * config/sparc/sparc.md (veltmode): New mode iterator.
18579 (vec_init<VMALL:mode>): Renamed to ...
18580 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18581 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18582 (vec_initv2sisi): ... this.
18583 (vec_initv2sf): Renamed to ...
18584 (vec_initv2sfsf): ... this.
18585 (vec_extractv2sf): Renamed to ...
18586 (vec_extractv2sfsf): ... this.
18587 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18588 (vec_init<mode>): Renamed to ...
18589 (vec_init<mode><VEC_base_l>): ... this.
18590 (vec_extract<mode>): Renamed to ...
18591 (vec_extract<mode><VEC_base_l>): ... this.
18592 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18593 (vec_initv2sfsf): ... this.
18594 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
18595 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18596 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
18597 gen_vec_init* calls.
18599 2017-08-01 Richard Biener <rguenther@suse.de>
18601 PR tree-optimization/81297
18602 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
18603 TREE_OVERFLOW from INTEGER_CSTs.
18605 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
18607 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
18609 2017-07-31 Carl Love <cel@us.ibm.com>
18611 * config/rs6000/rs6000-c: Add support for built-in functions
18612 vector signed char vec_xl_be (signed long long, signed char *);
18613 vector unsigned char vec_xl_be (signed long long, unsigned char *);
18614 vector signed int vec_xl_be (signed long long, signed int *);
18615 vector unsigned int vec_xl_be (signed long long, unsigned int *);
18616 vector signed long long vec_xl_be (signed long long, signed long long *);
18617 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
18618 vector signed short vec_xl_be (signed long long, signed short *);
18619 vector unsigned short vec_xl_be (signed long long, unsigned short *);
18620 vector double vec_xl_be (signed long long, double *);
18621 vector float vec_xl_be (signed long long, float *);
18622 * config/rs6000/altivec.h (vec_xl_be): Add #define.
18623 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
18624 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
18626 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
18627 (altivec_expand_builtin): Add switch statement to call
18628 altivec_expand_xl_be for each builtin.
18629 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
18630 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
18631 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
18632 __builtin_vsx_le_be_v16qi.
18633 * doc/extend.texi: Update the built-in documentation file for the
18634 new built-in functions.
18636 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18639 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
18641 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18643 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18645 * config.gcc: Add z14.
18646 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
18647 CPU model numbers for z13s and z14.
18648 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
18650 * config/s390/s390-opts.h (enum processor_type): Rename
18651 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18652 * config/s390/s390.c (processor_table): Add field for CPU name to
18653 be passed to Binutils.
18654 (s390_asm_output_machine_for_arch): Use the new field in
18655 processor_table for Binutils.
18656 (s390_expand_builtin): Replace arch12 with z14.
18657 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18658 (s390_get_sched_attrmask): Likewise.
18659 (s390_get_unit_mask): Likewise.
18660 * config/s390/s390.opt: Add z14 to processor_type enum.
18662 2017-07-31 Martin Jambor <mjambor@suse.cz>
18665 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
18666 regardless of optimization level.
18668 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
18669 Martin Liska <mliska@suse.cz>
18671 * predict.def: Remove old comment and adjust probability.
18672 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
18673 PREDICT statements.
18675 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18678 * config/i386/i386.c (ix86_function_naked): New function.
18679 (ix86_can_use_return_insn_p): Return false for naked functions.
18680 (ix86_expand_prologue): Skip prologue for naked functions.
18681 (ix86_expand_epilogue): Skip epilogue for naked functions
18682 and emit trap instruction.
18683 (ix86_warn_func_return): New function.
18684 (ix86_attribute_table): Add "naked" attribute specification.
18685 (TARGET_WARN_FUNC_RETURN): Define.
18686 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
18688 2017-07-31 Martin Liska <mliska@suse.cz>
18690 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
18691 (dump_gimple_bb_header): Always dump BB info.
18692 (pp_cfg_jump): Do not append info about BB when dumping a jump.
18694 2017-07-31 Martin Liska <mliska@suse.cz>
18697 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
18698 also with current_function_decl non-null equality.
18700 2017-07-31 Jakub Jelinek <jakub@redhat.com>
18703 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
18704 change type to the element type, instead add eltype variable and
18705 use it where we are interested in the element type.
18707 PR tree-optimization/81603
18708 * ipa-polymorphic-call.c
18709 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
18710 offset arithmetic in offset_int, bail out if the resulting bit offset
18711 doesn't fit into shwi.
18713 2017-07-31 Martin Liska <mliska@suse.cz>
18715 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
18716 (gimplify_save_expr): Fix comment.
18718 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
18721 * config/i386/i386.c (ix86_function_arg): Update arguments for
18723 (ix86_compute_frame_layout): Set the initial stack offset to
18724 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
18725 INCOMING_FRAME_SP_OFFSET.
18726 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
18727 stack before exception handler returns.
18728 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
18729 the 'ERROR_CODE' for exception handler.
18731 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
18733 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
18734 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
18735 (ASM_OUTPUT_REG_POP): Ditto.
18736 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
18737 instead of asm_fprintf to output pure string.
18739 2017-07-29 Jakub Jelinek <jakub@redhat.com>
18741 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
18742 to imported_module_or_decl hook.
18743 (debug_nothing_tree_tree_tree_bool): Remove.
18744 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
18745 * debug.c (do_nothing_debug_hooks): Use
18746 debug_nothing_tree_tree_tree_bool_bool instead of
18747 debug_nothing_tree_tree_tree_bool.
18748 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18749 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
18750 * sdbout.c (sdb_debug_hooks): Likewise.
18751 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
18752 (gen_namespace_die): Add DW_AT_export_symbols attribute if
18754 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
18755 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
18756 attribute, don't add anything.
18758 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18760 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
18761 (fold_build2_stat_loc): Likewise.
18762 (fold_build3_stat_loc): Likewise.
18763 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
18764 (fold_build1_loc): Remove macro.
18765 (fold_build2_loc): Likewise.
18766 (fold_build3_loc): Likewise.
18768 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18770 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
18771 (gimple_build_debug_bind_source_stat): Likewise.
18772 * gimple.h (gimple_build_debug_bind): Remove macro.
18773 (gimple_build_debug_bind_source): Likewise.
18775 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18777 * bitmap.c (bitmap_alloc): Adjust.
18778 (bitmap_gc_alloc): Likewise.
18779 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
18781 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18783 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
18784 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
18785 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
18786 (bitmap_gc_alloc_stat): Likewise.
18787 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
18789 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18791 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
18792 * rtl.h (shallow_copy_rtx): Remove macro.
18794 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18796 * emit-rtl.c (gen_raw_REG): Adjust.
18797 * gengenrtl.c (gendef): Likewise.
18798 * rtl.c (rtx_alloc_stat): Remove _stat from name.
18799 * rtl.h (rtx_alloc): Remove macro.
18801 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18803 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
18804 (build_tree_list_stat): Likewise.
18805 * tree.h (build_tree_list): Remove macro.
18806 (build_tree_list_vec): Likewise.
18808 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18810 * tree.c (make_vector_stat): Remove _stat from name.
18811 (build_vector_stat): Likewise.
18812 * tree.h (make_vector_stat): Remove macro.
18813 (build_vector_stat): Likewise.
18815 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18817 * tree.h (build_var_debug_value): Remove prototype.
18819 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18821 * tree.c (tree_cons_stat): Remove _stat from name.
18822 * tree.h (tree_cons): Remove macro.
18824 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18826 * tree.c (build_vl_exp_stat): Remove _stat from name.
18827 * tree.h (build_vl_exp): Remove macro.
18829 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18831 * tree.c (build_decl_stat): Remove _stat from name.
18832 * tree.h (build_decl): Remove macro.
18834 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18836 * gimple.c (gimple_build_with_ops_stat): Adjust.
18837 (gimple_alloc_stat): Remove _stat from name.
18838 * gimple.h (gimple_alloc): Remove macro.
18840 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18842 * tree.c (make_tree_vec_stat): Remove _stat from name.
18843 (grow_tree_vec_stat): Likewise.
18844 * tree.h (make_tree_vec_stat): Adjust prototype.
18845 (grow_tree_vec_stat): Likewise.
18846 (make_tree_vec): Remove macro.
18847 (grow_tree_vec): Likewise.
18849 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18851 * fold-const.c (fold_build1_stat_loc): Adjust.
18852 (fold_build2_stat_loc): Likewise.
18853 (fold_build3_stat_loc): Likewise.
18854 * tree.c (build0_stat): Remove _stat from name.
18855 (build1_stat): Likewise.
18856 (build2_stat): Likewise.
18857 (build3_stat): Likewise.
18858 (build4_stat): Likewise.
18859 (build5_stat): Likewise.
18860 * tree.h (build1_loc): Remove macro, and rename _stat function
18862 (build2_loc): Likewise.
18863 (build3_loc): Likewise.
18864 (build4_loc): Likewise.
18865 (build5_loc): Likewise.
18867 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18869 * tree.c (make_int_cst_stat): Remove _stat from name.
18870 * tree.h (make_int_cst_stat): Adjust prototype.
18871 (make_int_cst): Remove macro.
18873 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18875 * tree.c (make_tre_binfo_stat): Remove _stat from name.
18876 * tree.h (make_tree_binfo_stat): Adjust prototype.
18877 (make_tree_binfo): Remove.
18879 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18881 * tree.c (copy_node_stat): Rename to copy_node.
18882 (build_distinct_type_copy): Adjust.
18883 * tree.h (copy_node_stat): Adjust prototype.
18884 (copy_node): Remove macro.
18886 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18888 * tree.c (make_node_stat): rename to make_node.
18889 (build_tree_list_stat): Adjust.
18890 (build0_stat): Likewise.
18891 (build2_stat): Likewise.
18892 (build3_stat): Likewise.
18893 (build4_stat): Likewise.
18894 (build5_stat): Likewise.
18895 (build_decl_stat): Likewise.
18896 * tree.h (make_node_stat): Adjust prototype.
18897 (make_node): remove macro.
18899 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
18901 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
18902 (PPC_FEATURE2_SCV): Likewise.
18903 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
18905 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18907 * config/aarch64/aarch64.c
18908 (aarch64_internal_mov_immediate): Add new special pattern.
18909 * config/aarch64/aarch64.md (*movdi_aarch64):
18910 Add reg/32bit const mov case.
18912 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18913 Richard Sandiford <richard.sandiford@linaro.org>
18915 * config/aarch64/aarch64.md (mov<mode>): Generalize.
18916 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18917 Add integer and movi cases.
18918 (movi-split-hf-df-sf split, fp16): New.
18919 (enabled): Added TARGET_FP_F16INST.
18920 * config/aarch64/iterators.md (GPF_HF): New.
18921 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
18923 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18925 * config/aarch64/aarch64.c
18926 (aarch64_simd_container_mode): Add prototype.
18927 (aarch64_expand_mov_immediate): Add HI support.
18928 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
18929 (aarch64_can_const_movi_rtx_p): New.
18930 (aarch64_preferred_reload_class):
18931 Remove restrictions of using FP registers for certain SIMD operations.
18932 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
18933 (aarch64_valid_floating_const): Add integer move validation.
18934 (aarch64_simd_imm_scalar_p): Remove.
18935 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
18936 (aarch64_legitimate_constant_p): Expand list of supported cases.
18937 * config/aarch64/aarch64-protos.h
18938 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
18939 (aarch64_reinterpret_float_as_int): New.
18940 (aarch64_simd_imm_scalar_p): Remove.
18941 * config/aarch64/constraints.md (Uvi): New.
18942 (Dd): Split into Ds and new Dd.
18943 * config/aarch64/aarch64.md (*movsi_aarch64):
18945 (*movdi_aarch64): Add SIMD mov case.
18947 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18949 * tree-predcom.c: (struct chain): Handle store-store chain in which
18950 stores for elimination only store loop invariant values.
18951 (execute_pred_commoning_chain): Ditto.
18952 (prepare_initializers_chain_store_elim): Ditto.
18953 (prepare_finalizers): Ditto.
18954 (is_inv_store_elimination_chain): New function.
18955 (initialize_root_vars_store_elim_1): New function.
18957 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18959 * tree-predcom.c: Revise general description of the pass.
18960 (enum chain_type): New enum type for store elimination.
18961 (struct chain): New field supporting store elimination.
18962 (struct component): Ditto.
18963 (dump_chain): Dump store-stores chain.
18964 (release_chain): Release resources.
18965 (split_data_refs_to_components): Compute and create component
18966 contains only stores for elimination.
18967 (get_chain_last_ref_at): New function.
18968 (make_invariant_chain): Initialization.
18969 (make_rooted_chain): Specify chain type in parameter and record it.
18970 (add_looparound_copies): Skip for store-stores chain.
18971 (determine_roots_comp): Compute type of chain and pass it to
18973 (initialize_root_vars_store_elim_2): New function.
18974 (finalize_eliminated_stores): New function.
18975 (remove_stmt): Handle store for elimination.
18976 (execute_pred_commoning_chain): Execute predictive commoning on
18977 store-store chains.
18978 (determine_unroll_factor): Skip unroll for store-stores chain.
18979 (prepare_initializers_chain_store_elim): New function.
18980 (prepare_initializers_chain): Hanlde store-store chain.
18981 (prepare_finalizers_chain, prepare_finalizers): New function.
18982 (tree_predictive_commoning_loop): Return integer value indicating
18983 if loop is unrolled or lcssa form is corrupted.
18984 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
18986 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18988 * tree-predcom.c (initialize_root): Delete.
18989 (execute_pred_commoning_chain): Initialize root vars and replace
18990 reference of non-combined chain directly, rather than call above
18993 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18995 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
18996 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
18998 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19000 * tree-predcom.c (struct chain): New field init_seq.
19001 (release_chain): Release init_seq.
19002 (prepare_initializers_chain): Record intialization stmts in above
19004 (insert_init_seqs): New function.
19005 (tree_predictive_commoning_loop): Call insert_init_seqs.
19007 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19009 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19011 2017-07-28 Richard Biener <rguenther@suse.de>
19013 * match.pd: Remove superfluous :c.
19014 * genmatch.c (simplify::id): Add member.
19015 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19017 (current_id): New global.
19018 (dt_node::parent): Move from ...
19019 (dt_operand::parent): ... here. Add for_id member.
19020 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19021 (decision_tree::find_node): Relax order requirement when
19022 merging DT_TRUE nodes to ones inbetween the current simplify
19023 and the one we try to merge with. Add diagnostic whenever
19024 we need to enforce pattern order by not merging.
19025 (decision_tree::insert): Set current_id.
19026 (decision_tree::print_node): Dump parent node and for_id.
19027 (parser::last_id): Add member.
19028 (parser::push_simplify): Assign unique id.
19029 (parser::parser): Initialize last_id.
19031 2017-07-28 Martin Liska <mliska@suse.cz>
19034 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19035 gimple_build_debug_bind.
19037 2017-07-28 Richard Biener <rguenther@suse.de>
19039 PR tree-optimization/81502
19040 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19042 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19044 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19045 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19046 for BIT_FIELD_REF args.
19047 * fold-const.c (make_bit_field_ref): Likewise.
19048 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19050 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19053 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19054 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19055 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19056 Or it into SANITIZER_UNDEFINED.
19057 * ubsan.c: Include gimple-fold.h and varasm.h.
19058 (ubsan_expand_ptr_ifn): New function.
19059 (instrument_pointer_overflow): New function.
19060 (maybe_instrument_pointer_overflow): New function.
19061 (instrument_object_size): Formatting fix.
19062 (pass_ubsan::execute): Call instrument_pointer_overflow
19063 and maybe_instrument_pointer_overflow.
19064 * internal-fn.c (expand_UBSAN_PTR): New function.
19065 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19066 * sanitizer.def (__ubsan_handle_pointer_overflow,
19067 __ubsan_handle_pointer_overflow_abort): New builtins.
19068 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19069 * internal-fn.def (UBSAN_PTR): New internal function.
19070 * opts.c (sanitizer_opts): Add pointer-overflow.
19071 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19072 * fold-const.c (build_range_check): Compute pointer range check in
19073 integral type if pointer arithmetics would be needed. Formatting
19076 2017-07-28 Martin Liska <mliska@suse.cz>
19079 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19080 parameters that are of a variable-length.
19082 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19084 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19086 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19087 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19088 (CRT_CALL_STATIC_FUNCTION): Likewise.
19089 (ASM_DEFAULT_SPEC): New define.
19090 (ASM_SPEC32): Likewise.
19091 (ASM_SPEC64): Likewise.
19092 (ASM_SPEC_COMMON): Likewise.
19093 (ASM_SPEC): Likewise.
19094 (INVALID_64BIT): Likewise.
19095 (LINK_OS_DEFAULT_SPEC): Likewise.
19096 (LINK_OS_SPEC32): Likewise.
19097 (LINK_OS_SPEC64): Likewise.
19098 (POWERPC_LINUX): Likewise.
19099 (PTRDIFF_TYPE): Likewise.
19100 (RESTORE_FP_PREFIX): Likewise.
19101 (RESTORE_FP_SUFFIX): Likewise.
19102 (SAVE_FP_PREFIX): Likewise.
19103 (SAVE_FP_SUFFIX): Likewise.
19104 (SIZE_TYPE): Likewise.
19105 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19106 (TARGET_64BIT): Likewise.
19107 (TARGET_64BIT): Likewise.
19108 (TARGET_AIX): Likewise.
19109 (WCHAR_TYPE_SIZE): Likewise.
19110 (WCHAR_TYPE): Undefine.
19111 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19112 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19113 (CPP_OS_RTEMS_SPEC): Delete.
19114 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19115 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19117 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19119 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19121 PR tree-optimization/81578
19122 * tree-parloops.c (build_new_reduction): Bail out if
19123 reduction_code isn't one of the standard OpenMP reductions.
19124 Move the details printing after that decision.
19126 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19128 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19129 related to reload_in_progress.
19130 (splat_input_operand): Likewise.
19131 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19133 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19135 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19136 (TARGET_INSTANTIATE_DECLS): Likewise.
19137 (legitimate_indexed_address_p): Delete reload_in_progress code.
19138 (rs6000_debug_legitimate_address_p): Likewise.
19139 (rs6000_eliminate_indexed_memrefs): Likewise.
19140 (rs6000_emit_le_vsx_store): Likewise.
19141 (rs6000_emit_move_si_sf_subreg): Likewise.
19142 (rs6000_emit_move): Likewise.
19143 (register_to_reg_type): Likewise.
19144 (rs6000_pre_atomic_barrier): Likewise.
19145 (rs6000_machopic_legitimize_pic_address): Likewise.
19146 (rs6000_allocate_stack_temp): Likewise.
19147 (rs6000_address_for_fpconvert): Likewise.
19148 (rs6000_address_for_altivec): Likewise.
19149 (rs6000_secondary_memory_needed_rtx): Delete function.
19150 (rs6000_check_sdmode): Likewise.
19151 (rs6000_alloc_sdmode_stack_slot): Likewise.
19152 (rs6000_instantiate_decls): Likewise.
19153 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19154 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19155 Delete reload_in_progress.
19156 (*vec_reload_and_plus_<mptrsize>): Likewise.
19157 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19158 (vsx_div_v2di): Likewise.
19159 (vsx_udiv_v2di): Likewise.
19161 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19163 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19164 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19165 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19166 (rs6000_debug_reg_global): Delete print of LRA status.
19167 (rs6000_option_override_internal): Delete dead LRA related code.
19168 (rs6000_lra_p): Delete function.
19169 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19171 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19173 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19174 * config/riscv/rtems.h: New file.
19176 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19177 Sudakshina Das <sudi.das@arm.com>
19179 * config/aarch64/aarch64.md
19180 (define_split for and<mode>3nr_compare): Move
19181 non aarch64_logical_operand to a register.
19182 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19183 register immediate operand to a register.
19184 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19186 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19188 PR middle-end/81564
19189 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19191 2017-07-27 Richard Biener <rguenther@suse.de>
19193 PR tree-optimization/81573
19194 PR tree-optimization/81494
19195 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19196 multi defuse cycle case.
19198 2017-07-27 Richard Biener <rguenther@suse.de>
19200 PR tree-optimization/81571
19201 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19204 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19206 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19207 earlier and only if MASK_FPU is set. Adjust formatting.
19209 2017-07-27 Martin Liska <mliska@suse.cz>
19211 * opt-functions.awk: Add validation of value of Init.
19212 * optc-gen.awk: Pass new argument.
19214 2017-07-27 Martin Liska <mliska@suse.cz>
19216 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19217 Fix wrong condition.
19219 2017-07-27 Martin Liska <mliska@suse.cz>
19221 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19222 BBs and edges seen by autoFDO.
19224 2017-07-27 Richard Biener <rguenther@suse.de>
19226 PR tree-optimization/81502
19227 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19228 with incompatible but same sized type.
19229 (execute_update_addresses_taken): Likewise.
19231 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19233 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19234 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19236 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19239 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19240 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19241 Change s_operand to memory_operand.
19243 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19245 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19246 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19247 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19248 Emit instructions rather than returning an expression. Handle TFmode
19249 and KFmode by casting to TImode.
19250 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19251 (rs6000_emit_le_vsx_store): Likewise.
19252 * config/rs6000/vsx.md (VSX_TI): New iterator.
19253 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19254 (*vsx_le_undo_permute_<mode>): Likewise.
19255 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19256 emit the split sequence.
19257 (*vsx_le_perm_store_<mode>): Likewise.
19259 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19261 PR tree-optimization/81555
19262 PR tree-optimization/81556
19263 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19264 if true, force CHANGED for the recursive invocation.
19265 (reassociate_bb): Remember original length of ops array, pass
19266 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19268 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19269 attributes for noipa attribute. For naked attribute use
19270 lookup_attribute first before lookup_attribute_spec.
19271 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19273 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19274 for functions with noipa attribute.
19275 (cgraph_externally_visible_p): Return true for functions with noipa
19277 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19278 for functions with noipa attribute.
19279 * doc/extend.texi: Document noipa function attribute.
19280 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19281 also for functions with noipa attribute.
19282 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19284 2017-07-26 Andrew Pinski <apinski@cavium.com>
19286 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19287 vec_unalign_load_cost and vec_unalign_store_cost.
19289 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19291 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19292 -mvsx-small-integer option.
19293 (ISA_3_0_MASKS_IEEE): Likewise.
19294 (OTHER_VSX_VECTOR_MASKS): Likewise.
19295 (POWERPC_MASKS): Likewise.
19296 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19297 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19298 code, only testing for DImode being allowed in non-VSX floating
19300 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19301 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19303 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19304 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19305 TARGET_P8_VECTOR test.
19306 (rs6000_secondary_reload_simple_move): Likewise.
19307 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19308 since TARGET_P9_VECTOR was already tested.
19309 (rs6000_opt_masks): Remove -mvsx-small-integer.
19310 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19311 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19313 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19314 test for TARGET_VEXTRACTUB was used, and that uses
19316 (p9 extract splitter): Likewise.
19317 (vsx_extract_<mode>_di_p9): Likewise.
19318 (vsx_extract_<mode>_store_p9): Likewise.
19319 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19320 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19321 the elimination of TARGET_VSX_SMALL_INTEGER.
19322 (vsx_extract_<mode>_p8): Likewise.
19323 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19324 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19325 (vsx_set_<mode>_p9): Likewise.
19326 (vsx_set_v4sf_p9): Likewise.
19327 (vsx_set_v4sf_p9_zero): Likewise.
19328 (vsx_insert_extract_v4sf_p9): Likewise.
19329 (vsx_insert_extract_v4sf_p9_2): Likewise.
19330 * config/rs6000/rs6000.md (sign extend splitter): Change
19331 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19332 (floatsi<mode>2_lfiwax_mem): Likewise.
19333 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19334 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19335 since a test for TARGET_P9_VECTOR was used.
19336 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19337 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19338 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19339 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19340 TARGET_P8_VECTOR test.
19341 (fix_trunc<mode>si2_stfiwx): Likewise.
19342 (fix_trunc<mode>si2_internal): Likewise.
19343 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19344 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19346 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19347 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19348 TARGET_P8_VECTOR test.
19349 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19350 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19351 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19353 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19354 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19355 since a test for TARGET_P9_VECTOR was used.
19356 (splitter for loading small constants): Likewise.
19358 2017-07-26 Andrew Pinski <apinski@cavium.com>
19360 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19363 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19366 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19367 (fp_valid_at): Likewise.
19369 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19371 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19372 (qdf24xx_addrcost_table): Likewise.
19373 (cortexa57_tunings): Update to use generic_branch_cost.
19374 (cortexa72_tunings): Likewise.
19375 (cortexa73_tunings): Likewise.
19376 (qdf24xx_tunings): Likewise.
19378 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19380 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19381 (thunderx2t99_branch_cost): Likewise.
19382 (cortexa35_tunings): Update to use generic_branch_cost.
19383 (cortexa53_tunings): Likewise.
19384 (cortexa57_tunings): Likewise.
19385 (cortexa72_tunings): Likewise.
19386 (cortexa73_tunings): Likewise.
19387 (thunderx2t99_tunings): Likewise.
19389 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19391 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19392 (sparc_option_override): Honour MASK_FSMULD.
19393 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19394 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19395 * config/sparc/sparc.opt (mfsmuld): New option.
19396 * doc/invoke.texi (mfsmuld): Document option.
19398 2017-07-26 Marek Polacek <polacek@redhat.com>
19400 PR middle-end/70992
19401 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19403 2017-07-26 Richard Biener <rguenther@suse.de>
19405 * gimple-match-head.c (do_valueize): Return OP if valueize
19407 (get_def): New helper to get at the def stmt of a SSA name
19408 if valueize allows.
19409 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19410 do_valueize to get at the def stmt.
19411 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19413 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19415 PR middle-end/46932
19416 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19418 2017-07-26 Martin Liska <mliska@suse.cz>
19421 * function.c (expand_function_start): Make expansion of
19422 nonlocal_goto_save_area after parm_birth_insn.
19424 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19426 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19427 from all CPU target flags enable members.
19429 2017-07-26 Richard Biener <rguenther@suse.de>
19431 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19432 (decision_tree::gen): Make 'type' const.
19433 (write_predicate): Likewise.
19435 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19437 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19438 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19439 (rs6000_option_override_internal): Likewise.
19440 (rs6000_expand_vector_set): Likewise.
19441 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19442 (TARGET_UPPER_REGS_SF): Likewise.
19443 (TARGET_UPPER_REGS_DI): Likewise.
19444 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19445 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19446 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19447 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19448 (Splitters for DI constants in Altivec registers): Likewise.
19449 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19450 (vsx_set_v4sf_p9): Likewise.
19451 (vsx_set_v4sf_p9_zero): Likewise.
19452 (vsx_insert_extract_v4sf_p9): Likewise.
19453 (vsx_insert_extract_v4sf_p9_2): Likewise.
19455 2017-07-25 Carl Love <cel@us.ibm.com>
19457 * doc/extend.texi: Update the built-in documentation file for the
19458 existing built-in functions
19459 vector signed char vec_cnttz (vector signed char);
19460 vector unsigned char vec_cnttz (vector unsigned char);
19461 vector signed short vec_cnttz (vector signed short);
19462 vector unsigned short vec_cnttz (vector unsigned short);
19463 vector signed int vec_cnttz (vector signed int);
19464 vector unsigned int vec_cnttz (vector unsigned int);
19465 vector signed long long vec_cnttz (vector signed long long);
19466 vector unsigned long long vec_cnttz (vector unsigned long long);
19468 2017-07-25 Andrew Pinski <apinski@cavium.com>
19470 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19471 accesses where the use is for the first operand of a BIT_INSERT.
19473 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19476 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19477 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19479 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19481 * config/i386/gstabs.h: Delete.
19482 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19484 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19486 * config/i386/i386.c (ix86_decompose_address): Do not check for
19487 register RTX when looking at index_reg or base_reg.
19488 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19490 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19492 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19493 to update EH info here.
19495 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19497 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19499 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19501 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19503 2017-07-25 Torsten Duwe <duwe@suse.de>
19505 * common.opt: Introduce -fpatchable-function-entry
19506 command line option, and its variables function_entry_patch_area_size
19507 and function_entry_patch_area_start.
19508 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19509 including a two-value parser.
19510 * target.def (print_patchable_function_entry): New target hook.
19511 * targhooks.h (default_print_patchable_function_entry): New function.
19512 * targhooks.c (default_print_patchable_function_entry): Likewise.
19513 * toplev.c (process_options): Switch off IPA-RA if
19514 patchable function entries are being generated.
19515 * varasm.c (assemble_start_function): Look at the
19516 patchable-function-entry command line switch and current
19517 function attributes and maybe generate NOP instructions by
19518 calling the print_patchable_function_entry hook.
19519 * doc/extend.texi: Document patchable_function_entry attribute.
19520 * doc/invoke.texi: Document -fpatchable_function_entry
19521 command line option.
19522 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19524 * doc/tm.texi: Re-generate.
19526 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19529 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19530 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19532 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19534 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19536 2017-07-25 Richard Biener <rguenther@suse.de>
19538 PR tree-optimization/81455
19539 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19540 not walk in cycles when looking for guards.
19542 2017-07-25 Richard Biener <rguenther@suse.de>
19544 PR tree-optimization/81529
19545 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19546 when optimizing backedge uses.
19548 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19550 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19552 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19553 to dl_section_ref. On AIX, append an expression to subtract
19554 the size of the section length to dl_section_ref.
19556 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19558 * configure.ac: If any of the config.* scripts fail, exit 1.
19559 * configure: Regenerate.
19561 2017-07-25 Richard Biener <rguenther@suse.de>
19563 PR middle-end/81546
19564 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19565 of immediate uses, be more verbose on errors.
19567 2017-07-25 Richard Biener <rguenther@suse.de>
19569 PR tree-optimization/81510
19570 * tree-vect-loop.c (vect_is_simple_reduction): When the
19571 reduction stmt is not inside the loop bail out.
19573 2017-07-25 Richard Biener <rguenther@suse.de>
19575 PR tree-optimization/81303
19576 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19577 profitability check against LOOP_VINFO_NITERSM1.
19579 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19581 * domwalk.c (cmp_bb_postorder): Simplify.
19582 (sort_bbs_postorder): New function. Use it...
19583 (dom_walker::walk): ...here to optimize common cases.
19585 2017-07-25 Martin Liska <mliska@suse.cz>
19588 * ipa-visibility.c (function_and_variable_visibility): Make the
19589 redirection just on target that supports aliasing.
19590 Fix GNU coding style.
19592 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
19595 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
19597 * config/arm/rtems.h: Likewise.
19598 * config/bfin/rtems.h: Likewise.
19599 * config/i386/rtemself.h: Likewise.
19600 * config/lm32/rtems.h: Likewise.
19601 * config/m32c/rtems.h: Likewise.
19602 * config/m68k/rtemself.h: Likewise.
19603 * config/microblaze/rtems.h: Likewise.
19604 * config/mips/rtems.h: Likewise.
19605 * config/moxie/rtems.h: Likewise.
19606 * config/nios2/rtems.h: Likewise.
19607 * config/powerpcspe/rtems.h: Likewise.
19608 * config/rs6000/rtems.h: Likewise.
19609 * config/rtems.h: Likewise.
19610 * config/sh/rtems.h: Likewise.
19611 * config/sh/rtemself.h: Likewise.
19612 * config/sparc/rtemself.h: Likewise.
19614 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19617 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
19618 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
19619 * tree-ssa-structalias.c (alias_get_name): Same.
19621 2017-07-25 Bin Cheng <bin.cheng@arm.com>
19624 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
19625 instructions if no du chain is found.
19627 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19629 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
19631 2017-07-25 Richard Biener <rguenther@suse.de>
19633 PR middle-end/81505
19634 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
19637 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19639 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
19640 upper-regs options.
19641 (ISA_2_7_MASKS_SERVER): Likewise.
19642 (ISA_3_0_MASKS_IEEE): Likewise.
19643 (OTHER_P8_VECTOR_MASKS): Likewise.
19644 (OTHER_VSX_VECTOR_MASKS): Likewise.
19645 (POWERPC_MASKS): Likewise.
19646 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
19647 duplicate list of options.
19648 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
19649 explicit -mupper-regs options.
19650 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
19651 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
19652 alias for -mupper-regs-df.
19653 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19654 (rs6000_init_hard_regno_mode_ok): Likewise.
19655 (rs6000_option_override_internal): Likewise.
19656 (rs6000_opt_masks): Likewise.
19657 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
19658 options in terms of whether -mvsx or -mpower8-vector was used.
19659 (TARGET_UPPER_REGS_DI): Likewise.
19660 (TARGET_UPPER_REGS_SF): Likewise.
19661 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
19662 -mupper-regs-* options.
19664 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19666 * passes.c (emergency_dump_function): Print some empty lines and a
19667 header before the RTL dump.
19669 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19671 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
19673 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
19676 * config/aarch64/aarch64.c (aarch64_classify_symbol):
19677 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
19679 2017-07-24 Carl Love <cel@us.ibm.com>
19681 * config/rs6000/rs6000-c.c: Add support for built-in functions
19682 vector float vec_extract_fp32_from_shorth (vector unsigned short);
19683 vector float vec_extract_fp32_from_shortl (vector unsigned short);
19684 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
19685 vec_extract_fp_from_shortl): Add defines for the two builtins.
19686 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
19687 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
19689 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
19690 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
19691 * doc/extend.texi: Update the built-in documentation file for the
19692 new built-in function.
19694 2017-07-24 Jakub Jelinek <jakub@redhat.com>
19697 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
19699 * doc/generic.texi: Likewise.
19700 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19701 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19703 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
19705 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
19706 (aarch64_mls_elt_merge<mode>): Likewise.
19708 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
19710 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19711 having __cxa_atexit.
19713 2017-07-23 Michael Collison <michael.collison@arm.com>
19715 * config/arm/arm.c (arm_option_override): Deprecate
19716 use of -mstructure-size-boundary.
19717 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
19718 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
19720 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19723 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19724 Reduce cost estimate for direct moves.
19726 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
19729 * config/i386/i386.c (ix86_option_override_internal): Disable
19730 BMI, BMI2 and TBM instructions for -m16.
19732 2017-07-21 Carl Love <cel@us.ibm.com>
19734 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19735 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19736 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19737 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19738 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19739 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19740 VMULOSW): New enum "unspec" values.
19741 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19742 altivec_vmulosw): New patterns.
19743 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19744 VMULOSW): Add definitions.
19746 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
19748 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
19749 (qdf24xx): Likewise.
19750 * config/aarch64/aarch64-options-extensions.def (rdma); New.
19751 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
19752 (AARCH64_FL_V8_1): Renumber.
19753 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
19754 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
19755 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
19756 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
19757 rdma to feature modifiers list.
19759 2017-07-21 Yury Gribov <tetra2005@gmail.com>
19761 PR middle-end/56727
19762 * ipa-visibility (function_and_variable_visibility): Convert
19763 recursive PLT call to direct call if appropriate.
19765 2017-07-21 Andrew Pinski <apinski@cavium.com>
19767 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
19768 operand 1 to see if the types precision matches.
19769 * fold-const.c (operand_equal_p): Likewise.
19771 2017-07-21 Richard Biener <rguenther@suse.de>
19773 PR tree-optimization/81303
19774 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
19775 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
19776 (vect_peeling_hash_get_lowest_cost): Adjust.
19777 (vect_enhance_data_refs_alignment): Likewise. Use
19778 vect_get_peeling_costs_all_drs to compute the penalty for no
19779 peeling to match up costs.
19781 2017-07-21 Richard Biener <rguenther@suse.de>
19783 PR tree-optimization/81500
19784 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
19785 we didn't identify a reduction path.
19787 2017-07-21 Tom de Vries <tom@codesourcery.com>
19788 Cesar Philippidis <cesar@codesourcery.com>
19790 PR gcov-profile/81442
19791 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
19794 2017-07-21 Tom de Vries <tom@codesourcery.com>
19797 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
19799 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
19800 nvptx_override_options_after_change.
19802 2017-07-21 Ulrich Drepper <drepper@redhat.com>
19804 * dwarf2out.c (output_file_names): Avoid double testing for
19805 dwarf_version >= 5.
19807 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
19809 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
19811 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19813 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
19815 (try_crossjump_to_edge): Do not punt on partitioned functions.
19817 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19819 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
19820 Put all BBs reachable only via paths crossing cold region to cold
19822 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
19824 2016-07-21 Richard Biener <rguenther@suse.de>
19826 PR tree-optimization/81303
19827 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
19828 into account prologue and epilogue iterations when raising
19829 min_profitable_iters to sth at least covering one vector iteration.
19831 2017-07-21 Tamar Christina <tamar.christina@arm.com>
19833 * config/arm/arm.c (arm_test_cpu_arch_dat):
19836 2017-07-20 Nathan Sidwell <nathan@acm.org>
19838 Remove TYPE_METHODS.
19839 * tree.h (TYPE_METHODS): Delete.
19840 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
19841 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
19842 (dbxout_type_methods): Scan TYPE_FIELDS.
19843 (dbxout_type): Don't check TYPE_METHODS here.
19844 * function.c (use_register_for_decl): Always ignore register for
19845 class types when not optimizing.
19846 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
19847 * tree.c (free_lang_data_in_type): Stitch out member functions and
19848 templates from TYPE_FIELDS.
19849 (build_distinct_type_copy, verify_type_variant,
19850 verify_type): Member fns are on TYPE_FIELDS.
19851 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
19852 * tree-pretty-print.c (dump_generic_node): Likewise.
19854 2017-07-20 Jakub Jelinek <jakub@redhat.com>
19857 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
19858 V2TImode and V4TImode.
19859 (ix86_expand_vector_extract): Likewise.
19860 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
19861 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
19862 (ssescalarmode): Handle V4TImode and V2TImode.
19863 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
19864 (*vec_extractv2ti, *vec_extractv4ti): New insns.
19865 (VEXTRACTI128_MODE): New mode iterator.
19866 (splitter for *vec_extractv?ti first element): New.
19867 (VEC_INIT_MODE): New mode iterator.
19868 (vec_init<mode>): Consolidate 3 expanders into one using
19869 VEC_INIT_MODE mode iterator.
19871 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19873 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
19874 non_spilled_static_chain_regno_p.
19876 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19878 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
19880 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
19882 * bb-reorder.c (connect_traces): Allow copying of blocks within
19885 2017-07-20 Richard Biener <rguenther@suse.de>
19887 * gimple.h (gimple_phi_result): Add gphi * overload.
19888 (gimple_phi_result_ptr): Likewise.
19889 (gimple_phi_arg): Likewise. Adjust index assert to only
19890 allow actual argument accesses rather than all slots available
19892 (gimple_phi_arg_def): Add gphi * overload.
19893 * tree-phinodes.c (make_phi_node): Initialize only actual
19895 (resize_phi_node): Clear memory not covered by old node,
19896 do not initialize excess argument slots.
19897 (reserve_phi_args_for_new_edge): Initialize new argument slot
19900 2017-07-20 Bin Cheng <bin.cheng@arm.com>
19902 PR tree-optimization/81388
19904 2016-07-21 Bin Cheng <bin.cheng@arm.com>
19906 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
19907 by removing computation of may_be_zero.
19909 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19910 Tom de Vries <tom@codesourcery.com>
19912 PR middle-end/81030
19913 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
19914 when gimple level profile disagrees with what RTL expander did.
19916 2017-07-20 Richard Biener <rguenther@suse.de>
19918 PR tree-optimization/61171
19919 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
19920 (vect_analyze_stmt): Add slp instance parameter.
19921 (vectorizable_reduction): Likewise.
19922 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
19923 (vect_is_simple_reduction): Deal with chains not detected
19924 as SLP reduction chain, specifically not properly associated
19925 chains containing a mix of plus/minus.
19926 (get_reduction_op): Remove.
19927 (get_initial_defs_for_reduction): Simplify, pass in whether
19928 this is a reduction chain, pass in the SLP node for the PHIs.
19929 (vect_create_epilog_for_reduction): Get the SLP instance as
19931 (vectorizable_reduction): Get the SLP instance as arg.
19932 During analysis remember the SLP node with the PHIs in the
19933 instance. Simplify getting at the vectorized reduction PHIs.
19934 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
19935 through SLP instance.
19936 (vect_slp_analyze_operations): Likewise.
19937 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
19938 (vect_transform_stmt): Likewise.
19940 2017-07-20 Tom de Vries <tom@codesourcery.com>
19942 PR tree-optimization/81489
19943 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
19944 read of phi arg location to before loop that modifies phi.
19946 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19948 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
19951 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19953 PR middle-end/81331
19954 * except.c (execute): Fix ordering issue.
19956 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19958 PR rtl-optimization/81423
19959 * combine.c (make_compound_operation_int): Don't try to optimize
19960 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
19962 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19964 PR rtl-optimization/81423
19965 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
19966 with a constant that is -1 in the truncated to mode.
19968 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19970 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
19971 (determine_unlikely_bbs): ... here.
19972 * predict.h (propagate_unlikely_bbs_forward): Declare.
19973 * cfgexpand.c (pass_expand::execute): Use it.
19974 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
19976 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
19977 propagate_unlikely_bbs_forward.
19979 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19981 PR middle-end/81331
19982 * except.c (maybe_add_nop_after_section_switch): New function.
19985 2017-07-19 Tom de Vries <tom@codesourcery.com>
19987 * gimple.h (gimple_phi_set_arg): Make assert more strict.
19989 2017-07-19 Tom de Vries <tom@codesourcery.com>
19991 * gimple.h (gimple_phi_arg): Make assert more strict.
19993 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
19995 * config.gcc (powerpc*-*-*): Add mmintrin.h.
19996 * config/rs6000/mmintrin.h: New file.
19997 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
19999 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20001 PR tree-optimization/81346
20002 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20004 2017-07-19 Tom de Vries <tom@codesourcery.com>
20006 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20008 2017-07-19 Tom de Vries <tom@codesourcery.com>
20010 * config/nvptx/nvptx-modes.def: Add V2DImode.
20011 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20012 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20013 (nvptx_output_mov_insn): Handle lack of mov.b128.
20014 (nvptx_print_operand): Handle 'H' and 'L' codes.
20015 (nvptx_vector_mode_supported): Allow V2DImode.
20016 (nvptx_preferred_simd_mode): New function.
20017 (nvptx_data_alignment): New function.
20018 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20019 nvptx_preferred_simd_mode.
20020 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20022 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20024 2017-07-19 Tom de Vries <tom@codesourcery.com>
20026 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20027 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20028 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20029 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20030 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20031 (mov<VECIM>_insn): New define_insn.
20032 (define_expand "mov<VECIM>): New define_expand.
20034 2017-07-19 Tom de Vries <tom@codesourcery.com>
20036 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20038 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20040 PR tree-optimization/81346
20041 * fold-const.h (fold_div_compare, range_check_type): Declare.
20042 * fold-const.c (range_check_type): New function.
20043 (build_range_check): Use range_check_type.
20044 (fold_div_compare): No longer static, rewritten into
20045 a match.pd helper function.
20046 (fold_comparison): Don't call fold_div_compare here.
20047 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20048 as helper function.
20050 2017-07-19 Nathan Sidwell <nathan@acm.org>
20052 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20053 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20054 * tree.c (find_decls_types_r, verify_type): Use
20055 TYPE_{MIN,MAX}_VALUE_RAW.
20056 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20057 (hash_tree): Likewise.
20058 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20060 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20063 2017-07-18 Tom de Vries <tom@codesourcery.com>
20065 PR middle-end/81464
20066 * omp-expand.c (expand_omp_for_static_chunk): Handle
20067 equal-argument loop exit phi.
20069 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20072 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20073 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20074 operand 2 predicate.
20075 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20076 operand 2 predicate.
20077 (ror,rol -> rorx splitters): Use const_int_operand as
20078 operand 2 predicate.
20080 2017-06-18 Richard Biener <rguenther@suse.de>
20082 PR tree-optimization/81410
20083 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20084 the gap in the ! slp_perm SLP case after each group.
20086 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20088 PR middle-end/81463
20089 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20092 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20094 PR middle-end/81462
20095 * predict.c (set_even_probabilities): Cleanup; do not affect
20096 probabilities that are already known.
20097 (combine_predictions_for_bb): Call even when count is set.
20099 2017-07-18 Nathan Sidwell <nathan@acm.org>
20101 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20104 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20107 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20108 optimization for loop niter analysis.
20110 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20113 * config/avr/avr.c (avr_optimize_casesi): Don't use
20114 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20116 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20118 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20119 body_cost_vec from _vect_peel_extended_info.
20120 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20121 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20124 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20126 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20128 2017-07-18 Richard Biener <rguenther@suse.de>
20130 PR tree-optimization/80620
20131 PR tree-optimization/81403
20132 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20133 info when re-using a VN table entry.
20135 2017-07-18 Richard Biener <rguenther@suse.de>
20137 PR tree-optimization/81418
20138 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20139 vectype_in. Verify that with lane-reducing reduction operations
20140 we have a single def-use cycle.
20142 2017-07-17 Carl Love <cel@us.ibm.com>
20144 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20146 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20147 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20148 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20149 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20150 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20151 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20152 VMULOSW): New enum "unspec" values.
20153 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20154 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20155 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20156 altivec_vmulosw): New patterns.
20157 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20158 VMULOSW): Add definitions.
20160 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20162 * config/alpha/alpha.c: Include predict.h.
20164 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20166 * tree-vrp.c (compare_assert_loc): Fix comparison function
20167 to return predictable results.
20169 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20171 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20173 (subdi3): Likewise.
20174 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20175 * doc/invoke.texi (mexpand-adddi): Update text.
20177 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20179 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20180 that also clobbers the CC register. The old expand code is moved
20182 (*arc_clzsi2): ... here.
20183 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20184 the CC register. The old expand code is moved to ...
20185 (arc_ctzsi2): ... here.
20187 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20189 * config/arc/arc.opt (mindexed-loads): Use initial value
20190 TARGET_INDEXED_LOADS_DEFAULT.
20191 (mauto-modify-reg): Use initial value
20192 TARGET_AUTO_MODIFY_REG_DEFAULT.
20193 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20194 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20195 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20196 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20198 2017-07-17 Martin Liska <mliska@suse.cz>
20201 * opts.c (finish_options): Do not allow -fgnu-tm
20202 w/ -fsanitize={kernel-,}address. Say sorry.
20204 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20207 * tree-loop-distribution.c (classify_partition): Only assert on
20208 numer of iterations.
20209 (merge_dep_scc_partitions): Delete prameter. Update function call.
20210 (distribute_loop): Remove code handling loop with unknown niters.
20211 (pass_loop_distribution::execute): Skip loop with unknown niters.
20213 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20216 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20217 function sort_partitions_by_post_order.
20219 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20221 PR tree-optimization/81374
20222 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20223 the max index of basic blocks, rather than number of basic blocks.
20225 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20227 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20229 (arc_legitimate_pic_operand_p): Likewise.
20230 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20232 (arc_needs_pcl_p): Likewise.
20233 (arc_legitimate_pc_offset_p): Likewise.
20234 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20235 function is also used in constrains.md.
20236 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20237 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20238 PLUS. Only return true/false in known cases, otherwise assert.
20239 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20240 is already called in arc_legitimate_constant_p.
20241 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20243 (LEGITIMATE_PIC_OPERAND_P): Use
20244 arc_raw_symbolic_reference_mentioned_p function.
20245 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20250 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20251 Andrew Burgess <andrew.burgess@embecosm.com>
20253 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20254 (arc_return_address_register): New function.
20255 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20256 (arc_handle_fndecl_attribute): Add naked attribute.
20257 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20258 (TARGET_WARN_FUNC_RETURN): Likewise.
20259 (arc_allocate_stack_slots_for_args): New function.
20260 (arc_warn_func_return): Likewise.
20261 (machine_function): Change type fn_type.
20262 (arc_compute_function_type): Consider new naked function type,
20263 change function return type.
20264 (arc_must_save_register): Adapt to handle new
20265 arc_compute_function_type's return type.
20266 (arc_expand_prologue): Likewise.
20267 (arc_expand_epilogue): Likewise.
20268 (arc_return_address_regs): Delete.
20269 (arc_return_address_register): New function.
20270 (arc_epilogue_uses): Use above function.
20271 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20272 (arc_function_type): Change encoding, add naked type.
20273 (ARC_INTERRUPT_P): Change to handle the new encoding.
20274 (ARC_FAST_INTERRUPT_P): Likewise.
20275 (ARC_NORMAL_P): Define.
20276 (ARC_NAKED_P): Likewise.
20277 (arc_compute_function_type): Delete prototype.
20278 * config/arc/arc.md (in_ret_delay_slot): Use
20279 arc_return_address_register function.
20280 (simple_return): Likewise.
20281 (p_return_i): Likewise.
20283 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20285 PR tree-optimization/81428
20286 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20287 can't be built for those types.
20289 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20291 Remove stuff dead since r239246.
20293 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20294 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20295 (avr_inform_devices): Remove dead stuff.
20297 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20299 * config/arm/arm_neon.h: Fix softp typo.
20301 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20303 PR tree-optimization/81365
20304 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20305 aggregate moves onto bb predecessor edges, make sure there are no
20306 loads that could alias the lhs in between the start of bb and the
20309 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20312 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20313 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20314 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20316 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20318 PR tree-optimization/81396
20319 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20320 (init_symbolic_number): Initialize it to 1.
20321 (perform_symbolic_merge): Add n_ops from both operands into the new
20323 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20324 without base_addr as useless if they need more than one operation.
20325 (bswap_replace): Handle !bswap case for NULL base_addr.
20327 2017-07-17 Tom de Vries <tom@codesourcery.com>
20330 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20333 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20335 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20336 conditional builtin define __FIX_LEON3FT_B2BST.
20338 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20340 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20341 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20344 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20346 PR rtl-optimization/81424
20347 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20348 to remove potential trapping from operands if -fnon-call-exceptions.
20350 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20352 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20353 profile_proability for scalling.
20354 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20356 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20358 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20360 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20362 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20363 fixpoint arithmetics.
20365 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20367 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20368 fixpoint arithmetics.
20370 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20372 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20373 fixpoint arithmetics.
20375 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20377 * profile-count.h (profile_probability::from_reg_br_prob_note,
20378 profile_probability::to_reg_br_prob_note): New functions.
20379 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20380 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20381 * predict.c (probability_reliable_p): Update.
20382 (edge_probability_reliable_p): Update.
20383 (br_prob_note_reliable_p): Update.
20384 (invert_br_probabilities): Update.
20385 (add_reg_br_prob_note): New function.
20386 (combine_predictions_for_insn): Update.
20387 * asan.c (asan_clear_shadow): Update.
20388 * cfgbuild.c (compute_outgoing_frequencies): Update.
20389 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20390 (update_br_prob_note): Update.
20391 (rtl_verify_edges): Update.
20392 (purge_dead_edges): Update.
20393 (fixup_reorder_chain): Update.
20394 * emit-rtl.c (try_split): Update.
20395 * ifcvt.c (cond_exec_process_insns): Update.
20396 (cond_exec_process_if_block): Update.
20397 (dead_or_predicable): Update.
20398 * internal-fn.c (expand_addsub_overflow): Update.
20399 (expand_neg_overflow): Update.
20400 (expand_mul_overflow): Update.
20401 * loop-doloop.c (doloop_modify): Update.
20402 * loop-unroll.c (compare_and_jump_seq): Update.
20403 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20404 * predict.h: Update.
20405 * reorg.c (mostly_true_jump): Update.
20407 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20408 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20409 * config/arc/arc.c: (emit_unlikely_jump): Update.
20410 * config/arm/arm.c: (emit_unlikely_jump): Update.
20411 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20412 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20413 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20414 (ix86_print_operand): Update.
20415 (ix86_split_fp_branch): Update.
20416 (predict_jump): Update.
20417 * config/ia64/ia64.c (ia64_print_operand): Update.
20418 * config/mmix/mmix.c (mmix_print_operand): Update.
20419 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20420 (rs6000_expand_split_stack_prologue): Update.
20421 * config/rs6000/rs6000.c: Update.
20422 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20423 (s390_expand_vec_movstr): Update.
20424 (s390_expand_cs_tdsi): Update.
20425 (s390_expand_split_stack_prologue): Update.
20426 * config/sh/sh.c (sh_print_operand): Update.
20427 (expand_cbranchsi4): Update.
20428 (expand_cbranchdi4): Update.
20429 * config/sparc/sparc.c (output_v9branch): Update.
20430 * config/spu/spu.c (get_branch_target): Update.
20431 (ea_load_store_inline): Update.
20432 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20433 * config/tilepro/tilepro.c: Update.
20435 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20437 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20438 (gimplify_save_expr): Likewise.
20440 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20442 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20444 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20446 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20448 * timevar.def (TV_IPA_FNSUMMARY): Define.
20450 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20452 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20453 to back store errata sensitive sequence from being generated.
20454 (sqrtdf2_fix): Likewise.
20456 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20458 * tree-ssa-threadupdate.c (compute_path_counts,
20459 update_joiner_offpath_counts): Use profile_probability.
20461 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20464 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20466 * config/arm/arm-c.c (arm_cpu_builtins): Define
20467 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20469 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20471 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20472 array entries to represent __ieee128 versions of the
20473 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20474 scalar_extract_sig, and scalar_insert_exp built-in functions.
20475 (altivec_resolve_overloaded_builtin): Add special case handling
20476 for the __builtin_scalar_insert_exp function, as represented by
20477 the P9V_BUILTIN_VEC_VSIEDP constant.
20478 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20479 exponent support for __ieee128 argument.
20480 (VSESQP): Add scalar extract signature support for __ieee128
20482 (VSTDCNQP): Add scalar test negative support for __ieee128
20484 (VSIEQP): Add scalar insert exponent support for __int128 argument
20485 with __ieee128 result.
20486 (VSIEQPF): Add scalar insert exponent support for __ieee128
20487 argument with __ieee128 result.
20488 (VSTDCQP): Add scalar test data class support for __ieee128
20490 (VSTDCNQP): Add overload support for scalar test negative with
20491 __ieee128 argument.
20492 (VSTDCQP): Add overload support for scalar test data class
20493 __ieee128 argument.
20494 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20495 UNSPEC_VSX_SXSIGDP.
20496 (UNSPEC_VSX_SIEXPQP): New constant.
20497 (xsxexpqp): New insn for VSX scalar extract exponent quad
20499 (xsxsigqp): New insn for VSX scalar extract significand quad
20501 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20502 precision with floating point argument.
20503 (xststdcqp): New expand for VSX scalar test data class quad
20505 (xststdcnegqp): New expand for VSX scalar test negative quad
20507 (xststdcqp): New insn to match expansions for VSX scalar test data
20508 class quad precision and VSX scalar test negative quad precision.
20509 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20510 special case operand checking to enforce that second operand of
20511 VSX scalar test data class with quad precision argument is a 7-bit
20513 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20514 prototypes and descriptions of __ieee128 versions of
20515 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20516 scalar_test_data_class, and scalar_test_neg built-in functions.
20518 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20520 PR tree-optimization/81162
20521 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20522 replace a negate with an add.
20524 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20526 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20528 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20530 * config/arm/arm-c.c (arm_cpu_builtins): Define
20531 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20533 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20535 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20536 (armv8-r): Set ARM Cortex-R52 as default CPU.
20537 * config/arm/arm-tables.opt: Regenerate.
20538 * config/arm/arm-tune.md: Regenerate.
20539 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20541 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20542 extension for -mcpu=cortex-r52.
20544 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20546 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20547 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20548 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20549 (fp-armv8): Define it as FP_ARMv8 only.
20550 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20551 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20553 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20554 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20555 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20556 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20557 than TARGET_FPU_ARMV8.
20558 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20559 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20560 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20562 * config/arm/neon.md (neon_vrint): Likewise.
20563 (neon_vcvt): Likewise.
20564 (neon_<fmaxmin_op><mode>): Likewise.
20565 (<fmaxmin><mode>3): Likewise.
20566 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20567 * config/arm/predicates.md (arm_cond_move_operator): Check against
20568 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20570 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20572 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20573 to top of function.
20575 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20577 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20578 loop in comment with memset.
20580 2017-07-14 Martin Liska <mliska@suse.cz>
20582 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20583 * dwarf2out.c (is_java): Remove the function.
20584 (output_pubname): Remove usage of the function.
20585 (lower_bound_default): Remove usage of DW_LANG_Java.
20586 (gen_compile_unit_die): Likewise.
20587 * gcc.c: Remove compiler defaults for .java and .zip files.
20588 * gimple-expr.c (remove_suffix): Change as there's no longer
20589 extension than 4-letter one.
20590 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20591 (gimplify_save_expr): Likewise.
20592 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20593 as it's possible even for other languages than Java.
20594 * langhooks.h (struct lang_hooks): Remove Java from a comment.
20595 * lto-opts.c (lto_write_options): Remove reference to Java.
20596 * opts.c (strip_off_ending): Update file extension handling.
20597 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
20598 * tree-eh.c (lower_resx): Likewise.
20599 * tree.c (free_lang_data_in_type): Remove dead code.
20600 (find_decls_types_r): Likewise.
20601 (build_common_builtin_nodes): Remove Java from a comment.
20602 (verify_type): Remove dead code.
20603 * varasm.c (assemble_external): Remove Java from a comment.
20605 2017-07-14 Martin Liska <mliska@suse.cz>
20607 * opts.c (finish_options): Add quotes.
20608 (common_handle_option): Likewise.
20610 2017-07-14 Martin Liska <mliska@suse.cz>
20612 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
20613 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
20614 Remove N_SO_PASCAL.
20615 * dwarf2out.c (lower_bound_default): Do not handle
20617 (gen_compile_unit_die): Likewise.
20618 * gcc.c: Remove default extension binding for GNU Pascal.
20619 * stmt.c: Remove Pascal language from a comment.
20620 * xcoffout.c: Likewise.
20622 2017-07-13 David Malcolm <dmalcolm@redhat.com>
20625 * diagnostic-show-locus.c (fixit_cmp): New function.
20626 (layout::layout): Sort m_fixit_hints.
20627 (column_range::column_range): Assert that the values are valid.
20628 (struct char_span): New struct.
20629 (correction::overwrite): New method.
20630 (struct source_line): New struct.
20631 (line_corrections::add_hint): Add assertions. Reimplement memcpy
20632 calls in terms of classes source_line and char_span, and
20633 correction::overwrite.
20634 (selftest::test_overlapped_fixit_printing_2): New function.
20635 (selftest::diagnostic_show_locus_c_tests): Call it.
20637 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
20639 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
20640 early if there is no lhs.
20642 2017-07-13 Martin Liska <mliska@suse.cz>
20644 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
20645 (gen_reference_type_die): Likewise.
20646 * stor-layout.c: Remove Pascal-related comment.
20648 2017-07-13 Martin Liska <mliska@suse.cz>
20650 * opts.c (finish_options): Add quotes to error messages.
20651 (parse_sanitizer_options): Likewise.
20653 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
20655 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
20657 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
20659 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
20661 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
20663 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
20665 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
20667 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
20670 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
20671 provides the hardware capability bits, define the macro
20672 __BUILTIN_CPU_SUPPORTS__.
20673 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
20674 if GLIBC does not provide the hardware capability bits. Add a
20675 gcc_unreachable call if the built-in cpu function is neither
20676 __builtin_cpu_is nor __builtin_cpu_supports.
20677 (rs6000_get_function_versions_dispatcher): Change the warning
20678 that an old GLIBC is used which does not export the capability
20679 bits to be an error.
20680 * doc/extend.texi (target_clones attribute): Document the
20681 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
20682 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
20683 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
20684 the macros defined by GCC if the newer GLIBC is available.
20686 2017-07-12 Jeff Law <law@redhat.com>
20688 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
20689 remaining includes slightly.
20690 * config/riscv/riscv-builtins.c: Include profile-count.h.
20692 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20695 * config/avr/avr.c (avr_set_current_function): In diagnostic
20696 messages: Quote keywords and (parts of) identifiers.
20697 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
20700 2017-07-12 Carl Love <cel@us.ibm.com>
20702 * config/rs6000/rs6000-c.c: Add support for built-in functions
20703 vector bool char vec_revb (vector bool char);
20704 vector bool short vec_revb (vector short char);
20705 vector bool int vec_revb (vector bool int);
20706 vector bool long long vec_revb (vector bool long long);
20707 * doc/extend.texi: Update the built-in documentation file for the
20708 new built-in functions.
20710 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20712 * config/s390/s390.md: Remove movcc splitter.
20714 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20716 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
20717 load/store on condition.
20719 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20722 * config/avr/avr.c (avr_encode_section_info)
20723 [progmem && !TREE_READONLY]: Error if progmem object needs
20726 2017-07-11 Michael Collison <michael.collison@arm.com>
20728 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
20731 2017-07-11 Carl Love <cel@us.ibm.com>
20733 * config/rs6000/rs6000-c.c: Add support for builtins
20734 vector unsigned int vec_parity_lsbb (vector signed int);
20735 vector unsigned int vec_parity_lsbb (vector unsigned int);
20736 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
20737 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
20738 vector unsigned long long vec_parity_lsbb (vector signed long long);
20739 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
20740 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
20741 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
20742 * doc/extend.texi: Update the built-in documentation file for the
20743 new built-in functions.
20745 2017-07-11 David Malcolm <dmalcolm@redhat.com>
20747 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
20748 (layout::m_primary_loc): New field.
20749 (layout::layout): Initialize new field. Move location filtering
20750 logic from here to...
20751 (layout::maybe_add_location_range): ...this new method. Add
20752 support for filtering to just the lines already specified by other
20754 (layout::will_show_line_p): New method.
20755 (gcc_rich_location::add_location_if_nearby): New method.
20756 (selftest::test_add_location_if_nearby): New test function.
20757 (selftest::diagnostic_show_locus_c_tests): Call it.
20758 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
20761 2017-07-11 Tom de Vries <tom@codesourcery.com>
20763 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
20764 (bb_first_real_insn): New function.
20765 (nvptx_single): Add extra initialization of broadcasted condition
20768 2017-07-11 Nathan Sidwell <nathan@acm.org>
20770 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
20772 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
20774 * doc/extend.texi (AVR Function Attributes): Remove weblink to
20775 Binutils doc as TEXI will mess them up.
20776 * doc/invoke.texi (AVR Options): Same here.
20778 2017-07-11 Daniel Cederman <cederman@gaisler.com>
20780 * config/sparc/sparc.opt (mfix-ut700): New option.
20781 (mfix-gr712rc): Likewise.
20782 (sparc_fix_b2bst): New variable.
20783 * doc/invoke.texi (SPARC options): Document them.
20784 (ARM options): Fix warnings.
20785 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
20786 instructions to prevent sequences that can trigger the store-store
20787 errata for certain LEON3FT processors.
20788 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
20789 (sparc_option_override): Set sparc_fix_b2bst appropriately.
20790 * config/sparc/sparc.md (fix_b2bst): New attribute.
20791 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
20793 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
20796 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
20798 (*rsqrtsf2_sse): Ditto.
20800 (div<mode>3): Macroize insn from divdf3 and divsf3
20801 using MODEF mode iterator.
20803 2017-07-10 Martin Sebor <msebor@redhat.com>
20805 PR tree-optimization/80397
20806 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
20807 instead of testing for equality to INTEGER_TYPE.
20809 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
20811 * config.gcc: Remove uclibc from arc target spec.
20813 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
20815 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
20817 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20820 * lto-wrapper.c (remove_option): New function.
20821 (merge_and_complain): Merge PIC/PIE options more realistically.
20823 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20825 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
20829 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
20830 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
20831 * config.in: Regenerate.
20832 * configure: Regenerate.
20833 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
20834 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
20835 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
20836 (TARGET_GASISR_PROLOGUES): ...target mask.
20837 * common/config/avr/avr-common.c
20838 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
20839 Set -mgas-isr-prologues.
20840 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
20841 INSERT_PASS_BEFORE for it.
20842 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
20843 * config/avr/avr.c (avr_option_override)
20844 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
20845 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
20846 (avr_attribute_table) <no_gccisr>: Add new function attribute.
20847 (avr_set_current_function) <is_no_gccisr>: Init machine field.
20848 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
20850 (make_avr_pass_pre_proep): New function.
20851 (emit_push_sfr) <treg>: Add argument to function and use it
20852 instead of TMP_REG.
20853 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
20854 and set machine->gasisr.yes.
20855 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
20856 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
20857 __gcc_isr.n_pushed to .L__stack_usage.
20858 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
20859 (avr_asm_final_postscan_insn): ...this new static function.
20860 * config/avr/avr.h (machine_function)
20861 <is_no_gccisr, use_L__stack_usage>: New fields.
20862 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
20863 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
20864 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
20865 (gasisr, *gasisr): New expander and insn.
20866 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20867 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
20868 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
20870 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
20872 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
20875 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20877 Move jump-tables out of .text again.
20880 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
20881 (ASM_OUTPUT_ADDR_VEC): New function.
20882 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
20883 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
20884 INSN_ADDRESSes as asm comment.
20885 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
20886 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
20887 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
20888 * config/avr/avr.md (*tablejump): Adjust comment.
20889 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
20890 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
20892 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
20893 (avr_output_addr_vec): New proto.
20894 (avr_log_t) <insn_addresses>: New field.
20896 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
20899 * config/i386/i386.c (ix86_function_arg_advance): Set
20900 outgoing_args_on_stack to true if there are outgoing arguments
20902 (ix86_function_arg): Likewise.
20903 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
20904 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
20905 * config/i386/i386.h (machine_function): Add
20906 outgoing_args_on_stack.
20908 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
20910 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20911 supporting pthreds.
20912 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
20914 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
20916 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
20917 (REAL_H): Remove $(MACHMODE_H).
20918 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
20920 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
20921 $(MACHMODE_H) and double-int.h.
20922 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
20924 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
20927 2017-07-07 Andrew Pinski <apinski@cavium.com>
20929 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
20930 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
20932 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20934 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
20935 Add warning if GCC was not configured to link against a GLIBC that
20936 exports the hardware capability bits.
20937 (make_resolver_func): Make resolver function private and not a
20938 COMDAT function. Create the name with clone_function_name instead
20939 of make_unique_name.
20942 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
20943 correct operand in doing the split.
20945 2017-07-07 Carl Love <cel@us.ibm.com>
20947 * config/rs6000/rs6000-c: Add support for built-in function
20948 vector unsigned short vec_pack_to_short_fp32 (vector float,
20950 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
20951 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
20952 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
20953 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
20954 (convert_4f32_8i16): Add define_expand.
20955 * doc/extend.texi: Update the built-in documentation file for the
20956 new built-in function.
20958 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20960 * config/sparc/m8.md: New file.
20961 * config/sparc/sparc.md: Include m8.md.
20963 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20965 * config/sparc/sparc.opt: New option -mvis4b.
20966 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
20967 (sparc_option_override): Handle VIS4B.
20968 (enum sparc_builtins): Define
20969 SPARC_BUILTIN_DICTUNPACK{8,16,32},
20970 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
20971 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
20972 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
20973 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
20974 (check_constant_argument): New function.
20975 (sparc_vis_init_builtins): Define builtins
20976 __builtin_vis_dictunpack{8,16,32},
20977 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
20978 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
20979 __builtin_vis_fpcmpde{8,16,32}shl and
20980 __builtin_vis_fpcmpur{8,16,32}shl.
20981 (sparc_expand_builtin): Check that the constant operands to
20982 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
20983 constant and in range.
20984 * config/sparc/sparc-c.c (sparc_target_macros): Handle
20986 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
20987 (SPARC_IMM5_P): Likewise.
20988 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
20989 (enabled): Handle vis4b.
20990 (UNSPEC_DICTUNPACK): New unspec.
20991 (UNSPEC_FPCMPSHL): Likewise.
20992 (UNSPEC_FPUCMPSHL): Likewise.
20993 (UNSPEC_FPCMPDESHL): Likewise.
20994 (UNSPEC_FPCMPURSHL): Likewise.
20995 (cpu_feature): New CPU feature `vis4b'.
20996 (dictunpack{8,16,32}): New insns.
20997 (FPCSMODE): New mode iterator.
20998 (fpcscond): New code iterator.
20999 (fpcsucond): Likewise.
21000 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21001 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21002 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21003 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21004 * config/sparc/constraints.md: Define constraints `q' for unsigned
21005 2-bit integer constants and `t' for unsigned 5-bit integer
21007 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21009 (imm5_operand_dictunpack16): Likewise.
21010 (imm5_operand_dictunpack32): Likewise.
21011 (imm2_operand): Likewise.
21012 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21013 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21014 ditunpack* and fpcmp*shl builtins.
21016 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21018 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21019 * config.in: Add HAVE_AS_SPARC6 define.
21020 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21022 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21024 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21025 (CPP_CPU_SPEC): Handle m8.
21026 (ASM_CPU_SPEC): Likewise.
21027 * config/sparc/sparc-opts.h (enum processor_type): Add
21029 * config/sparc/sparc.c (m8_costs): New struct.
21030 (sparc_option_override): Handle TARGET_CPU_m8.
21031 (sparc32_initialize_trampoline): Likewise.
21032 (sparc64_initialize_trampoline): Likewise.
21033 (sparc_issue_rate): Likewise.
21034 (sparc_register_move_cost): Likewise.
21035 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21036 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21037 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21038 (CPP_CPU_SPEC): Handle M8.
21039 (ASM_CPU_SPEC): Likewise.
21040 (AS_M8_FLAG): Define.
21041 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21042 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21043 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21045 * configure: Regenerate.
21046 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21049 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21051 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21053 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21054 ("*movdi_insn_sp32"): Do not set v3pipe.
21055 ("*movsi_insn"): Likewise.
21056 ("*movdi_insn_sp64"): Likewise.
21057 ("*movsf_insn"): Likewise.
21058 ("*movdf_insn_sp32"): Likewise.
21059 ("*movdf_insn_sp64"): Likewise.
21060 ("*zero_extendsidi2_insn_sp64"): Likewise.
21061 ("*sign_extendsidi2_insn"): Likewise.
21062 ("*mov<VM32:mode>_insn"): Likewise.
21063 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21064 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21065 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21066 ("<vlop:code><VL:mode>3"): Likewise.
21067 ("*not_<vlop:code><VL:mode>3"): Likewise.
21068 ("*nand<VL:mode>_vis"): Likewise.
21069 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21070 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21071 ("one_cmpl<VL:mode>2"): Likewise.
21072 ("faligndata<VM64:mode>_vis"): Likewise.
21073 ("alignaddrsi_vis"): Likewise.
21074 ("alignaddrdi_vis"): Likweise.
21075 ("alignaddrlsi_vis"): Likewise.
21076 ("alignaddrldi_vis"): Likewise.
21077 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21078 ("bmaskdi_vis"): Likewise.
21079 ("bmasksi_vis"): Likewise.
21080 ("bshuffle<VM64:mode>_vis"): Likewise.
21081 ("cmask8<P:mode>_vis"): Likewise.
21082 ("cmask16<P:mode>_vis"): Likewise.
21083 ("cmask32<P:mode>_vis"): Likewise.
21084 ("pdistn<P:mode>_vis"): Likewise.
21085 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21087 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21089 * config/sparc/sparc.md ("subtype"): New insn attribute.
21090 ("*wrgsr_sp64"): Set insn subtype.
21091 ("*rdgsr_sp64"): Likewise.
21092 ("alignaddrsi_vis"): Likewise.
21093 ("alignaddrdi_vis"): Likewise.
21094 ("alignaddrlsi_vis"): Likewise.
21095 ("alignaddrldi_vis"): Likewise.
21096 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21097 ("fexpand_vis"): Likewise.
21098 ("fpmerge_vis"): Likewise.
21099 ("faligndata<VM64:mode>_vis"): Likewise.
21100 ("bshuffle<VM64:mode>_vis"): Likewise.
21101 ("cmask8<P:mode>_vis"): Likewise.
21102 ("cmask16<P:mode>_vis"): Likewise.
21103 ("cmask32<P:mode>_vis"): Likewise.
21104 ("fchksm16_vis"): Likewise.
21105 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21106 ("fmean16_vis"): Likewise.
21107 ("fp<plusminus_insn>64_vis"): Likewise.
21108 ("<plusminus_insn>v8qi3"): Likewise.
21109 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21110 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21111 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21112 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21113 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21114 ("*movqi_insn"): Likewise.
21115 ("*movhi_insn"): Likewise.
21116 ("*movsi_insn"): Likewise.
21117 ("movsi_pic_gotdata_op"): Likewise.
21118 ("*movdi_insn_sp32"): Likewise.
21119 ("*movdi_insn_sp64"): Likewise.
21120 ("movdi_pic_gotdata_op"): Likewise.
21121 ("*movsf_insn"): Likewise.
21122 ("*movdf_insn_sp32"): Likewise.
21123 ("*movdf_insn_sp64"): Likewise.
21124 ("*zero_extendhisi2_insn"): Likewise.
21125 ("*zero_extendqihi2_insn"): Likewise.
21126 ("*zero_extendqisi2_insn"): Likewise.
21127 ("*zero_extendqidi2_insn"): Likewise.
21128 ("*zero_extendhidi2_insn"): Likewise.
21129 ("*zero_extendsidi2_insn_sp64"): Likewise.
21130 ("ldfsr"): Likewise.
21131 ("prefetch_64"): Likewise.
21132 ("prefetch_32"): Likewise.
21133 ("tie_ld32"): Likewise.
21134 ("tie_ld64"): Likewise.
21135 ("*tldo_ldub_sp32"): Likewise.
21136 ("*tldo_ldub1_sp32"): Likewise.
21137 ("*tldo_ldub2_sp32"): Likewise.
21138 ("*tldo_ldub_sp64"): Likewise.
21139 ("*tldo_ldub1_sp64"): Likewise.
21140 ("*tldo_ldub2_sp64"): Likewise.
21141 ("*tldo_ldub3_sp64"): Likewise.
21142 ("*tldo_lduh_sp32"): Likewise.
21143 ("*tldo_lduh1_sp32"): Likewise.
21144 ("*tldo_lduh_sp64"): Likewise.
21145 ("*tldo_lduh1_sp64"): Likewise.
21146 ("*tldo_lduh2_sp64"): Likewise.
21147 ("*tldo_lduw_sp32"): Likewise.
21148 ("*tldo_lduw_sp64"): Likewise.
21149 ("*tldo_lduw1_sp64"): Likewise.
21150 ("*tldo_ldx_sp64"): Likewise.
21151 ("*mov<VM32:mode>_insn"): Likewise.
21152 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21153 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21155 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21157 * config/sparc/sparc.md ("type"): New insn type viscmp.
21158 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21160 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21161 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21162 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21163 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21165 ("n7_vis_logical_11cycle"): Likewise.
21166 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21167 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21168 * config/sparc/niagara.md ("niag_vis"): Likewise.
21169 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21170 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21172 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21174 * config/sparc/sparc.md: New instruction type `bmask'.
21175 (bmaskdi_vis): Use the `bmask' type.
21176 (bmasksi_vis): Likewise.
21177 * config/sparc/ultra3.md (us3_array): Likewise.
21178 * config/sparc/niagara7.md (n7_array): Likewise.
21179 * config/sparc/niagara4.md (n4_array): Likewise.
21180 * config/sparc/niagara2.md (niag2_vis): Likewise.
21181 (niag3_vis): Likewise.
21182 * config/sparc/niagara.md (niag_vis): Likewise.
21184 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21186 * ipa-comdats.c: Remove optimize check from gate.
21187 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21188 for functions not optimized.
21189 (ipa_fn_summary_read): Skip optimize check.
21190 (ipa_fn_summary_write): Likewise.
21191 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21193 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21195 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21196 (check_callers): Give up on caller which is not optimized.
21197 (inline_small_functions): Likewise.
21198 (ipa_inline): Do not give up when not optimizing.
21199 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21200 away unoptimizes cdtors.
21201 (whole_program_function_and_variable_visibility): Do
21202 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21203 * ipa.c (process_references): Do not check optimize.
21204 (symbol_table::remove_unreachable_nodes): Update optimize check.
21205 (set_writeonly_bit): Update optimize check.
21206 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21207 (pass_ipa_single_use::gate): Remove.
21209 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21211 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21212 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21213 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21214 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21215 permute_load, permute_store, adjust_extract, adjust_splat,
21216 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21217 replace_swap_with_copy, dump_swap_insn_table,
21218 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21219 recombine_lvx_pattern, recombine_stvx_pattern,
21220 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21221 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21222 to file rs6000-p8swap.c.
21223 * config/rs6000/rs6000-p8swap.c: New file.
21224 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21225 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21226 and rs6000*-*-* targets.
21228 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21230 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21232 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21234 * lto-wrapper.c (merge_and_complain): Do not merge
21235 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21236 fsigned_zeros, ftrapping_math, fwrapv.
21237 (append_compiler_options): Do not track these options.
21238 (append_linker_options): Likewie
21240 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21242 * cgraphunit.c (cgraph_node::finalize_function): When
21243 !flag_toplevel_reorde set no_reorder flag.
21244 (varpool_node::finalize_decl): Likewise.
21245 (symbol_table::compile): Drop no toplevel reorder path.
21247 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21249 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21250 edges; zero probability is not better than uninitialized.
21252 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21254 * asan.h (asan_sanitize_allocas_p): Declare.
21255 * asan.c (asan_sanitize_allocas_p): New function.
21256 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21257 (handle_builtin_alloca): Likewise.
21258 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21259 if !asan_sanitize_allocas_p.
21260 * params.def (asan-instrument-allocas): Add new option.
21261 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21262 * opts.c (common_handle_option): Disable allocas sanitization for
21265 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21267 * asan.c: Include gimple-fold.h.
21268 (get_last_alloca_addr): New function.
21269 (handle_builtin_stackrestore): Likewise.
21270 (handle_builtin_alloca): Likewise.
21271 (asan_emit_allocas_unpoison): Likewise.
21272 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21273 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21274 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21275 (instrument_builtin_call): Pass gimple iterator to
21276 get_mem_refs_of_builtin_call.
21277 (last_alloca_addr): New global.
21278 * asan.h (asan_emit_allocas_unpoison): Declare.
21279 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21280 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21281 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21282 if function calls alloca.
21283 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21284 * gimple-fold.h (replace_call_with_value): Declare.
21285 * internal-fn.c: Include asan.h.
21286 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21287 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21289 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21291 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21292 (C_SELFTEST_FLAGS): New.
21293 (CPP_SELFTEST_FLAGS): New.
21294 (SELFTEST_DEPS): New, from deps of s-selftest.
21295 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21296 (CPP_SELFTEST_DEPS): New.
21297 (selftest): Add dependency on s-selftest-c++.
21298 (s-selftest): Rename to...
21299 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21300 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21301 than SELFTEST_FLAGS.
21302 (selftest-gdb): Rename to...
21303 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21305 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21306 (selftest-valgrind): Rename to...
21307 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21309 (selftest-valgrind): Reintroduce as an alias for
21310 selftest-c-valgrind.
21311 (s-selftest-c++): New.
21312 (selftest-c++-gdb): New.
21313 (selftest-c++-valgrind): New.
21315 2017-07-06 Olivier Hainque <hainque@adacore.com>
21317 * gcc.c (process_command): When deciding if undefined variables
21318 should be ignored when processing specs, accept "gcc -v" as well.
21320 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21322 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21323 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21325 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21327 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21328 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21329 * config/arm/arm-tables.opt: Regenerate.
21330 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21332 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21334 2017-07-06 Carl Love <cel@us.ibm.com>
21336 * ChangeLog: Clean up from mid air collision
21338 2017-07-06 Carl Love <cel@us.ibm.com>
21340 * config/rs6000/rs6000-c.c: Add support for built-in functions
21341 vector signed int vec_subc (vector signed int, vector signed int);
21342 vector signed __int128 vec_subc (vector signed __int128,
21343 vector signed __int128);
21344 vector unsigned __int128 vec_subc (vector unsigned __int128,
21345 vector unsigned __int128);
21346 vector signed int vec_sube (vector signed int, vector signed int,
21347 vector signed int);
21348 vector unsigned int vec_sube (vector unsigned int,
21349 vector unsigned int,
21350 vector unsigned int);
21351 vector signed __int128 vec_sube (vector signed __int128,
21352 vector signed __int128,
21353 vector signed__int128);
21354 vector unsigned __int128 vec_sube (vector unsigned __int128,
21355 vector unsigned __int128,
21356 vector unsigned __int128);
21357 vector signed int vec_subec (vector signed int, vector signed int,
21358 vector signed int);
21359 vector unsigned int vec_subec (vector unsigned int,
21360 vector unsigned int,
21361 vector unsigned int);
21362 vector signed __int128 vec_subec (vector signed __int128,
21363 vector signed __int128,
21364 vector signed__int128);
21365 vector unsigned __int128 vec_subec (vector unsigned __int128,
21366 vector unsigned __int128,
21367 vector unsigned __int128);
21368 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21369 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21370 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21371 BU_ALTIVEC_OVERLOAD_X definitions.
21372 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21373 * doc/extend.texi: Update the built-in documentation file for the new
21374 built-in functions.
21376 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21379 * diagnostic-show-locus.c (layout::layout): Use start and finish
21380 spelling location for the start and finish of each range.
21381 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21382 Add unused aspect param.
21383 * input.c (expand_location_1): Add "aspect" param, and use it
21384 to access the correct part of the location.
21385 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21387 (expand_location_to_spelling_point): Likewise.
21388 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21389 param, and pass it to expand_location_1.
21391 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21393 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21394 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21395 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21396 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21397 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21398 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21399 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21400 _mm_maskz_getmant_ss): New intrinsics.
21401 (__builtin_ia32_getexpss128_mask): Changed to ...
21402 __builtin_ia32_getexpss128_round ... this.
21403 (__builtin_ia32_getexpsd128_mask): Changed to ...
21404 __builtin_ia32_getexpsd128_round ... this.
21405 * config/i386/i386-builtin-types.def
21406 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21407 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21408 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21409 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21410 __builtin_ia32_getmantss_mask_round): New builtins.
21411 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21412 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21413 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21414 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21415 * config/i386/sse.md
21416 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21417 avx512f_sgetexp<mode><mask_scalar_name>
21418 <round_saeonly_scalar_name> ... this.
21419 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21420 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21421 vgetexp<ssescalarmodesuffix>
21422 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21423 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21424 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21425 avx512f_vgetmant<mode><mask_scalar_name>
21426 <round_saeonly_scalar_name> ... this.
21427 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21428 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21429 vgetmant<ssescalarmodesuffix>
21430 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21431 %0<mask_scalar_operand4>, %1, %2
21432 <round_saeonly_scalar_mask_op4>, %3} ... this.
21433 * config/i386/subst.md (mask_scalar_operand4,
21434 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21435 round_saeonly_scalar_nimm_predicate): New subst attributes.
21437 2017-07-06 Julia Koval <julia.koval@intel.com>
21439 * config/i386/i386.c (ix86_erase_embedded_rounding):
21440 Remove code for old rounding pattern.
21442 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21444 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21446 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21448 * doc/sourcebuild.texi (Test Directives, Variants of
21449 dg-require-support): Add documentation for dg-require-stack-check.
21451 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21453 * config/i386/subst.md (mask_scalar, round_scalar,
21454 round_saeonly_scalar): New meta-templates.
21455 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21456 round_scalar_mask_operand3, round_scalar_mask_op3,
21457 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21458 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21459 round_saeonly_scalar_constraint,
21460 round_saeonly_scalar_prefix): New subst attribute.
21461 * config/i386/sse.md
21462 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21463 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21464 <round_scalar_name> ... this.
21465 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21466 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21467 <round_scalar_name> ... this.
21468 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21469 <sse>_vm<code><mode>3<mask_scalar_name>
21470 <round_saeonly_scalar_name> ... this.
21471 (v<plusminus_mnemonic><ssescalarmodesuffix>
21472 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21473 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21474 v<plusminus_mnemonic><ssescalarmodesuffix>
21475 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21476 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21477 (v<multdiv_mnemonic><ssescalarmodesuffix>
21478 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21479 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21480 v<multdiv_mnemonic><ssescalarmodesuffix>
21481 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21482 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21483 (v<maxmin_float><ssescalarmodesuffix>
21484 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21485 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21486 v<maxmin_float><ssescalarmodesuffix>
21487 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21488 %0<mask_scalar_operand3>, %1, %<iptr>2
21489 <round_saeonly_scalar_mask_op3>} ... this.
21491 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21493 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21494 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21496 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21497 Alan Hayward <alan.hayward@arm.com>
21498 David Sherwood <david.sherwood@arm.com>
21500 * combine.c (simplify_if_then_else): Remove "enum" before
21502 * compare-elim.c (can_eliminate_compare): Likewise.
21503 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21505 (aarch64_lookup_simd_builtin_type): Likewise.
21506 (aarch64_simd_builtin_type): Likewise.
21507 (aarch64_init_simd_builtin_types): Likewise.
21508 (aarch64_simd_expand_args): Likewise.
21509 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21511 (aarch64_reverse_mask): Likewise.
21512 (aarch64_simd_emit_reg_reg_move): Likewise.
21513 (aarch64_gen_adjusted_ldpstp): Likewise.
21514 (aarch64_ccmp_mode_to_code): Likewise.
21515 (aarch64_operands_ok_for_ldpstp): Likewise.
21516 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21517 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21519 (aarch64_min_divisions_for_recip_mul): Likewise.
21520 (aarch64_reassociation_width): Likewise.
21521 (aarch64_get_condition_code_1): Likewise.
21522 (aarch64_simd_emit_reg_reg_move): Likewise.
21523 (aarch64_simd_attr_length_rglist): Likewise.
21524 (aarch64_reverse_mask): Likewise.
21525 (aarch64_operands_ok_for_ldpstp): Likewise.
21526 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21527 (aarch64_gen_adjusted_ldpstp): Likewise.
21528 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21530 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21531 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21532 (arm_lookup_simd_builtin_type): Likewise.
21533 (arm_simd_builtin_type): Likewise.
21534 (arm_init_simd_builtin_types): Likewise.
21535 (arm_expand_builtin_args): Likewise.
21536 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21537 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21538 (ft32_setup_incoming_varargs): Likewise.
21539 (ft32_function_arg): Likewise.
21540 (ft32_function_arg_advance): Likewise.
21541 (ft32_pass_by_reference): Likewise.
21542 (ft32_arg_partial_bytes): Likewise.
21543 (ft32_valid_pointer_mode): Likewise.
21544 (ft32_addr_space_pointer_mode): Likewise.
21545 (ft32_addr_space_legitimate_address_p): Likewise.
21546 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21548 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21549 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21550 (iamcu_alignment): Likewise.
21551 (canonicalize_vector_int_perm): Likewise.
21552 (ix86_noce_conversion_profitable_p): Likewise.
21553 (ix86_mpx_bound_mode): Likewise.
21554 (ix86_operands_ok_for_move_multiple): Likewise.
21555 * config/microblaze/microblaze-protos.h
21556 (microblaze_expand_conditional_branch_reg): Likewise.
21557 * config/microblaze/microblaze.c
21558 (microblaze_expand_conditional_branch_reg): Likewise.
21559 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21561 (rs6000_reassociation_width): Likewise.
21562 (rs6000_invalid_binary_op): Likewise.
21563 (fusion_p9_p): Likewise.
21564 (emit_fusion_p9_load): Likewise.
21565 (emit_fusion_p9_store): Likewise.
21566 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21568 (riscv_hard_regno_mode_ok_p): Likewise.
21569 (riscv_address_insns): Likewise.
21570 (riscv_split_symbol): Likewise.
21571 (riscv_legitimize_move): Likewise.
21572 (riscv_function_value): Likewise.
21573 (riscv_hard_regno_nregs): Likewise.
21574 (riscv_expand_builtin): Likewise.
21575 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21576 (riscv_build_integer): Likewise.
21577 (riscv_split_integer): Likewise.
21578 (riscv_legitimate_constant_p): Likewise.
21579 (riscv_cannot_force_const_mem): Likewise.
21580 (riscv_regno_mode_ok_for_base_p): Likewise.
21581 (riscv_valid_base_register_p): Likewise.
21582 (riscv_valid_offset_p): Likewise.
21583 (riscv_valid_lo_sum_p): Likewise.
21584 (riscv_classify_address): Likewise.
21585 (riscv_legitimate_address_p): Likewise.
21586 (riscv_address_insns): Likewise.
21587 (riscv_load_store_insns): Likewise.
21588 (riscv_force_binary): Likewise.
21589 (riscv_split_symbol): Likewise.
21590 (riscv_force_address): Likewise.
21591 (riscv_legitimize_address): Likewise.
21592 (riscv_move_integer): Likewise.
21593 (riscv_legitimize_const_move): Likewise.
21594 (riscv_legitimize_move): Likewise.
21595 (riscv_address_cost): Likewise.
21596 (riscv_subword): Likewise.
21597 (riscv_output_move): Likewise.
21598 (riscv_canonicalize_int_order_test): Likewise.
21599 (riscv_emit_int_order_test): Likewise.
21600 (riscv_function_arg_boundary): Likewise.
21601 (riscv_pass_mode_in_fpr_p): Likewise.
21602 (riscv_pass_fpr_single): Likewise.
21603 (riscv_pass_fpr_pair): Likewise.
21604 (riscv_get_arg_info): Likewise.
21605 (riscv_function_arg): Likewise.
21606 (riscv_function_arg_advance): Likewise.
21607 (riscv_arg_partial_bytes): Likewise.
21608 (riscv_function_value): Likewise.
21609 (riscv_pass_by_reference): Likewise.
21610 (riscv_setup_incoming_varargs): Likewise.
21611 (riscv_print_operand): Likewise.
21612 (riscv_elf_select_rtx_section): Likewise.
21613 (riscv_save_restore_reg): Likewise.
21614 (riscv_for_each_saved_reg): Likewise.
21615 (riscv_register_move_cost): Likewise.
21616 (riscv_hard_regno_mode_ok_p): Likewise.
21617 (riscv_hard_regno_nregs): Likewise.
21618 (riscv_class_max_nregs): Likewise.
21619 (riscv_memory_move_cost): Likewise.
21620 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
21621 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
21622 (rl78_addr_space_address_mode): Likewise.
21623 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21625 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21626 (rs6000_reassociation_width): Likewise.
21627 (rs6000_invalid_binary_op): Likewise.
21628 (fusion_p9_p): Likewise.
21629 (emit_fusion_p9_load): Likewise.
21630 (emit_fusion_p9_store): Likewise.
21631 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
21632 (ok_for_simple_move_operands): Likewise.
21633 (ok_for_simple_move_strict_operands): Likewise.
21634 (ok_for_simple_arith_logic_operands): Likewise.
21635 (visium_legitimize_reload_address): Likewise.
21636 (visium_select_cc_mode): Likewise.
21637 (output_cbranch): Likewise.
21638 (visium_split_double_move): Likewise.
21639 (visium_expand_copysign): Likewise.
21640 (visium_expand_int_cstore): Likewise.
21641 (visium_expand_fp_cstore): Likewise.
21642 * config/visium/visium.c (visium_pass_by_reference): Likewise.
21643 (visium_function_arg): Likewise.
21644 (visium_function_arg_advance): Likewise.
21645 (visium_libcall_value): Likewise.
21646 (visium_setup_incoming_varargs): Likewise.
21647 (visium_legitimate_constant_p): Likewise.
21648 (visium_legitimate_address_p): Likewise.
21649 (visium_legitimize_address): Likewise.
21650 (visium_secondary_reload): Likewise.
21651 (visium_register_move_cost): Likewise.
21652 (visium_memory_move_cost): Likewise.
21653 (prepare_move_operands): Likewise.
21654 (ok_for_simple_move_operands): Likewise.
21655 (ok_for_simple_move_strict_operands): Likewise.
21656 (ok_for_simple_arith_logic_operands): Likewise.
21657 (visium_function_value_1): Likewise.
21658 (rtx_ok_for_offset_p): Likewise.
21659 (visium_legitimize_reload_address): Likewise.
21660 (visium_split_double_move): Likewise.
21661 (visium_expand_copysign): Likewise.
21662 (visium_expand_int_cstore): Likewise.
21663 (visium_expand_fp_cstore): Likewise.
21664 (visium_split_cstore): Likewise.
21665 (visium_select_cc_mode): Likewise.
21666 (visium_split_cbranch): Likewise.
21667 (output_cbranch): Likewise.
21668 (visium_print_operand_address): Likewise.
21669 * expmed.c (flip_storage_order): Likewise.
21670 * expmed.h (emit_cstore): Likewise.
21671 (flip_storage_order): Likewise.
21672 * genrecog.c (validate_pattern): Likewise.
21673 * hsa-gen.c (gen_hsa_addr): Likewise.
21674 * internal-fn.c (expand_arith_overflow): Likewise.
21675 * ira-color.c (allocno_copy_cost_saving): Likewise.
21676 * lra-assigns.c (find_hard_regno_for_1): Likewise.
21677 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
21678 (process_invariant_for_inheritance): Likewise.
21679 * lra-eliminations.c (move_plus_up): Likewise.
21680 * omp-low.c (lower_oacc_reductions): Likewise.
21681 * simplify-rtx.c (simplify_subreg): Likewise.
21682 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
21683 (TARGET_CHKP_BOUND_MODE): Likewise..
21684 * targhooks.c (default_chkp_bound_mode): Likewise.
21685 (default_setup_incoming_vararg_bounds): Likewise.
21686 * targhooks.h (default_chkp_bound_mode): Likewise.
21687 (default_setup_incoming_vararg_bounds): Likewise.
21688 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
21689 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
21690 (have_whole_vector_shift): Likewise.
21691 * tree-vect-stmts.c (vectorizable_load): Likewise.
21692 * doc/tm.texi: Regenerate.
21694 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21696 Graceful degrade if Binutils PR21472 is not available.
21699 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
21700 .rodata in flash test fails.
21701 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
21702 * confgure: Regenerate.
21703 * config.in: Regenerate.
21704 * config/avr/avr.c (avr_asm_named_section)
21705 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
21706 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
21707 (avr_asm_init_sections): Same.
21709 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21711 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
21712 (fma<VH:mode>4_intrinsic): Likewise.
21713 (*fmsub<VCVTF:mode>4): Likewise.
21714 (*fmsub<VH:mode>4_intrinsic): Likewise.
21716 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21719 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
21720 Don't depend on "optimize > 0".
21721 (out_movhi_r_mr, out_movqi_mr_r): Same.
21722 (out_movhi_mr_r, out_movqi_r_mr): Same.
21723 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
21724 io_address_operand on "optimize > 0".
21726 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21728 * tree-loop-distribution.c: Add general explanantion on the pass.
21729 (generate_loops_for_partition): Mark distributed loop.
21730 (pg_add_dependence_edges): New parameter. Handle alias data
21731 dependence specially and record it in the parameter if asked.
21732 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
21733 (init_partition_graph_vertices, add_partition_graph_edge): New.
21734 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
21735 (free_partition_graph_vdata, build_partition_graph): New.
21736 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
21737 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
21738 (data_ref_segment_size, latch_dominated_by_data_ref): New.
21739 (compute_alias_check_pairs, version_loop_by_alias_check): New.
21740 (version_for_distribution_p, finalize_partitions): New.
21741 (distribute_loop): Handle alias data dependence specially. Factor
21742 out loop fusion code as functions and call these functions.
21744 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21746 * tree-loop-distribution.c (classify_partition): New parameter and
21747 better handle reduction statement.
21748 (rdg_build_partitions): Revise comment.
21749 (distribute_loop): Compute statements in all partitions and pass it
21750 to classify_partition.
21752 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21754 * tree-loop-distribution.c (enum partition_type): New.
21755 (struct partition): New field type.
21756 (partition_merge_into): Add parameter. Update partition type.
21757 (data_dep_in_cycle_p, update_type_for_merge): New functions.
21758 (build_rdg_partition_for_vertex): Compute partition type.
21759 (rdg_build_partitions): Dump partition type.
21760 (distribute_loop): Update calls to partition_merge_into.
21762 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21764 * tree-loop-distribution.c (struct ddr_hasher): New.
21765 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
21767 (classify_partition): Call get_data_dependence.
21768 (pg_add_dependence_edges): Ditto.
21769 (distribute_loop): Release data dependence hash table.
21771 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21773 * tree-loop-distribution.c (ref_base_address): Delete.
21774 (similar_memory_accesses): Rename ...
21775 (share_memory_accesses): ... to this. Check if partitions access
21776 the same memory reference.
21777 (distribute_loop): Call share_memory_accesses.
21779 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21781 * tree-loop-distribution.c (struct partition): New field recording
21782 its data reference.
21783 (partition_alloc, partition_free): Init and release data refs.
21784 (partition_merge_into): Merge data refs.
21785 (build_rdg_partition_for_vertex): Collect data refs for partition.
21786 (pg_add_dependence_edges): Change parameters from vector to bitmap.
21788 (distribute_loop): Remve data refs from vertice data of partition
21791 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21793 * tree-loop-distribution.c (params.h): Include header file.
21794 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
21795 (datarefs_vec): New global var.
21796 (create_rdg_vertices): Use datarefs_vec directly.
21797 (free_rdg): Don't free data references.
21798 (build_rdg): Update use. Don't free data references.
21799 (distribute_loop): Compute global variable for data references.
21800 Bail out if there are too many data references.
21802 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21804 * tree-loop-distribution.c (loop_nest): New global var.
21805 (build_rdg): Use loop directly, rather than loop nest.
21806 (pg_add_dependence_edges): Remove loop nest parameter. Use global
21808 (distribute_loop): Compute global variable loop nest. Update use.
21810 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21812 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
21813 (partition_merge_into): New parameter. Dump reason for fusion.
21814 (distribute_loop): Update use of partition_merge_into.
21816 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21818 * tree-loop-distribution.c (bb_top_order_index): New.
21819 (bb_top_order_index_size, bb_top_order_cmp): New.
21820 (stmts_from_loop): Use topological order.
21821 (pass_loop_distribution::execute): Compute and release topological
21822 order for basic blocks.
21824 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21826 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
21829 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21831 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
21832 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
21833 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
21834 * internal-fn.def (LOOP_DIST_ALIAS): New.
21835 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
21836 (fold_loop_internal_call): ... this.
21837 (vect_loop_dist_alias_call): New function.
21838 (set_uid_loop_bbs): Call fold_loop_internal_call.
21839 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
21842 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21845 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
21846 Require dead FLAGS_REG at the beginning of a peephole.
21848 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21851 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
21852 arguments in the call to __builtin_ia32_sbb_u32.
21853 (_subborrow_u64): Swap _X and _Y arguments in the call to
21854 __builtin_ia32_sbb_u64.
21856 2017-07-04 Jakub Jelinek <jakub@redhat.com>
21859 * tree-vrp.c (compare_assert_loc): Turn into a function template
21860 with stable template parameter. Only test if a->e is NULL,
21861 !a->e == !b->e has been verified already. Use e == NULL or
21862 e != NULL instead of e or ! e tests. If stable is true, don't use
21863 iterative_hash_expr, on the other side allow a or b or both NULL
21864 and sort the NULLs last.
21865 (process_assert_insertions): Sort using compare_assert_loc<false>
21866 instead of compare_assert_loc, later sort using
21867 compare_assert_loc<true> before calling process_assert_insertions_for
21868 in a loop. Use break instead of continue once seen NULL pointer.
21870 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21872 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21873 Cortex-R7 and Cortex-R8 processors.
21875 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21877 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
21878 uninitialized while src is not.
21880 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
21882 * common/config/arm/arm-common.c: Adjust include path for
21884 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
21885 (arm-cpu.h): Create in build directory. Adjust dependency rules.
21886 (arm-cpu-data.h): Likewise.
21887 (arm-cpu-cdata.h): Likewise.
21888 * config/arm/arm-cpu.h: Delete.
21889 * config/arm/arm-cpu-cdata.h: Delete.
21890 * config/arm/arm-cpu-data.h: Delete.
21892 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
21894 * config/arm/arm-cpus.in (cortex-a55): New.
21895 (cortex-a75): Likewise.
21896 (cortex-a75.cortex-a55): Likewise.
21897 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
21899 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
21900 * config/arm/arm-cpu-cdata.h: Regenerate.
21901 * config/arm/arm-cpu-data.h: Regenerate.
21902 * config/arm/arm-cpu.h: Regenerate.
21903 * config/arm/arm-tables.opt: Regenerate.
21904 * config/arm/arm-tune.md: Regenerate.
21906 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21908 * haifa-sched.c (sched_create_recovery_edges): Update profile.
21910 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21912 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
21915 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
21917 PR tree-optimization/81292
21918 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
21919 full_string_p, also call adjust_related_strinfos if the adjustment
21920 is simple, otherwise invalidate related strinfos.
21922 2017-07-04 Martin Liska <mliska@suse.cz>
21925 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
21926 newly created variable as DECL_IGNORED_P.
21928 2017-07-04 Martin Liska <mliska@suse.cz>
21931 * ipa-inline.c (inline_small_functions):
21932 Use xstrdup_for_dump.
21934 2017-07-04 Tom de Vries <tom@codesourcery.com>
21936 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
21938 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21941 * config/darwin.c (darwin_function_switched_text_sections):
21944 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
21946 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
21948 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21950 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
21952 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21954 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
21955 min_profitable_iters, and th as inclusive lower bounds.
21956 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
21957 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
21958 for min_profitable_iters and min_profitable_estimate.
21959 (vect_transform_loop): Treat th as an inclusive lower bound.
21960 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21962 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21965 * config/darwin.c (darwin_function_switched_text_sections):
21966 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
21967 in two pieces, and suppress the use of buf.
21969 2017-07-03 Nathan Sidwell <nathan@acm.org>
21971 * hash-table.h (hash_table_mod1): Fix indentation.
21973 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21975 PR middle-end/81290
21976 * predict.c (force_edge_cold): Be more careful about propagation
21978 * profile-count.h (profile_probability::guessed,
21979 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
21981 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
21983 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
21985 * doc/invoke.texi (rcpc architecture extension): Document it.
21987 2017-07-03 Richard Biener <rguenther@suse.de>
21989 PR tree-optimization/60510
21990 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
21991 the scalar reduction PHI and use it.
21992 (vectorizable_reduction): Properly guard the single_defuse_cycle
21993 path for non-SLP reduction chains where we cannot use it.
21994 Rework reduc_def/index and vector type deduction. Rework
21995 vector operand gathering during reduction op code-gen.
21996 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
21997 chains dissolve the chain and leave it to non-SLP reduction
22000 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22002 * tree-data-ref.h (dr_alignment): Declare.
22003 * tree-data-ref.c (dr_alignment): New function.
22004 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22005 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22007 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22009 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22011 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22012 and base_misalignment fields.
22013 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22014 * tree-data-ref.c: Include builtins.h.
22015 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22016 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22017 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22018 * tree-vect-data-refs.c: Include tree-cfg.h.
22019 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22020 fields instead of calculating an alignment here.
22021 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22022 innermost_loop_behavior fields.
22024 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22026 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22028 (DR_STEP_ALIGNMENT): New macro.
22029 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22030 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22031 (create_data_ref): Print it.
22032 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22033 to tell whether the step preserves vector (mis)alignment.
22034 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22035 Move the check for an integer step and generalise to all INTEGER_CST.
22036 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22037 Print the outer step alignment.
22039 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22041 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22042 with offset_alignment.
22043 (DR_ALIGNED_TO): Delete.
22044 (DR_OFFSET_ALIGNMENT): New macro.
22045 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22046 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22047 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22048 (create_data_ref): Likewise.
22049 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22050 (vect_analyze_data_refs): Likewise.
22051 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22052 creating dummy innermost behavior.
22054 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22056 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22057 with a "innermost_loop_behavior *" and refeence tree.
22058 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22059 (create_data_ref): Update call accordingly.
22060 * tree-predcom.c (find_looparound_phi): Likewise.
22062 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22064 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22065 fields with dr_wrt_vec_loop.
22066 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22067 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22068 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22069 (vect_dr_behavior): New function.
22070 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22071 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22072 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22073 track whether the step preserves the misalignment.
22074 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22075 Use vect_dr_behavior.
22076 (vect_setup_realignment): Update call accordingly.
22077 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22078 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22079 call to vect_create_addr_base_for_vector_ref.
22080 (vect_create_cond_for_align_checks): Likewise.
22081 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22082 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22083 (vect_recog_mask_conversion_pattern): Likewise.
22084 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22085 (new_stmt_vec_info): Remove redundant zeroing.
22087 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22089 * common/config/arm/arm-common.c (arm_be8_option): New function.
22090 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22091 (ISA_ARMv6): Add isa_bit_be8.
22092 * config/arm/arm.h (arm_be8_option): Add prototype.
22093 (BE8_SPEC_FUNCTION): New define.
22094 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22095 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22096 (mlittle-endian): Similarly.
22097 (mbe8, mbe32): New options.
22098 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22099 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22101 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22103 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22105 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22107 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22108 (cleanup_tree_cfg_bb): Use it.
22109 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22111 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22113 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22116 * loop-doloop.c (add_test): Update profile.
22118 2017-07-03 Martin Liska <mliska@suse.cz>
22121 * sanopt.c (rewrite_usage_of_param): New function.
22122 (sanitize_rewrite_addressable_params): Likewise.
22123 (pass_sanopt::execute): Call rewrite_usage_of_param.
22125 2017-07-03 Richard Biener <rguenther@suse.de>
22127 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22128 back to using VIEW_CONVERT_EXPR.
22130 2017-07-03 Martin Liska <mliska@suse.cz>
22133 * doc/extend.texi: Document when a resolver function is
22134 generated for target_clones.
22136 2017-07-03 Martin Liska <mliska@suse.cz>
22138 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22139 and shadow memory of auto variables which are subject of
22140 use-after-scope sanitization.
22141 (asan_expand_mark_ifn): Add do set only when is_poison.
22143 2016-07-03 Richard Biener <rguenther@suse.de>
22145 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22147 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22148 (vectorizable_reduction): Perform reduction PHI creation when
22149 visiting a reduction PHI and adjust and simplify code generation
22150 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22151 (vect_transform_loop): Visit reduction PHIs.
22152 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22153 defs into the SLP tree.
22154 (vect_build_slp_tree): Reduction defs terminate the recursion.
22155 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22157 (vect_get_vec_defs_for_stmt_copy): Export.
22158 (vect_get_vec_defs): Likewise.
22159 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22161 (vect_get_vec_defs_for_stmt_copy): Declare.
22162 (vect_get_vec_defs): Likewise.
22164 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22166 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22167 parameter with a "loop" parameter and use it instead of the
22168 loop containing DR_STMT. Don't check simple_iv when doing
22169 BB analysis. Describe the two analysis modes in the comment.
22171 2017-07-03 Tom de Vries <tom@codesourcery.com>
22173 PR tree-optimization/69468
22174 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22175 (find_same_succ_bb): Handle ignore_edge_flags.
22177 2017-07-03 Tom de Vries <tom@codesourcery.com>
22179 PR tree-optimization/81192
22180 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22182 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22184 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22186 2017-07-03 Tom de Vries <tom@codesourcery.com>
22188 PR tree-optimization/81192
22189 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22190 BB_SAME_SUCC (bb) == NULL.
22192 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22194 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22197 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22199 * dumpfile.c: Include profile-count.h
22200 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22202 (insert_cond_bb): Update profile.
22203 * tree-cfg.h (insert_cond_bb): Update prototype.
22204 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22205 * tree-dump.c: Do not include tree-cfg.
22207 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22209 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22211 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22213 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22216 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22218 * tree-complex.c (expand_complex_div_wide): update profile.
22220 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22221 Alan Hayward <alan.hayward@arm.com>
22222 David Sherwood <david.sherwood@arm.com>
22224 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22225 (CORETYPES_H): New define.
22226 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22227 (insn-modes-inline.h, s-modes-inline-h): New rules.
22228 (generated_files): Add insn-modes-inline.h.
22229 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22230 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22231 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22232 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22233 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22234 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22235 (build/gencodes.o, build/genconditions.o): Likewise.
22236 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22237 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22238 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22239 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22240 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22241 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22242 * coretypes.h: Include everything up to real.h for generators.
22243 Include insn-modes.h first. Include wide-int-print.h after
22244 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22245 * machmode.h: Don't include insn-modes.h here.
22246 * function-tests.c: Remove includes of signop.h, machmode.h,
22247 double-int.h and wide-int.h.
22249 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22251 * optc-save-gen.awk: Likewise.
22252 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22253 * godump.c: Remove include of wide-int-print.h.
22254 * pretty-print.h: Likewise.
22255 * wide-int-print.cc: Likewise.
22256 * wide-int.cc: Likewise.
22257 * hash-map-tests.c: Remove include of signop.h.
22258 * hash-set-tests.c: Likewise.
22259 * rtl-tests.c: Likewise.
22260 * mkconfig.sh: Remove include of machmode.h.
22261 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22263 (emit_insn_modes_inline_h): ...this new function. Emit the code
22264 into an insn-modes-inline.h header file, adding appropriate
22265 include guards and end comments.
22266 (emit_insn_modes_c_header): Remove include of machmode.h.
22267 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22269 (main): Handle -i flag and call emit_insn_modes_inline_h when
22272 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22274 * tree-ssa-strlen.c (strinfo): Rename the length field to
22275 nonzero_chars. Add a full_string_p field.
22276 (compare_nonzero_chars, zero_length_string_p): New functions.
22277 (get_addr_stridx): Add an offset_out parameter.
22278 Use compare_nonzero_chars.
22279 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22280 (new_strinfo): Update after above changes to strinfo.
22281 (set_endptr_and_length): Set full_string_p.
22282 (get_string_length): Update after above changes to strinfo.
22283 (unshare_strinfo): Update call to new_strinfo.
22284 (maybe_invalidate): Likewise.
22285 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22286 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22287 as a uhwi instead of an shwi. Update after above changes to
22288 strinfo and new_strinfo.
22289 (zero_length_string): Assert that chainsi contains full strings.
22290 Use zero_length_string_p. Update call to new_strinfo.
22291 (adjust_related_strinfos): Update after above changes to strinfo.
22292 Copy full_string_p from origsi.
22293 (adjust_last_stmt): Use zero_length_string_p.
22294 (handle_builtin_strlen): Update after above changes to strinfo and
22295 new_strinfo. Install the lhs as the string length if the previous
22296 entry didn't describe a full string.
22297 (handle_builtin_strchr): Update after above changes to strinfo
22299 (handle_builtin_strcpy): Likewise.
22300 (handle_builtin_strcat): Likewise.
22301 (handle_builtin_malloc): Likewise.
22302 (handle_pointer_plus): Likewise.
22303 (handle_builtin_memcpy): Likewise. Track nonzero characters
22304 that aren't necessarily followed by a nul terminator.
22305 (handle_char_store): Likewise.
22307 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22309 PR tree-optimization/80769
22310 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22311 for malloc and calloc. Document the new invariant that all related
22312 strinfos have delayed lengths or none do.
22313 (verify_related_strinfos): Move earlier in file.
22314 (set_endptr_and_length): New function, split out from...
22315 (get_string_length): ...here. Also set the lengths of related
22317 (zero_length_string): Assert that chainsi has known (rather than
22319 (adjust_related_strinfos): Likewise.
22321 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22323 PR tree-optimization/81136
22324 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22325 assert that two references with the same misalignment have the same
22326 compile-time misalignment if those compile-time misalignments
22329 2017-07-01 Andi Kleen <ak@linux.intel.com>
22331 * print-tree.c (print_node): Print all attributes.
22333 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22335 * cfg.c (scale_bbs_frequencies): New function.
22336 * cfg.h (scale_bbs_frequencies): Declare it.
22337 * cfgloopanal.c (single_likely_exit): Cleanup.
22338 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22340 (scale_loop_profile): Likewise.
22341 (loop_version): Likewise.
22342 (create_empty_loop_on_edge): Update.
22343 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22344 scale_loop_frequencies, scale_loop_profile, loopify,
22345 loop_version): Update prototypes.
22346 * modulo-sched.c (sms_schedule): Update.
22347 * predict.c (unlikely_executed_edge_p): Also check probability.
22348 (probably_never_executed_edge_p): Fix typo.
22349 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22350 * tree-parloops.c (gen_parallel_loop): Update.
22351 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22352 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22353 * tree-ssa-loop-split.c (split_loop): Update.
22354 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22355 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22356 (vect_loop_versioning): Update.
22357 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22359 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22361 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22363 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22365 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22366 to keep profile consistent.
22368 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22370 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22371 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22372 * profile-count.h (max_safe_multiplier): Make unsigned.
22373 (profile_count::guessed_zero): New.
22375 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22377 * bb-reorder.c (fix_up_crossing_landing_pad,
22378 fix_crossing_conditional_branches): Use make_single_succ_edge
22379 to keep profile consistent.
22381 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22383 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22386 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22389 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22390 the right scopes, make sure cond_jump isn't preserved between multiple
22391 iterations. Search for fallthru edge whenever there are 3+ edges and
22392 use find_fallthru_edge for it.
22394 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22396 Patch by Alexander Monakov <amonakov@ispras.ru>
22397 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22398 probabilities consistently.
22400 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22402 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22403 emit_cmp_and_jump_insns.
22405 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22408 * tree-inline.c (expand_call_inline): Combine profile statuses.
22410 2017-06-30 Andrew Pinski <apinski@cavium.com>
22412 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22413 fold_stmt returned true.
22415 2017-06-30 Nathan Sidwell <nathan@acm.org>
22417 * ggc.h (empty_string): Delete.
22418 * cfgexpand.c (expand_asm_stmt): Use plain "".
22419 * optabs.c (expand_asm_memory_barrier): Likewise.
22420 * stringpool.c (empty_string): Delete.
22421 (digit_vector, digit_string): Delete.
22422 (ggc_alloc_string): Use plain "", don't optimize single digit
22423 strings. Use ggc_alloc_atomic.
22425 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22427 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22428 comparison set and one other set, use the cost of the non-comparison
22431 2017-06-30 Nathan Sidwell <nathan@acm.org>
22433 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22436 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22438 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22439 loops. Remove now unneeded calls to gimple_switch_set_label() that
22440 just set removed labels to NULL_TREE.
22442 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22444 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22445 (set_range_info): ...here. Only call set_range_info_raw if domain
22447 (set_nonzero_bits): Call set_range_info_raw.
22448 * tree-ssanames.h (set_range_info_raw): New.
22450 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22453 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22454 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22455 of nonimmediate_operand and <store_mask_constraint> instead of m
22456 for the input operand. For V8FI iterator, always split if input
22457 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22458 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22459 <store_mask_predicate> instead of register_operand and
22460 <store_mask_constraint> instead of v for the input operand. Make
22461 sure both operands aren't MEMs for if not <mask_applied>.
22463 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22465 * lto-wrapper.c (copy_file) Close both file descriptors before
22468 2017-06-30 Martin Liska <mliska@suse.cz>
22471 * multiple_target.c (create_dispatcher_calls): Make ifunc
22472 also for function that don't have calls or are not referenced.
22474 2017-06-30 Richard Biener <rguenther@suse.de>
22476 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22477 analyze the first scalar stmt. Move vector type computation
22478 for the BB case here from ...
22479 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22480 live operation processing in the SLP case properly.
22482 2017-06-30 Richard Biener <rguenther@suse.de>
22484 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22486 2017-06-30 Martin Liska <mliska@suse.cz>
22489 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22490 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22492 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22494 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22497 2017-06-30 Richard Biener <rguenther@suse.de>
22499 PR tree-optimization/81249
22500 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22501 condition reduction result to original scalar type.
22503 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22505 * profile-count.h (enum profile_quality): Fix typos and whitespace
22508 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22510 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22511 type for branch probabilities.
22513 2017-06-29 Julian Brown <julian@codesourcery.com>
22514 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22516 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22517 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22518 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22519 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22521 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22523 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22524 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22525 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22526 CC usage from generic code to here.
22527 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22528 CC usage into the target macros.
22530 2017-06-29 Maya Rashish <coypu@sdf.org>
22532 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22535 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22537 * arm/arm-builtins.c: Include profile-count.h
22538 * except.c (sjlj_emit_function_enter): Use
22539 profile_probability::unlikely.
22541 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22543 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22544 and tocrel_offset be pointer args rather than implicitly using
22546 (legitimate_constant_pool_address_p, rs6000_emit_move,
22547 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22548 tocrel_offset and use in toc_relative_expr_p call.
22549 (print_operand, print_operand_address): Use static tocrel_base_oac
22550 and tocrel_offset_oac.
22551 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22554 2017-06-29 Maya Rashish <coypu@sdf.org>
22556 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22558 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22560 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22561 objects, take into account only the alignment of 'op0' and 'mode1' if
22564 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22566 * ccmp.c (ccmp_tree_comparison_p): New function.
22567 (ccmp_candidate_p): Update to use above function.
22568 (get_compare_parts): New function.
22569 (expand_ccmp_next): Update to use new functions.
22570 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22572 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22573 take mode as argument.
22574 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22575 * expr.c (expand_expr_real_1): Pass mode as argument.
22577 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22579 * combine.c (combine_instructions): Print insns to dump_file, together
22582 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22584 * asan.c (asan_emit_stack_protection): Update.
22585 (create_cond_insert_point): Update.
22586 * auto-profile.c (afdo_propagate_circuit): Update.
22587 * basic-block.h (struct edge_def): Turn probability to
22588 profile_probability.
22589 (EDGE_FREQUENCY): Update.
22590 * bb-reorder.c (find_traces_1_round): Update.
22591 (better_edge_p): Update.
22592 (sanitize_hot_paths): Update.
22593 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22594 (make_single_succ_edge): Update.
22595 (check_bb_profile): Update.
22596 (dump_edge_info): Update.
22597 (update_bb_profile_for_threading): Update.
22598 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
22599 probabilitycount to 0.
22600 * cfgbuild.c (compute_outgoing_frequencies): Update.
22601 * cfgcleanup.c (try_forward_edges): Update.
22602 (outgoing_edges_match): Update.
22603 (try_crossjump_to_edge): Update.
22604 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
22605 (expand_gimple_tailcall): Update.
22606 (construct_init_block): Use make_single_succ_edge.
22607 (construct_exit_block): Use make_single_succ_edge.
22608 * cfghooks.c (verify_flow_info): Update.
22609 (redirect_edge_succ_nodup): Update.
22610 (split_edge): Update.
22611 (account_profile_record): Update.
22612 * cfgloopanal.c (single_likely_exit): Update.
22613 * cfgloopmanip.c (scale_loop_profile): Update.
22614 (set_zero_probability): Remove.
22615 (duplicate_loop_to_header_edge): Update.
22616 * cfgloopmanip.h (loop_version): Update prototype.
22617 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
22618 (force_nonfallthru_and_redirect): Update.
22619 (update_br_prob_note): Update.
22620 (rtl_verify_edges): Update.
22621 (purge_dead_edges): Update.
22622 (rtl_lv_add_condition_to_bb): Update.
22623 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
22624 * cgraphunit.c (init_lowered_empty_function): Update.
22625 (cgraph_node::expand_thunk): Update.
22626 * cilk-common.c: Include profile-count.h
22627 * dojump.c (inv): Remove.
22628 (jumpifnot): Update.
22629 (jumpifnot_1): Update.
22630 (do_jump_1): Update.
22632 (do_jump_by_parts_greater_rtx): Update.
22633 (do_compare_rtx_and_jump): Update.
22634 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
22635 do_jump_1. do_compare_rtx_and_jump): Update prototype.
22636 * dwarf2cfi.c: Include profile-count.h
22637 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
22638 (sjlj_emit_dispatch_table): Likewise.
22639 * explow.c: Include profile-count.h
22640 * expmed.c (emit_store_flag_force): Update.
22641 (do_cmp_and_jump): Update.
22642 * expr.c (compare_by_pieces_d::generate): Update.
22643 (compare_by_pieces_d::finish_mode): Update.
22644 (emit_block_move_via_loop): Update.
22645 (store_expr_with_bounds): Update.
22646 (store_constructor): Update.
22647 (expand_expr_real_2): Update.
22648 (expand_expr_real_1): Update.
22649 * expr.h (try_casesi, try_tablejump): Update prototypes.
22650 * gimple-pretty-print.c (dump_probability): Update.
22651 (dump_profile): New.
22652 (dump_gimple_label): Update.
22653 (dump_gimple_bb_header): Update.
22654 * graph.c (draw_cfg_node_succ_edges): Update.
22655 * hsa-gen.c (convert_switch_statements): Update.
22656 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
22657 (find_if_case_1): Update.
22658 (find_if_case_2): Update.
22659 * internal-fn.c (expand_arith_overflow_result_store): Update.
22660 (expand_addsub_overflow): Update.
22661 (expand_neg_overflow): Update.
22662 (expand_mul_overflow): Update.
22663 (expand_vector_ubsan_overflow): Update.
22664 * ipa-cp.c (good_cloning_opportunity_p): Update.
22665 * ipa-split.c (split_function): Use make_single_succ_edge.
22666 * ipa-utils.c (ipa_merge_profiles): Update.
22667 * loop-doloop.c (add_test): Update.
22668 (doloop_modify): Update.
22669 * loop-unroll.c (compare_and_jump_seq): Update.
22670 (unroll_loop_runtime_iterations): Update.
22671 * lra-constraints.c (lra_inheritance): Update.
22672 * lto-streamer-in.c (input_cfg): Update.
22673 * lto-streamer-out.c (output_cfg): Update.
22674 * mcf.c (adjust_cfg_counts): Update.
22675 * modulo-sched.c (sms_schedule): Update.
22676 * omp-expand.c (expand_omp_for_init_counts): Update.
22677 (extract_omp_for_update_vars): Update.
22678 (expand_omp_ordered_sink): Update.
22679 (expand_omp_for_ordered_loops): Update.
22680 (expand_omp_for_generic): Update.
22681 (expand_omp_for_static_nochunk): Update.
22682 (expand_omp_for_static_chunk): Update.
22683 (expand_cilk_for): Update.
22684 (expand_omp_simd): Update.
22685 (expand_omp_taskloop_for_outer): Update.
22686 (expand_omp_taskloop_for_inner): Update.
22687 * omp-simd-clone.c (simd_clone_adjust): Update.
22688 * optabs.c (expand_doubleword_shift): Update.
22689 (expand_abs): Update.
22690 (emit_cmp_and_jump_insn_1): Update.
22691 (expand_compare_and_swap_loop): Update.
22692 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
22693 * predict.c (predictable_edge_p): Update.
22694 (edge_probability_reliable_p): Update.
22695 (set_even_probabilities): Update.
22696 (combine_predictions_for_insn): Update.
22697 (combine_predictions_for_bb): Update.
22698 (propagate_freq): Update.
22699 (estimate_bb_frequencies): Update.
22700 (force_edge_cold): Update.
22701 * profile-count.c (profile_count::dump): Add missing space into dump.
22702 (profile_count::debug): Add newline.
22703 (profile_count::differs_from_p): Explicitly convert to unsigned.
22704 (profile_count::stream_in): Update.
22705 (profile_probability::dump): New member function.
22706 (profile_probability::debug): New member function.
22707 (profile_probability::differs_from_p): New member function.
22708 (profile_probability::differs_lot_from_p): New member function.
22709 (profile_probability::stream_in): New member function.
22710 (profile_probability::stream_out): New member function.
22711 * profile-count.h (profile_count_quality): Rename to ...
22712 (profile_quality): ... this one.
22713 (profile_probability): New.
22714 (profile_count): Update.
22715 * profile.c (compute_branch_probabilities): Update.
22716 * recog.c (peep2_attempt): Update.
22717 * sched-ebb.c (schedule_ebbs): Update.
22718 * sched-rgn.c (find_single_block_region): Update.
22719 (compute_dom_prob_ps): Update.
22720 (schedule_region): Update.
22721 * sel-sched-ir.c (compute_succs_info): Update.
22722 * stmt.c (struct case_node): Update.
22723 (do_jump_if_equal): Update.
22724 (get_outgoing_edge_probs): Update.
22725 (conditional_probability): Update.
22726 (emit_case_dispatch_table): Update.
22727 (expand_case): Update.
22728 (expand_sjlj_dispatch_table): Update.
22729 (emit_case_nodes): Update.
22730 * targhooks.c: Update.
22731 * tracer.c (better_p): Update.
22732 (find_best_successor): Update.
22733 * trans-mem.c (expand_transaction): Update.
22734 * tree-call-cdce.c: Update.
22735 * tree-cfg.c (gimple_split_edge): Upate.
22736 (move_sese_region_to_fn): Upate.
22737 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
22738 * tree-eh.c (lower_resx): Upate.
22739 (cleanup_empty_eh_move_lp): Upate.
22740 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22741 * tree-inline.c (copy_edges_for_bb): Update.
22742 (copy_cfg_body): Update.
22743 * tree-parloops.c (gen_parallel_loop): Update.
22744 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
22745 (gimple_gen_time_profiler): Update.
22746 * tree-ssa-dce.c (remove_dead_stmt): Update.
22747 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22748 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
22749 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
22750 (unloop_loops): Update.
22751 (try_peel_loop): Update.
22752 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22753 * tree-ssa-loop-split.c (connect_loops): Update.
22754 (split_loop): Update.
22755 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22756 (hoist_guard): Update.
22757 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
22758 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
22759 (value_replacement): Update.
22760 * tree-ssa-reassoc.c (branch_fixup): Update.
22761 * tree-ssa-tail-merge.c (replace_block_by): Update.
22762 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
22763 (create_edge_and_update_destination_phis): Update.
22764 (compute_path_counts): Update.
22765 (recompute_probabilities): Update.
22766 (update_joiner_offpath_counts): Update.
22767 (freqs_to_counts_path): Update.
22768 (duplicate_thread_path): Update.
22769 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
22770 (struct switch_conv_info): Update.
22771 (gen_inbound_check): Update.
22772 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22773 (vect_do_peeling): Update.
22774 (vect_loop_versioning): Update.
22775 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22776 (optimize_mask_stores): Update.
22777 * ubsan.c (ubsan_expand_null_ifn): Update.
22778 * value-prof.c (gimple_divmod_fixed_value): Update.
22779 (gimple_divmod_fixed_value_transform): Update.
22780 (gimple_mod_pow2): Update.
22781 (gimple_mod_pow2_value_transform): Update.
22782 (gimple_mod_subtract): Update.
22783 (gimple_mod_subtract_transform): Update.
22784 (gimple_ic): Update.
22785 (gimple_stringop_fixed_value): Update.
22786 (gimple_stringops_transform): Update.
22787 * value-prof.h: Update.
22789 2017-06-29 Carl Love <cel@us.ibm.com>
22791 * config/rs6000/rs6000-c.c: Add support for built-in functions
22792 vector signed int vec_signed (vector float);
22793 vector signed long long vec_signed (vector double);
22794 vector signed int vec_signed2 (vector double, vector double);
22795 vector signed int vec_signede (vector double);
22796 vector signed int vec_signedo (vector double);
22797 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
22798 instruction generator.
22799 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22800 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
22801 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
22803 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
22804 vunsignede_v2df): Add define_expands.
22805 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
22806 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
22807 VEC_UNSIGNEDO): Add definitions.
22808 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22809 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
22810 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
22811 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
22812 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
22813 * config/rs6000/altivec.h (vec_signed, vec_signed2,
22814 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
22815 vec_unsignede, vec_unsignedo): Add builtin defines.
22816 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
22818 * doc/extend.texi: Update the built-in documentation file for the
22819 new built-in functions.
22821 2017-06-29 Richard Biener <rguenther@suse.de>
22823 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
22824 reduction chains to LOOP_VINFO_REDUCTIONS.
22825 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
22826 SLP reductions after processing reduction chains.
22828 2017-06-29 Nathan Sidwell <nathan@acm.org>
22830 * builtins.c (fold_builtin_FUNCTION): Use
22831 lang_hooks.decl_printable_name.
22833 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
22835 PR middle-end/81194
22836 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
22837 with only one label.
22838 * stmt.c (expand_case): Assert NCASES is greater than one.
22840 2017-06-29 Richard Biener <rguenther@suse.de>
22842 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
22844 (group_case_labels): Likewise.
22845 (find_taken_edge): Push sanity checking on val to workers...
22846 (find_taken_edge_cond_expr): ... here
22847 (find_taken_edge_switch_expr): ... and here, handle cases
22848 with just a default label.
22849 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
22850 (group_case_labels): Likewise.
22851 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
22852 group_case_labels does anything cleanup the CFG again.
22854 2017-06-29 Bin Cheng <bin.cheng@arm.com>
22856 PR tree-optimization/81196
22857 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
22858 exit condition comparing two IVs.
22860 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
22862 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
22863 profile to the dummy entry at the end of the list of architectures.
22864 * config/arm/arm-cpu-cdata.h: Regenerated.
22866 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22867 Michael Collison <michael.collison@arm.com>
22870 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
22872 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
22873 (*aarch64_reg_<mode>3_minus_mask): New pattern.
22874 (*aarch64_<optab>_reg_di3_mask2): New pattern.
22875 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
22876 of shift when the shift amount is masked with constant equal to
22877 the size of the mode.
22878 * config/aarch64/predicates.md (subreg_lowpart_operator): New
22881 2017-06-29 Martin Liska <mliska@suse.cz>
22883 * config/i386/i386.opt: Change range from [1,5] to [0,5].
22885 2017-06-29 Yury Gribov <tetra2005@gmail.com>
22888 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
22891 (edge_growth_cache_entry::edge_growth_cache_entry): New
22893 (reset_edge_growth_cache): Update to use constructor.
22895 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22897 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
22898 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22899 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22901 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
22903 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
22904 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
22906 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
22908 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
22909 (*-linux-uclibc*): Add t-uclibc tmake_file.
22910 * config/t-musl: New.
22911 * config/t-uclibc: New.
22913 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
22915 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
22917 (gen_comm_data): Emit architectural setting of arch_prof.
22918 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
22920 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
22921 (armv8-m.base, armv8-m.main): Likewise.
22922 * arm-protos.h (arm_build_target): Add profile field.
22923 (arch_option): Likewise.
22924 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
22926 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
22927 arm_active_target.profile.
22929 2017-06-28 Richard Biener <rguenther@suse.de>
22931 PR middle-end/81227
22932 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
22933 TYPE_OVERFLOW_WRAPS.
22934 * match.pd (negate_expr_p): Likewise.
22935 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
22936 fold_build2, not fold_binary.
22938 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22940 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
22941 Convert memory address to Pmode.
22942 (aarch64_print_operand): Assert MEM operands are always Pmode.
22944 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22947 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
22948 Remove redundant if.
22949 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
22950 * config/arm/aarch-common-protos.h
22951 (aarch_forward_to_shift_is_not_shifted_re): Remove.
22952 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
22954 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
22957 * multiple_target.c (create_dispatcher_calls): Set the default
22958 clone to be static, not public.
22960 2017-06-28 Richard Biener <rguenther@suse.de>
22962 * tree-vect-loop.c (vectorizable_reduction): Move special
22963 cond reduction IV var creation ...
22964 (vect_create_epilog_for_reduction): ... here. Remove induction_index
22965 parameter. Use STMT_VINFO_VECTYPE.
22966 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
22969 2017-06-28 Martin Liska <mliska@suse.cz>
22972 * ipa-visibility.c (non_local_p): Handle visibility.
22974 2017-06-28 Martin Liska <mliska@suse.cz>
22977 * common.opt: Add IntegerRange to various options.
22978 * opt-functions.awk (integer_range_info): New function.
22979 * optc-gen.awk: Add integer_range_info to cl_options struct.
22980 * opts-common.c (decode_cmdline_option): Handle
22981 CL_ERR_INT_RANGE_ARG.
22982 (cmdline_handle_error): Likewise.
22983 * opts.c (print_filtered_help): Show valid interval in
22984 when --help is provided.
22985 * opts.h (struct cl_option): Add range_min and range_max fields.
22986 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
22988 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
22990 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
22991 (x * C EQ/NE y * C): New transformation.
22993 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
22995 * genmultilib (combination_space): Accept '+' in option names.
22997 2017-06-28 Martin Liska <mliska@suse.cz>
23000 * asan.c (instrument_derefs): Bail out inner references
23001 that are hard register variables.
23003 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23006 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23007 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23009 2017-06-28 Richard Biener <rguenther@suse.de>
23011 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23012 (vect_get_slp_defs): Adjust.
23013 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23015 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23017 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23018 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23019 (vectorizable_reduction): Adjust.
23020 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23022 (vect_get_slp_defs): Likewise.
23023 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23024 (vectorizable_bswap): Adjust.
23025 (vectorizable_call): Likewise.
23026 (vectorizable_conversion): Likewise.
23027 (vectorizable_assignment): Likewise.
23028 (vectorizable_shift): Likewise.
23029 (vectorizable_operation): Likewise.
23030 (vectorizable_store): Likewise.
23031 (vectorizable_condition): Likewise.
23032 (vectorizable_comparison): Likewise.
23034 2017-06-28 Michael Collison <michael.collison@arm.com>
23037 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23039 (arm_gen_movmemqi): Removed unused variable 'i'.
23040 Convert 'for' loop into 'while' loop.
23041 (arm_expand_prologue): Remove last unnecessary set of insn.
23042 (thumb_pop): Remove unused variable 'pushed_words'.
23043 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23045 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23047 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23048 * config/s390/s390-protos.h: Add prototype of
23049 s390_rel_address_ok_p.
23050 * config/s390/s390.c (s390_got_symbol): New function.
23051 (s390_rel_address_ok_p): New function.
23052 (legitimize_pic_address): Use s390_rel_address_ok_p.
23053 (s390_load_got): Use s390_got_symbol.
23054 (s390_option_override): Issue error if
23055 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23056 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23058 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23060 2017-06-27 Andrew Pinski <apinski@cavium.com>
23062 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23063 (X * copysign (1.0, X)): New pattern.
23064 (X * copysign (1.0, -X)): New pattern.
23065 (copysign (-1.0, CST)): New pattern.
23067 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23069 * genmultilib (combination_space): Remove variable.
23070 Validate reuse rules against regular expression for any sequence
23071 of multilib options in any order.
23073 2017-06-27 Michael Collison <michael.collison@arm.com>
23075 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23076 call aarch64_split_simd_combine.
23077 * (aarch64_combine_internal<mode>): Delete pattern.
23078 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23079 Allow register and subreg operands.
23081 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23083 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23084 specific need, just fallback on defaults.
23085 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23087 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23088 Olivier Hainque <hainque@adacore.com>
23090 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23092 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23093 targets. Pick a default if no particular attempt applied.
23094 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23097 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23099 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23100 (x86_64-wrs-vxworks7): Likewise.
23102 2017-06-27 Marek Polacek <polacek@redhat.com>
23105 * ubsan.c (instrument_null): Check get_base_address's result for null.
23107 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23109 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23111 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23113 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23114 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23115 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23116 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23117 New function types.
23118 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23119 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23120 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23121 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23122 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23123 BUILT_IN_FEUPDATEENV): New builtins.
23124 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23125 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23126 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23127 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23129 (builtin_structptr_types): Adjust size.
23130 * tree.c (builtin_structptr_types): Add four entries.
23132 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23133 Olivier Hainque <hainque@adacore.com>
23135 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23136 (TLS_SYM): New local macro, forcing reference to __tls__ on
23137 link command lines for VxWorks 7 RTPs, triggering initialization
23139 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23140 OS features TLS support, true for RTPs on VxWorks 7.
23141 * config/vxworks.c (vxworks_override_options): Setup emutls
23144 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23146 * predict.c (test_prediction_value_range): Use -1U instead of -1
23147 to avoid narrowing conversion warning.
23148 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23149 to avoid narrowing conversion warning.
23150 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23152 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23154 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23156 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23157 64bit configurations.
23158 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23159 (SIZE_TYPE): Likewise.
23160 * config/vxworks.c (vxworks_emutls_var_fields): Use
23161 long_unsigned_type_node instead of unsigned_type_node as the offset
23162 field type, which is "pointer" mode in emutls.c.
23164 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23167 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23169 PR middle-end/81207
23170 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23171 gimple_vuse copying separately from gimple_vdef copying.
23173 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23175 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23178 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23179 Olivier Hainque <hainque@adacore.com>
23181 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23182 for all vxworks7 targets.
23183 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23184 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23185 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23186 variations for VX6/VX7 and 32/64bits later on in ...
23187 (VXWORKS_LIB_SPEC): Leverage new macros.
23188 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23189 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23191 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23193 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23194 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23196 2017-06-26 Carl Love <cel@us.ibm.com>
23198 * config/rs6000/rs6000-c.c: Add support for built-in functions
23199 vector bool char vec_reve (vector bool char);
23200 vector signed char vec_reve (vector signed char);
23201 vector unsigned char vec_reve (vector unsigned char);
23202 vector bool int vec_reve (vector bool int);
23203 vector signed int vec_reve (vector signed int);
23204 vector unsigned int vec_reve (vector unsigned int);
23205 vector bool long long vec_reve (vector bool long long);
23206 vector signed long long vec_reve (vector signed long long);
23207 vector unsigned long long vec_reve (vector unsigned long long);
23208 vector bool short vec_reve (vector bool short);
23209 vector signed short vec_reve (vector signed short);
23210 vector double vec_reve (vector double);
23211 vector float vec_reve (vector float);
23212 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23213 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23214 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23215 (altivec_vreve): New pattern.
23216 * config/rs6000/altivec.h (vec_reve): New define.
23217 * doc/extend.texi (vec_rev): Update the built-in documentation file
23218 for the new built-in functions.
23220 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23222 PR tree-optimization/71815
23223 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23225 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23227 (slsr_process_phi): Likewise.
23228 (replace_uncond_cands_and_profitable_phis): Don't replace a
23229 multiply candidate with a stride of 1 (copy or cast).
23230 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23232 (lowest_cost_path): Likewise.
23233 (total_savings): Likewise.
23235 2017-06-26 Richard Biener <rguenther@suse.de>
23238 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23239 Use def_builtin_pure for all gather builtins.
23241 2017-06-26 Richard Biener <rguenther@suse.de>
23243 PR tree-optimization/81203
23244 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23245 non-dominating BBs.
23247 2017-06-26 Marek Polacek <polacek@redhat.com>
23250 * doc/invoke.texi: Document -Wmultistatement-macros.
23252 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23254 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23255 arm_neon_ok_no_float_abi effective target.
23257 2017-06-26 Richard Biener <rguenther@suse.de>
23259 PR tree-optimization/80928
23260 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23261 (copy_bbs): Set BB_DUPLICATED flag early.
23262 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23264 (execute_on_shrinking_pred): Likewise.
23265 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23266 BB_DUPLICATED blocks.
23267 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23268 iterate over all PHIs considering removal of *gsi.
23270 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23272 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23275 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23277 * config/rs6000/rs6000-string.c: (expand_block_clear,
23278 do_load_for_compare, select_block_compare_mode,
23279 compute_current_alignment, expand_block_compare,
23280 expand_strncmp_align_check, expand_strn_compare,
23281 expand_block_move, rs6000_output_load_multiple)
23282 Move functions related to string/block move/compare
23283 to a separate file.
23284 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23285 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23286 for this function which is now used in two files.
23287 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23288 * config.gcc: Add rs6000-string.o to extra_objs for
23289 targets powerpc*-*-* and rs6000*-*-*.
23291 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23294 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23295 32-bit, since indexed is not valid for DImode.
23296 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23297 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23298 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23299 (define_peephole2 for Altivec d-form store): Likewise.
23302 * multiple_target.c (create_dispatcher_calls): Only create the
23303 dispatcher call if the function is the default clone of a
23304 versioned function.
23306 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23308 PR middle-end/80902
23309 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23310 a call, force the call to not be a tail call.
23312 2017-06-23 Jeff Law <law@redhat.com>
23314 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23317 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23319 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23320 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23321 handling for early expansion of vector shifts (sl,sr,sra,rl).
23322 (builtin_function_type): Add vector shift right instructions
23323 to the unsigned argument list.
23325 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23327 rtl-optimizatoin/79286
23328 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23329 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23330 trap. PIC register plus a const unspec without offset can never trap.
23332 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23334 * tree.h (builtin_structptr_type): New type.
23335 (builtin_structptr_types): Declare new array.
23336 * tree.c (builtin_structptr_types): New array.
23337 (free_lang_data, build_common_tree_nodes): Use it.
23339 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23342 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23345 2017-06-22 Matt Turner <mattst88@gmail.com>
23347 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23348 Lake models to skylake case. Assume skylake for unknown
23349 models with clflushopt.
23351 2017-06-22 Jeff Law <law@redhat.com>
23353 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23354 frame sizes that do not satisfy aarch64_uimm12_shift.
23356 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23358 * profile-count.h (apply_probability,
23359 apply_scale, probability_in): Fix checks for zero.
23361 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23363 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23364 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23366 2016-06-22 Richard Biener <rguenther@suse.de>
23368 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23369 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23370 REDUC_MAX_EXPR support.
23371 (vectorizable_reduction): Likewise.
23372 (vect_create_epilog_for_reduction): Likewise.
23374 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23376 * match.pd (A / (1 << B) -> A >> B): New.
23377 * generic-match-head.c: Include optabs-tree.h.
23378 * gimple-match-head.c: Likewise.
23379 * optabs-tree.h (target_supports_op_p): New.
23380 * optabs-tree.c (target_supports_op_p): New.
23382 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23384 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23385 $gcc_cv_ld --help output.
23386 (gcc_cv_ld_demangle): Likewise.
23387 (gcc_cv_ld_eh_frame_hdr): Likewise.
23388 (gcc_cv_ld_pie): Likewise.
23389 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23390 (gcc_cv_ld_buildid): Likewise.
23391 (gcc_cv_ld_sysroot): Likewise.
23392 (ld_bndplt_support): Likewise.
23393 (ld_pushpopstate_support): Likewise.
23394 * configure: Regenerate.
23395 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23397 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23400 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23401 operands indexes to avoid gap between operands and match_dups.
23403 2017-06-21 Andrew Pinski <apinski@cavium.com>
23405 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23406 Increment Arith_shift and Arith_shift_reg by 1.
23407 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23409 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23410 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23411 (aarch64_strip_extend): Add new argument and test for it.
23412 (aarch64_cheap_mult_shift_p): New function.
23413 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23414 add a cost if it is true.
23415 Update calls to aarch64_strip_extend.
23416 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23418 2017-06-21 Andrew Pinski <apinski@cavium.com>
23420 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23422 (thunderxt88): Likewise.
23423 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23424 (thunderx_prefetch_tune): New variable.
23425 (thunderx2t99_prefetch_tune): Update for the correct values.
23426 (thunderxt88_tunings): New variable.
23427 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23428 generic_prefetch_tune.
23429 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23431 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23433 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23434 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23435 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23436 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23437 (aarch64_atomic_cas<mode>, GPI): Likewise.
23439 2017-06-21 Martin Liska <mliska@suse.cz>
23441 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23442 statements on cold and hot labels.
23443 * predict.c (tree_estimate_probability_bb): Remove the
23444 prediction from this place.
23446 2017-06-21 Martin Liska <mliska@suse.cz>
23448 PR tree-optimization/79489
23449 * gimplify.c (maybe_add_early_return_predict_stmt): New
23451 (gimplify_return_expr): Call the function.
23452 * predict.c (tree_estimate_probability_bb): Remove handling
23454 * predict.def: Update comment about early return predictor.
23455 * gimple-predict.h (is_gimple_predict): New function.
23456 * predict.def: Change default value of early return to 66.
23457 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23459 * passes.def: Put pass_strip_predict_hints to the beginning of
23462 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23464 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23465 FUNCTION_DECL declarations.
23466 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23468 (dwaf2out_decl): Likewise.
23469 * godump.c (go_early_global_decl): Skip call to the real debug hook
23470 for FUNCTION_DECL declarations.
23471 * passes.c (rest_of_decl_compilation): Skip call to the
23472 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23473 -fdump-go-spec is passed.
23475 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23477 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23478 Reorder for compactness.
23479 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23480 (def_builtin_pure, def_builtin_pure2): New functions.
23481 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23483 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23485 * match.pd (nop_convert): New predicate.
23486 ((A +- CST1) +- CST2): Allow some NOP conversions.
23488 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23491 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23492 with ctors/dtors if GOVD_SHARED is set.
23494 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23496 * config/aarch64/aarch64.md (movti_aarch64):
23497 Emit mov rather than orr.
23498 (movtf_aarch64): Likewise.
23499 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23500 Emit mov rather than orr.
23502 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23504 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23505 Swap alternatives, make integer dup more expensive.
23507 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23509 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23510 Return true for non-tls symbols.
23512 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23514 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23515 (cortex-a75): Likewise.
23516 (cortex-a75.cortex-a55): Likewise.
23517 * config/aarch64/aarch64-tune.md: Regenerate.
23518 * doc/invoke.texi (-mtune): Document new values for -mtune.
23520 2017-06-21 Tom de Vries <tom@codesourcery.com>
23522 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23523 stack_size feature.
23524 (Effective-Target Keywords, Other attributes): Suggest using
23525 dg-add-options stack_size feature to get stack limit in stack_size
23526 effective target documentation.
23528 2017-06-21 Julian Brown <julian@codesourcery.com>
23529 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23531 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23532 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23533 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23535 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23536 attribute type list for neon_multiply.
23537 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23538 attribute type list for neon_multiply.
23539 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23540 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23541 attribute type list for neon_multiply.
23542 * config/arm/types.md (crypto_pmull): Add.
23543 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23544 attribute type list.
23546 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23548 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23551 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23553 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23554 to make sure not to dereference a NULL cost_classes_ptr pointer.
23556 2017-06-20 Carl Love <cel@us.ibm.com>
23558 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23559 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23560 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23561 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23562 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23563 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23564 VMULOSW): New enum "unspec" values.
23565 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23566 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23567 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23568 altivec_vmulosw): New patterns.
23569 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23570 VMULOSW): Add definitions.
23572 2017-06-20 Julia Koval <julia.koval@intel.com>
23574 * config/i386/i386.c: Fix rounding expand for new pattern.
23575 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23577 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23579 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23580 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23582 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23584 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23587 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23589 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23590 architecture, then by implementer ID.
23591 * config/aarch64/aarch64-tune.md: Regenerate.
23593 2017-06-20 Richard Biener <rguenther@suse.de>
23595 PR middle-end/81097
23596 * fold-const.c (split_tree): Fold to type before negating.
23598 2017-06-20 David Malcolm <dmalcolm@redhat.com>
23600 * diagnostic-show-locus.c
23601 (selftest::test_fixit_deletion_affecting_newline): New function.
23602 (selftest::diagnostic_show_locus_c_tests): Call it.
23604 2017-06-20 Andreas Schwab <schwab@suse.de>
23607 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
23610 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23612 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
23613 __ARM_FEATURE_COPROC according to support.
23615 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23617 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23618 Rewritten to avoid overflow for > 32-bit pointers.
23621 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
23622 by removing enum keyword.
23623 (ubsan_type_descriptor): Likewise. Formatting fix.
23626 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
23627 splitter): Require TARGET_SSE2 in the condition.
23629 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
23632 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
23633 for doing vector set of SFmode on ISA 3.0.
23634 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
23635 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
23637 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
23638 SFmode value into a V4SF variable that was extracted from another
23639 V4SF variable without converting the element to double precision
23640 and back to single precision vector format.
23641 (vsx_insert_extract_v4sf_p9_2): Likewise.
23643 2017-06-19 Jakub Jelinek <jakub@redhat.com>
23645 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
23646 in UWHI to avoid undefined overflow.
23649 * ubsan.h (enum ubsan_encode_value_phase): New.
23650 (ubsan_encode_value): Change second argument to
23651 enum ubsan_encode_value_phase with default value of
23652 UBSAN_ENCODE_VALUE_GENERIC.
23653 * ubsan.c (ubsan_encode_value): Change second argument to
23654 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
23655 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
23656 create_tmp_var_raw instead of create_tmp_var and use a
23658 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
23659 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
23660 ubsan_encode_value callers.
23663 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
23664 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
23665 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
23667 2017-06-19 Richard Biener <rguenther@suse.de>
23669 PR middle-end/81118
23670 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
23671 estimates if we changed anything.
23673 2017-06-19 Richard Biener <rguenther@suse.de>
23675 PR tree-optimization/80887
23676 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
23677 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
23678 simplified lookups, then reset mprts_hook.
23679 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
23681 (try_to_simplify): Likewise.
23683 2017-06-19 Martin Liska <mliska@suse.cz>
23686 * gimplify.c (gimplify_switch_expr):
23687 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
23689 2017-06-19 Martin Liska <mliska@suse.cz>
23691 * doc/install.texi: Document that PGO runs in 4 stages.
23693 2017-06-19 Martin Liska <mliska@suse.cz>
23696 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
23697 to dispatcher function name.
23698 * multiple_target.c (replace_function_decl): New function.
23699 (create_dispatcher_calls): Redirect both edges and references.
23701 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
23703 * profile-count.c (profile_count::dump): Dump quality.
23704 (profile_count::differs_from_p): Update for unsigned val.
23705 * profile-count.h (profile_count_quality): New enum.
23706 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
23708 2017-06-19 Richard Biener <rguenther@suse.de>
23710 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
23711 struct function as arg.
23712 (estimate_numbers_of_iterations): Export overload with loop arg.
23713 (free_numbers_of_iterations_estimates_loop): Use an overload of
23714 free_numbers_of_iterations_estimates instead.
23715 * tree-cfg.c (remove_bb): Adjust.
23716 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
23717 * tree-parloops.c (gen_parallel_loop): Likewise.
23718 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
23720 (tree_unroll_loops_completely): Likewise.
23721 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
23722 Use an overload instead and export.
23723 (estimated_loop_iterations): Adjust.
23724 (max_loop_iterations): Likewise.
23725 (likely_max_loop_iterations): Likewise.
23726 (estimate_numbers_of_iterations): Take struct function as arg
23728 (loop_exits_before_overflow): Adjust.
23729 (free_numbers_of_iterations_estimates_loop): Use an overload.
23730 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
23731 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
23733 2017-06-19 Richard Biener <rguenther@suse.de>
23736 * ipa-prop.c (find_constructor_constant_at_offset): Handle
23737 RANGE_EXPR conservatively.
23739 2017-06-16 Carl Love <cel@us.ibm.com>
23741 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23742 definitions for vec_float, vec_float2, vec_floato,
23743 vec_floate built-ins.
23744 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
23745 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
23747 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
23748 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
23749 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
23750 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
23751 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
23752 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
23753 vec_floato): Add builtin defines.
23754 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
23755 Update the built-in documentation file for the new built-in
23758 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23760 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
23761 (mthumb): Mark as the negative of -marm.
23763 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23765 * doc/invoke.texi (ARM Options, -mcpu): Document supported
23767 (ARM Options, -mtune): Document that this accepts the same
23768 extension options as -mcpu.
23769 (ARM Options, -mfpu): Document addition of -mfpu=auto.
23771 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23773 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
23774 permitted extensions.
23776 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23778 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
23779 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
23780 (armv8-m.main): Add option +nodsp.
23781 * config/arm/arm-cpu-cdata.h: Regenerated.
23783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23785 * config/arm/t-fuchsia: New file.
23786 * config.gcc (arm*-*-fuchsia*): Use it.
23788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23790 * config/arm/t-symbian: Rewrite for new option infrastructure.
23792 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23794 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
23795 (MULTILIB_REQUIRED): Likewise.
23797 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23799 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
23800 (MULTILIB_RESUE): Likewise.
23801 (MULTILIB_MATCHES): Likewise.
23802 (MULTLIB_REQUIRED): Likewise.
23804 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23806 * config/arm/t-rtems: Rewrite for new option framework.
23808 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23810 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
23811 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
23812 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
23813 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
23814 * config/arm/t-multilib: ... here.
23815 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
23816 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
23817 armv7-a and armv8*-a when A-profile libraries have not been built.
23818 * config/arm/t-rmprofile: Rewrite.
23820 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23822 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
23823 with a backslash. Remove the backslash after substituting unescaped
23825 * doc/fragments.texi (MULTILIB_REUSE): Document it.
23827 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23829 * config.gcc: (arm*-*-*): When building a-profile libraries, force
23830 the driver to pass through the default setting of -mfloat-abi.
23831 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
23833 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
23834 (all_feat_combs): New rule.
23835 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
23837 * config/arm/t-aprofile: Rewrite.
23839 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23841 * config/arm/arm.h (FPUTYPE_AUTO): Define.
23842 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
23843 fpu is not specified by the user/command-line.
23844 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
23845 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
23846 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
23847 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
23848 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
23849 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
23851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23853 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
23854 * config/arm/t-arm-elf: Rewritten.
23856 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23858 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
23859 have some floating-point instructions.
23860 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
23861 (TARGET_MAYBE_HARD_FLOAT): New macro.
23862 * config/arm/arm-builtins.c (arm_init_builtins): Use
23863 TARGET_MAYBE_HARD_FLOAT.
23864 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
23866 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23868 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
23869 (configargs.h): Include it.
23870 (arm_print_hint_for_fpu_option): New function.
23871 (arm_parse_fpu_option): New function.
23872 (candidate_extension): New class.
23873 (arm_canon_for_multilib): New function.
23874 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
23875 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
23876 (ARCH_CANONICAL_SPECS): New macro.
23877 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
23879 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23881 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
23882 are set after handling multilib fragments. Set target_cpu_default2
23885 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23887 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
23889 (arm*-*-*): Set target_cpu_default2 to a quoted string.
23890 * config/arm/parsecpu.awk (check_cpu): Validate any extension
23892 (check_arch): Likewise.
23893 * config/arm/arm.c (arm_configure_build_target): Handle
23894 TARGET_CPU_DEFAULT being a string constant. Scan any feature
23895 options in the default.
23897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23899 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
23900 when an option is an alias of another.
23901 * config/arm/parsecpu.awk (optalias): New parser token.
23902 (gen_comm_data): Mark non-alias options as such. Emit entries
23903 for extension aliases.
23904 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
23905 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
23906 (armv6kz, armv6zk, armv6t2): Likewise.
23907 (armv7): Make vfpv3-d16 an alias.
23908 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
23910 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
23911 Sort in canonical order.
23912 (armv8-a): Sort in canonical order.
23913 (armv8.1-a, armv8.2-a): Likewise.
23914 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
23916 (cortex-a9): Sort in canonical order.
23917 * config/arm/arm.c (selftests.h): Include it.
23918 (arm_test_cpu_arch_data): New function.
23919 (arm_run_self_tests): New function.
23920 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
23921 (targetm): Move declaration to the end of the file.
23922 * arm-cpu-cdata.h: Regenerated.
23924 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23926 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
23927 call to target_mode_check describing the type of option passed.
23928 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
23929 (arm_target_thumb_only): Use arm_parse_arch_option_name or
23930 arm_parse_cpu_option_name to match parameters against list of
23932 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
23933 arm_arch_core_flags data structure.
23934 * config/arm/arm-cpu_cdata.h: Regenerated.
23936 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23938 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
23940 (arm_print_hint_for_cpu_option): Likewise.
23941 (arm_print_hint_for_arch_option): Likewise.
23942 (arm_parse_cpu_option_name): Likewise.
23943 (arm_parse_arch_option_name): Likewise.
23944 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
23945 of entries in the all_fpus list.
23946 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
23947 (arm_parse_cpu_option_name): Declare.
23948 (arm_parse_arch_option_name): Declare.
23949 (arm_parse_option_features): Declare.
23950 (arm_intialize_isa): Declare.
23951 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
23953 (gen_comm_data): ... here. Make definitions non-static.
23954 * config/arm/arm-cpu-data.h: Regenerated.
23955 * config/arm/arm-cpu-cdata.h: Regenerated.
23957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23959 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
23960 (cpu_arch_extension): New structure.
23961 (cpu_arch_option, arch_option, cpu_option): New structures.
23962 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
23963 architecture types.
23964 (gen_data): Generate new format data tables.
23965 * config/arm/arm.c (cpu_tune): New structure.
23966 (cpu_option, processors): Delete.
23967 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
23968 (arm_print_hint_for_cpu_option): ... this and ...
23969 (arm_print_hint_for_arch_option): ... this.
23970 (arm_parse_arch_cpu_name): Delete. Replace with ...
23971 (arm_parse_cpu_option_name): ... this and ...
23972 (arm_parse_arch_option_name): ... this.
23973 (arm_unrecognized_feature): Change type of target parameter to
23975 (arm_parse_arch_cpu_features): Delete. Replace with ...
23976 (arm_parse_option_features): ... this.
23977 (arm_configure_build_target): Rework to use new configuration data
23979 (arm_print_tune_info): Rework for new configuration data tables.
23980 * config/arm/arm-cpu-data.h: Regenerated.
23981 * config/arm/arm-cpu.h: Regenerated.
23983 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23985 * Makefile.in (OBJS): Move sbitmap.o from here ...
23986 (OBJS-libcommon): ... to here.
23988 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23990 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
23991 (ISA_ALL_CRYPTO): New macro.
23992 (ISA_ALL_SIMD): New macro
23993 (ISA_ALL_FP): New macro.
23994 * config/arm/arm.c (fpu_bitlist): Update initializer.
23995 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
23997 (arm9e): Add fpu. Add option for nofp
23998 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
23999 (arm926ej-s, arm1026ej-s): Likewise.
24000 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24001 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24002 neon-fp16, neon-vfpv4, nofp and nosimd.
24003 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24004 (cortex-a8): Add fpu. Add option for nofp.
24005 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24006 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24007 (cortex-r4f): Add fpu.
24008 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24009 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24011 (cortex-r8): Likewise.
24012 (cortex-m4): Add fpu. Add option for nofp.
24013 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24014 (cortex-a17.cortex-a7): Likewise.
24015 (cortex-a32): Add fpu. Add options for crypto and nofp.
24016 (cortex-a35, cortex-a53): Likewise.
24017 (cortex-a57): Add fpu. Add option for crypto.
24018 (cortex-a72, cortex-a73): Likewise.
24019 (exynos-m1): Likewise.
24020 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24021 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24022 (cortex-m33): Add fpu. Add option for nofp.
24023 * config/arm/arm-cpu-cdata.h: Regenerated
24024 * config/arm/arm-cpu-data.h: Regenerated.
24026 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24028 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24029 (armv5te, armv5tej): Likewise.
24030 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24031 (armv7): Add options fp and vfpv3-d16.
24032 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24033 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24035 (armv7ve): Likewise.
24036 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24037 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24038 (armv8-a): Add nocrypto option.
24039 (armv8.1-a, armv8.2-a): Likewise.
24040 (armv8-m.main): add options fp, fp.dp and nofp.
24042 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24044 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24046 (armv8-a+crc): Delete.
24047 (armv8.1-a): Add options simd, crypto and nofp.
24048 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24049 (armv8.2-a+fp16): Delete.
24050 (armv8-m.main): Add option dsp.
24051 (armv8-m.main+dsp): Delete.
24052 (cortex-a8): Add fpu. Add nofp option.
24053 (cortex-a9): Add fpu. Add nofp and nosimd options.
24054 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24055 link to main cpu and architecture data structures.
24056 (gen_comm_data): Only put isa attributes from the main architecture
24058 (option): New statement for architecture and CPU entries.
24059 * arm.c (struct cpu_option): New structure.
24060 (struct processors): Add entry for options.
24061 (arm_unrecognized_feature): New function.
24062 (arm_parse_arch_cpu_name): Ignore any characters after the first
24064 (arm_parse_arch_cpu_feature): New function.
24065 (arm_configure_build_target): Separate out any CPU and architecture
24066 features and parse separately. Don't error out if -mfpu=auto is
24067 used with only an architecture string.
24068 (arm_print_asm_arch_directives): New function.
24069 (arm_file_start): Call it.
24070 * config/arm/arm-cpu-cdata.h: Regenerated.
24071 * config/arm/arm-cpu-data.h: Likewise.
24072 * config/arm/arm-tables.opt: Likewise.
24074 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24076 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24077 assembler when it is not -mfpu=auto.
24079 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24081 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24082 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24083 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24084 (ASM_CPU_SPEC): Rewrite.
24085 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24086 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24087 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24088 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24089 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24090 copied string is NUL-terminated. Also strip any characters prefixed
24092 (arm_rewrite_selected_arch): New function.
24093 (arm_rewrite_march): New function.
24095 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24097 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24098 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24099 (march, mcpu, mtune): Convert to string-based options.
24100 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24101 (arm_parse_arch_cpu_name): New function.
24102 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24103 identify selected architecture or CPU.
24104 (arm_option_save): New function.
24105 (TARGET_OPTION_SAVE): Redefine.
24106 (arm_option_restore): Restore string options.
24107 (arm_option_print): Print string options.
24109 2017-06-16 Martin Sebor <msebor@redhat.com>
24111 PR tree-optimization/80933
24112 PR tree-optimization/80934
24113 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24114 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24115 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24116 (gimple_fold_builtin): Call them.
24118 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24120 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24121 as unlikely; update profile.
24123 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24125 * predict.c (force_edge_cold): Handle declaring edges impossible
24128 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24130 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24132 (try_unroll_loop_completely): Fix reporting.
24134 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24136 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24138 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24141 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24142 if given a non-constant argument for an intrinsic which requires a
24145 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24147 * profile.c (compare_freqs): New function.
24148 (branch_prob): Sort edge list.
24149 (find_spanning_tree): Assume that the list is priority sorted.
24151 2017-06-16 Richard Biener <rguenther@suse.de>
24153 PR tree-optimization/81090
24154 * passes.def (pass_record_bounds): Remove.
24155 * tree-pass.h (make_pass_record_bounds): Likewise.
24156 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24157 make_pass_record_bounds): Likewise.
24158 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24159 not free niter estimates at the beginning but at the end.
24160 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24162 2017-06-16 Richard Biener <rguenther@suse.de>
24164 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24165 initializer to workaround ICE in host GCC 4.8.
24167 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24169 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24171 (clone_inlined_nodes): Update.
24173 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24175 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24176 prefetch settings, and enable prefetching by default at -O3.
24178 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24180 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24181 Set flag_prefetch_loop_arrays according to tuning data.
24183 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24185 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24186 New tune structure.
24187 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24188 [Unrelated to main purpose of the patch] Place the pointer field last
24189 to enable type checking errors when tune structure are wrongly merged.
24190 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24191 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24192 (thunderx2t99_prefetch_tune): New tune constants.
24193 (tune_params *_tunings): Update all tunings (no functional change).
24194 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24195 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24196 from tunings structures.
24198 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24201 * ubsan.c (instrument_null): Add T argument, use it instead
24202 of computing it based on IS_LHS.
24203 (instrument_object_size): Likewise.
24204 (pass_ubsan::execute): Adjust instrument_null and
24205 instrument_object_size callers to pass gimple_get_lhs or
24206 gimple_assign_rhs1 result to it. Use instrument_null instead of
24207 calling get_base_address and instrument_mem_ref. Handle
24208 aggregate call arguments for object-size sanitization.
24210 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24212 PR tree-optimization/81089
24213 * tree-vrp.c (is_masked_range_test): Validate operands of
24216 2017-06-15 Martin Sebor <msebor@redhat.com>
24219 * dumpfile.c (dump_register): Avoid calling memset to initialize
24220 a class with a default ctor.
24221 * gcc.c (struct compiler): Remove const qualification.
24222 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24223 * hash-table.h: Ditto.
24224 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24226 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24227 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24229 * params.h (struct param_info): Make struct members non-const.
24230 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24231 with default initialization.
24232 * vec.h (vec_copy_construct, vec_default_construct): New helper
24234 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24235 with vec_copy_construct.
24236 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24237 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24238 * doc/invoke.texi (-Wclass-memaccess): Document.
24240 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24242 * emit-rtl.h (is_leaf): Update comment about local
24243 register allocator.
24245 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24248 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24249 for a variable to have a section before checking if the section has a
24251 Set section to.persistent if persistent attribute is set.
24252 Warn if .persistent attribute is used on an automatic variable.
24254 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24256 PR rtl-optimization/80474
24257 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24259 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24261 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24264 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24266 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24267 (casesi): Emit bounds checking as RTL.
24268 (casesi_internal_mips16_<mode>): Remove bounds checking.
24270 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24272 * config/xtensa/xtensa.c (xtensa_option_override): Append
24273 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24274 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24275 xtensa_doloop_hooks): Define unconditionally.
24276 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24278 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24279 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24280 for it in xtensa_option_override.
24281 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24282 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24284 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24286 * doc/cppopts.texi: Document '-' special value to -MF.
24288 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24290 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24291 (cortex_a53_fconst): Likewise.
24292 (cortex_a53_fpmul): Likewise.
24293 (cortex_a53_f_load_64): Likewise.
24294 (cortex_a53_f_load_many): Likewise.
24295 (cortex_a53_advsimd_alu): Likewise.
24296 (cortex_a53_advsimd_alu_q): Likewise.
24297 (cortex_a53_advsimd_mul): Likewise.
24298 (cortex_a53_advsimd_mul_q): Likewise.
24299 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24300 Add missing fmul, r2f_cvt and fconst cases.
24302 2017-06-14 Richard Biener <rguenther@suse.de>
24304 PR middle-end/81088
24305 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24307 (fold_binary_loc): When associating do not treat pre-existing
24308 TREE_OVERFLOW on literal constants as a reason to allow
24309 TREE_OVERFLOW on associated literal constants.
24311 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24313 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24314 (MASK_FEATURES): New macro.
24315 * config/sparc/sparc.c (sparc_option_override): Remove the special
24316 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24318 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24320 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24321 a division of 0 if non-call exceptions are enabled.
24323 2017-06-14 Andrew Pinski <apinski@cavium.com>
24324 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24327 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24328 Improve vector initialization code gen for only variable case.
24330 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24332 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24334 2017-06-14 Richard Biener <rguenther@suse.de>
24336 PR tree-optimization/81083
24337 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24340 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24342 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24343 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24344 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24345 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24346 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24347 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24349 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24351 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24352 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24354 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24356 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24358 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24360 * config/rs6000/t-rtems: Don't handle SPE.
24362 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24364 * config/rs6000/t-linux: Don't handle SPE.
24366 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24368 * config/rs6000/eabispe.h: Delete file.
24370 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24372 * config/rs6000/t-spe: Delete file.
24374 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24376 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24377 (rs6000_legitimate_offset_address_p): Return false for anything in
24378 V2SImode or V2SFmode.
24380 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24382 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24383 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24384 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24386 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24387 (rs6000_legitimate_offset_address_p): Ditto.
24388 (rs6000_emit_move): Ditto.
24389 (rs6000_init_builtins): Remove V4HI_type_node.
24391 2017-06-13 Martin Liska <mliska@suse.cz>
24394 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24395 (gate_asan): Likewise.
24396 * asan.h (asan_no_sanitize_address_p): Remove the function.
24397 (sanitize_flags_p): New function.
24398 * builtins.def: Fix coding style.
24399 * common.opt: Use renamed enum value.
24400 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24401 * doc/extend.texi: Document no_sanitize attribute.
24402 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24403 to SANITIZE_UNDEFINED_NONDEFAULT.
24404 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24405 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24406 Use sanitize_flags_p.
24407 * gimplify.c (gimplify_function_tree): Likewise.
24408 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24409 * opts.c (parse_no_sanitize_attribute): New function.
24410 (common_handle_option): Use renamed enum value.
24411 * opts.h (parse_no_sanitize_attribute): Declare.
24412 * tree.c (sanitize_flags_p): New function.
24413 * tree.h: Declared here.
24414 * tsan.c: Use sanitize_flags_p.
24415 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24416 (instrument_mem_ref): Likewise.
24417 (instrument_bool_enum_load): Likewise.
24418 (do_ubsan_in_current_function): Remove the function.
24419 (pass_ubsan::execute): Use sanitize_flags_p.
24420 * ubsan.h: Remove do_ubsan_in_current_function
24421 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24422 (dump_function_to_file): Use it here.
24424 2017-06-13 Martin Jambor <mjambor@suse.cz>
24426 PR tree-optimization/80803
24427 PR tree-optimization/81063
24428 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24429 (propagate_subaccesses_across_link): Enqueue subtree whenever
24430 necessary instead of relying on the caller.
24432 2017-06-13 Martin Jambor <mjambor@suse.cz>
24434 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24435 that have a first_link.
24436 (sort_and_splice_var_accesses): Do not check first_link before
24438 (subtree_mark_written_and_enqueue): Likewise.
24439 (propagate_all_subaccesses): Likewise and do not stop at first
24440 parent with a first_link.
24442 2017-06-13 Martin Jambor <mjambor@suse.cz>
24444 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24447 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24449 * match.pd: New pattern.
24451 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24453 * tree-vrp.c (is_masked_range_test): New function.
24454 (register_edge_assert_for): Determine ranges for
24457 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24459 PR tree-optimization/67328
24460 * fold-const.c (maskable_range_p): New function.
24461 (build_range_check): Generate bittests if possible.
24463 2017-06-13 Martin Liska <mliska@suse.cz>
24465 * gimple-pretty-print.c (dump_probability): Add new argument.
24466 (dump_edge_probability): Dump both probability and count.
24467 (dump_gimple_label): Likewise.
24468 (dump_gimple_bb_header): Likewise.
24470 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24473 * config/avr/avr-devices.c: Fix indentation.
24474 * config/avr/gen-avr-mmcu-specs.c: Dito.
24476 2017-06-13 Richard Biener <rguenther@suse.de>
24478 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24479 instead get vector type from stmt_info.
24480 (vectorizable_reduction): Adjust. Remove dead code.
24482 2017-06-13 Richard Biener <rguenther@suse.de>
24484 PR middle-end/81065
24485 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24486 case of C * (x * C2 + C3).
24487 (fold_addr_of_array_ref_difference): Properly fold index difference.
24489 2017-06-12 David S. Miller <davem@davemloft.net>
24492 * config/sparc/sparc.md (return expander): Emit frame blockage if
24493 function uses alloca.
24495 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24497 * combine.c (make_field_assignment): Check len rather than the mode
24498 precision when calling force_to_mode.
24500 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24502 Support multilibs and devices that see flash in RAM address range.
24505 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24506 (avr_mcu_t) <flash_pm_offset>: New field.
24507 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24508 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24509 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24510 (AVR_TINY_PM_OFFSET): Remove macro.
24511 * config/avr/avr.opt (-mshort-calls): New option.
24512 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24513 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24514 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24515 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24516 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24517 instead of avr_arch->have_jmp_call.
24518 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24519 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24520 avr_arch->flash_pm_offset to define.
24521 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24522 new field flash_pm_offset. Add entry for avrxmega3.
24523 (avr_texinfo): Add entry for avrxmega3.
24524 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24525 attiny212, attiny214,
24526 attiny412, attiny414, attiny416, attiny417,
24527 attiny814, attiny816, attiny817,
24528 attiny1614, attiny1616, attiny1617,
24529 attiny3214, attiny3216, attiny3217.
24530 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24531 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24532 (avr_print_operand_address) [AVR_TINY]: Same.
24533 (avr_asm_init_sections) <readonly_data_section>: Only patch
24534 callback if avr_arch->flash_pm_offset = 0.
24535 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24536 for rodata if avr_arch->flash_pm_offset != 0.
24537 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24538 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24539 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24540 (m_options): Append opt_rcall.
24541 (m_dirnames): Append dir_rcall.
24542 * config/avr/t-multilib: Regenerate.
24544 * configure.ac [target=avr]: Check whether avrxmega3 default
24545 linker description file works as needed.
24546 * configure: Regenerate.
24547 * doc/avr-mmcu.texi: Regenerate.
24548 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24549 <__AVR_ARCH__>: Document avrxmega3 and 103.
24550 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24551 <__AVR_SHORT_CALLS__>: Document it.
24552 <__AVR_PM_BASE_ADDRESS__>: Document it.
24553 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24554 (AVR Variable Attributes) <progmem>: Document this is
24555 not needed for avrxmega3.
24556 (AVR Named Address Spaces) <__flash>: Dito.
24558 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24560 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24562 2017-06-12 Doug Rupp <rupp@adacore.com>
24564 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24565 Append vxworks-stdint.h to the tm_file list.
24566 * config/vxworks-stdint.h: New file.
24568 2017-06-12 Martin Liska <mliska@suse.cz>
24570 PR tree-optimization/81041
24571 * tree-profile.c (gimple_gen_ic_func_profiler):
24572 Create an extra BB in profile-generate
24573 (gimple_gen_time_profiler): Likewise.
24575 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24577 PR tree-optimization/81003
24578 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24579 (update_range_test): Use it instead of force_gimple_operand_gsi.
24581 2017-06-12 Richard Biener <rguenther@suse.de>
24583 PR tree-optimization/81053
24584 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24585 with backedge value not defined in loop. Simplify def stmt
24588 2017-06-11 Tom de Vries <tom@codesourcery.com>
24591 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24593 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24594 nvptx_cannot_force_const_mem.
24596 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24598 * opts.c (finish_options): Move test for flag_split_stack after
24599 it has been initialized.
24601 2017-06-11 Jason Merrill <jason@redhat.com>
24603 * tree.h (id_equal): New.
24604 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
24605 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
24606 instead of strcmp of IDENTIFIER_POINTER.
24608 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24610 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
24611 (mark_all_inlined_calls_cdtor): Fix formating.
24612 (inline_transform): Rescale profile before inlining.
24614 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24616 * cgraph.h (cgraph_edge::clone): Update prototype.
24617 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
24618 (cgraph_node::create_clone): Update.
24619 (cgraph_node::create_version_clone): Update.
24620 * tree-inline.c (copy_bb): Update.
24621 (expand_call_inline): Update.
24623 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24625 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
24626 factored out from ...
24627 (rs6000_emit_prologue): ... here.
24629 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24631 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
24632 factored out from ...
24633 (rs6000_emit_prologue): ... here.
24635 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24637 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
24639 (handle_missing_profiles): Fix computation of tp_first_run.
24640 (counts_to_freqs): Do not touch freqs when count is 0.
24642 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24644 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
24647 2017-06-10 Tom de Vries <tom@codesourcery.com>
24649 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24650 attributes): Document signal effective target.
24652 2017-06-10 Tom de Vries <tom@codesourcery.com>
24654 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24655 Document effective target stack_size.
24657 2017-06-09 David Malcolm <dmalcolm@redhat.com>
24659 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
24660 to the edit_context if they can be auto-applied.
24662 2017-06-9 Ian Lance Taylor <iant@golang.org>
24664 * opts.c (finish_options): If -fsplit-stack, disable implicit
24665 -forder-blocks-and-partition.
24666 * doc/invoke.texi (Optimize Options): Document that when using
24667 -fsplit-stack -forder-blocks-and-partition is not implicitly
24670 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24672 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
24673 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
24674 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
24675 * builtins.def (abort, trap, unreachable): Declare cold.
24676 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
24677 * tree-core.h (ECF_COLD): New.
24678 * tree.c (set_call_expr_flags): Handle ECF_COLD.
24679 (build_common_builtin_nodes): Mark unreachable and abort as cold.
24681 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24683 * predict.c (unlikely_executed_stmt_p): Cleanup.
24685 2017-06-09 Richard Biener <rguenther@suse.de>
24687 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
24688 model if the ref is always written to.
24690 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24692 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
24694 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24696 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
24699 2017-06-09 Tom de Vries <tom@codesourcery.com>
24702 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
24703 "target cannot support label values" when encountering LABEL_REF.
24705 2017-06-09 Martin Liska <mliska@suse.cz>
24707 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
24708 (gimple_gen_ic_func_profiler): Emit direct comparison
24709 of __gcov_indirect_call_callee with NULL.
24710 (gimple_gen_time_profiler): Change probability from
24711 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
24713 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24715 * profile.c (edge_gcov_counts): Turn to pointer.
24716 (compute_branch_probabilities, compute_branch_probabilities): Update.
24717 (branch_prob): Do not clear edge_gcov_count.
24718 * profile.h (edge_gcov_counts): Turn to pointer.
24719 (edge_gcov_count): Update.
24721 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24723 * gimple.h (gimple_check_failed): Mark cold.
24725 2017-06-09 Richard Biener <rguenther@suse.de>
24727 PR tree-optimization/66623
24728 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
24729 refactor check_reduction into two parts, properly computing
24730 whether we have to check reduction validity for outer loop
24733 2017-06-09 Richard Biener <rguenther@suse.de>
24735 PR tree-optimization/79483
24736 * graphite-scop-detection.c (order): New global.
24737 (get_order): Compute bb to order mapping that satisfies code
24738 generation constraints.
24739 (cmp_pbbs): New helper.
24740 (build_scops): Start domwalk at entry block, sort generated
24743 2017-06-09 Richard Biener <rguenther@suse.de>
24745 PR middle-end/81007
24746 * ipa-polymorphic-call.c
24747 (ipa_polymorphic_call_context::restrict_to_inner_class):
24748 Skip FIELD_DECLs with error_mark_node type.
24749 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
24752 2017-06-09 Martin Liska <mliska@suse.cz>
24754 * predict.c (struct branch_predictor): New struct.
24755 (test_prediction_value_range): New test.
24756 (predict_c_tests): New function.
24757 * selftest-run-tests.c (selftest::run_tests): Run the function.
24758 * selftest.h: Declare new tests.
24760 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
24763 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
24764 gen_add3_insn did not fail.
24765 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
24766 r0, construct that number in a temporary reg and add that reg to r0.
24767 If asked to put the result in r0 as well, fail.
24769 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
24771 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24772 for early expansion of vec_eqv.
24774 2017-06-08 Jakub Jelinek <jakub@redhat.com>
24776 PR middle-end/81005
24777 * ubsan.c (instrument_null): Avoid pointless code temporary.
24778 (pass_ubsan::execute): Instrument aggregate arguments of calls.
24780 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
24784 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
24787 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
24790 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24792 * predict.c (unlikely_executed_edge_p): Move ahead.
24793 (probably_never_executed_edge_p): Use it.
24795 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24797 PR middle-end/79988
24798 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
24799 gimple_call_builtin_p call.
24801 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24803 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
24804 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
24805 rtl_check_failed_type2, rtl_check_failed_code1,
24806 rtl_check_failed_code2, rtl_check_failed_code_mode,
24807 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
24808 rtvec_check_failed_bounds, rtl_check_failed_flag,
24809 _fatal_insn_not_found, _fatal_insn): Likewise.
24810 * tree.h (tree_contains_struct_check_failed,
24811 tree_check_failed, tree_not_check_failed,
24812 tree_class_check_failed, tree_range_check_failed,
24813 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
24814 tree_vec_elt_check_failed, phi_node_elt_check_failed,
24815 tree_operand_check_failed, omp_clause_check_failed,
24816 omp_clause_operand_check_failed, omp_clause_range_check_failed):
24819 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24821 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
24822 flag_branch_probabilities.
24823 * ipa-inline.c (edge_badness): Likewise.
24824 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
24825 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
24826 * predict.c (maybe_hot_frequency_p): Likewise.
24827 (probably_never_executed): Likewise.
24828 * sched-ebb.c (schedule_ebbs): Likewise.
24829 * sched-rgn.c (find_single_block_region): Likewise.
24830 * tracer.c (tail_duplicate): Likewise.
24832 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24834 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
24835 longer requires x_flag_profile_use.
24837 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24839 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
24840 instead of flag_reorder_blocks_and_partition.
24841 * dbxout.c (dbxout_function_end): Likewise.
24842 * dwarf2out.c (gen_subprogram_die): Likewise.
24843 * haifa-sched.c (sched_create_recovery_edges): Likewise.
24844 * hw-doloop.c (reorg_loops): Likewise.
24845 * varasm.c (assemble_start_function,
24846 assemble_end_function): Likewise.
24847 (decide_function_section): Do not check for
24848 flag_reorder_blocks_and_partition.
24850 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24852 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24854 (chkp_get_hard_register_fake_addr_expr): Ditto.
24855 (chkp_build_addr_expr): Add check for hard reg case.
24856 (chkp_parse_array_and_component_ref): Ditto.
24857 (chkp_find_bounds_1): Ditto.
24858 (chkp_process_stmt): Don't generate bounds store for
24861 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24863 * predict.c (maybe_hot_bb_p): Do not check profile status.
24864 (maybe_hot_edge_p): Likewise.
24865 (probably_never_executed): Check for zero counts even if profile
24867 (unlikely_executed_edge_p): New function.
24868 (unlikely_executed_stmt_p): New function.
24869 (unlikely_executed_bb_p): New function.
24870 (set_even_probabilities): Use unlikely predicates.
24871 (combine_predictions_for_bb): Likewise.
24872 (predict_paths_for_bb): Likewise.
24873 (predict_paths_leading_to_edge): Likewise.
24874 (determine_unlikely_bbs): New function.
24875 (estimate_bb_frequencies): Use it.
24876 (compute_function_frequency): Use zero counts even if profile is
24878 * profile-count.h: Fix typo.
24880 2017-08-08 Julia Koval <julia.koval@intel.com>
24882 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
24883 _mm512_mask_cvtsepi16_storeu_epi8,
24884 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
24885 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
24886 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
24887 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
24888 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
24889 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
24890 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
24891 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
24892 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
24893 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
24894 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
24895 __builtin_ia32_pmovuswb256mem_mask,
24896 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
24897 __builtin_ia32_pmovwb512mem_mask): New builtins.
24899 2017-08-08 Julia Koval <julia.koval@intel.com>
24901 PR target/73350,80862
24902 * config/i386/subst.md (round): Fix round pattern.
24903 * config/i386/i386.c (ix86_erase_embedded_rounding):
24904 Fix erasing rounding for the fixed pattern.
24906 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24908 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
24910 2017-06-08 Martin Liska <mliska@suse.cz>
24912 PR gcov-profile/80911
24913 * gcov.c (block_info::block_info): New constructor.
24915 2017-06-07 Carl Love <cel@us.ibm.com>
24917 * config/rs6000/rs6000-c: The return type of the following
24918 built-in functions was implemented as int not long long. Fix sign
24919 of return value for the unsigned version of vec_mulo and vec_mule.
24920 vector unsigned long long vec_bperm (vector unsigned long long,
24921 vector unsigned char)
24922 vector signed long long vec_mule (vector signed int,
24924 vector unsigned long long vec_mule (vector unsigned int,
24925 vector unsigned int)
24926 vector signed long long vec_mulo (vector signed int,
24928 vector unsigned long long vec_mulo (vector unsigned int,
24929 vector unsigned int)
24930 * doc/extend.texi: Fix the documentation for the built-in
24933 2017-06-07 Carl Love <cel@us.ibm.com>
24936 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
24939 2017-06-07 Carl Love <cel@us.ibm.com>
24941 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
24942 support, Generate doublehv for signed int/float for BE case only.
24944 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
24946 * doc/invoke.texi (mcx16): Rewrite.
24948 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24950 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
24951 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
24952 *mov<mode>_softfloat, and an anonymous splitter): Use
24953 nonimmediate_operand instead of rs6000_nonimmediate_operand.
24955 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24957 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
24959 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24960 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
24961 (rs6000_debug_reg_global): Adjust.
24962 (rs6000_init_hard_regno_mode_ok): Adjust.
24963 (rs6000_dbx_register_number): Adjust.
24964 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
24965 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24966 Remove SPE_ACC and SPEFSCR.
24967 (REG_ALLOC_ORDER): Ditto.
24968 (FRAME_POINTER_REGNUM): Change to 111.
24969 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
24970 (REG_CLASS_NAMES): Ditto.
24971 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
24972 (REGISTER_NAMES): Ditto.
24973 (ADDITIONAL_REG_NAMES): Ditto.
24974 (rs6000_reg_names): Ditto.
24975 * config/rs6000/rs6000.md: Renumber some register number
24978 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24980 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
24982 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24983 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
24985 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
24986 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24987 Delete the SPE high registers.
24988 (REG_ALLOC_ORDER): Ditto.
24989 (enum reg_class): Remove SPE_HIGH_REGS.
24990 (REG_CLASS_NAMES): Ditto.
24991 (REG_CLASS_CONTENTS): Delete the SPE high registers.
24992 (REGISTER_NAMES): Ditto.
24993 (rs6000_reg_names): Ditto.
24994 * doc/tm.texi.in: Remove SPE as example.
24995 * doc/tm.texi: Regenerate.
24997 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24999 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25000 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25001 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25002 * config/rs6000/rs6000.md (type): Remove "brinc".
25004 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25006 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25007 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25008 * config/rs6000/linuxspe.h: Delete file.
25009 * config/rs6000/rs6000.md: Don't include spe.md.
25010 * config/rs6000/spe.h: Delete file.
25011 * config/rs6000/spe.md: Delete file.
25012 * config/rs6000/t-rs6000: Remove spe.md.
25014 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25016 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25017 (reg_or_none500mem_operand): Delete.
25018 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25019 instead of reg_or_none500mem_operand.
25021 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25023 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25024 handling of SPE flags.
25025 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25027 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25029 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25031 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25032 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25033 paired_divv2sf3): Similar.
25034 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25035 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25036 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25038 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25039 Rename the paired_* instruction patterns.
25040 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25042 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25043 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25044 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25045 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25046 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25047 PAIRED_VECTOR_MODE.
25048 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25049 (spe_func_has_64bit_regs_p): Delete.
25050 (spe_expand_predicate_builtin): Delete.
25051 (spe_expand_evsel_builtin): Delete.
25052 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25053 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25054 (invalid_e500_subreg): Delete.
25055 (rs6000_legitimize_address): Always force_reg op2 as well, for
25056 paired single memory accesses.
25057 (rs6000_member_type_forces_blk): Delete.
25058 (rs6000_spe_function_arg): Delete.
25059 (rs6000_expand_unop_builtin): Delete SPE handling.
25060 (rs6000_expand_binop_builtin): Ditto.
25061 (spe_expand_stv_builtin): Delete.
25062 (bdesc_2arg_spe): Delete.
25063 (spe_expand_builtin): Delete.
25064 (spe_expand_predicate_builtin): Delete.
25065 (spe_expand_evsel_builtin): Delete.
25066 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25067 (spe_init_builtins): Delete.
25068 (spe_func_has_64bit_regs_p): Delete.
25069 (savres_routine_name): Delete "info" parameter. Adjust callers.
25070 (rs6000_emit_stack_reset): Ditto.
25071 (rs6000_dwarf_register_span): Delete.
25072 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25073 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25074 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25076 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25078 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25079 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25080 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25081 mulv2sf3, divv2sf3): Delete expanders.
25083 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25085 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25087 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25089 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25090 * config/rs6000/rs6000.c: Ditto.
25092 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25094 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25095 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25096 comparison_operator.
25098 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25100 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25101 * config/rs6000/rs6000.opt: Ditto.
25102 * config/rs6000/t-rtems: Ditto.
25104 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25106 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25107 TARGET_E500_SINGLE by 0, simplify.
25108 * config/rs6000/rs6000.c: Ditto.
25109 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25110 (spe_build_register_parallel): Delete.
25111 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25112 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25113 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25114 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25115 (E500_CONVERT): Delete.
25116 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25118 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25120 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25121 * config/rs6000/dfp.md: Ditto.
25122 (negdd2, *negdd2_fpr): Merge.
25123 (absdd2, *absdd2_fpr): Merge.
25124 (negtd2, *negtd2_fpr): Merge.
25125 (abstd2, *abstd2_fpr): Merge.
25126 * config/rs6000/e500.h: Delete file.
25127 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25128 TARGET_FPRS by 1 and simplify.
25129 * config/rs6000/rs6000-c.c: Ditto.
25130 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25131 TARGET_DF_SPE by 0.
25132 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25134 * config/rs6000/rs6000.md: Ditto.
25135 (floatdidf2, *floatdidf2_fpr): Merge.
25136 (move_from_CR_gt_bit): Delete.
25137 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25138 (E500_CR_IOR_COMPARE): Delete.
25139 (All patterns that require !TARGET_FPRS): Delete.
25140 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25142 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25144 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25146 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25148 * graphds.c (add_edge): Intitialize edge's attached data.
25149 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25150 pointer parameter. Call pointed function on each edge during
25151 graph traversing. Skip traversing the edge when the function
25153 (graphds_dfs, graphds_scc): Ditto.
25154 (for_each_edge): New parameter. Pass the new parameter to callback
25156 * graphds.h (skip_edge_callback): New function pointer type.
25157 (graphds_dfs, graphds_scc): New function pointer parameter.
25158 (graphds_edge_callback, for_each_edge): New parameter.
25160 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25162 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25163 out code checking if runtime alias check is possible to below ...
25164 Call the new function.
25165 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25166 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25168 2017-06-07 Marek Polacek <polacek@redhat.com>
25171 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25172 TYPE_OVERFLOW_WRAPS check.
25174 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25176 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25177 if versioning is required.
25178 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25179 peeling with the check for versioning.
25181 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25183 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25184 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25185 Set true to new parameter if new ssa variable is defined.
25186 (vect_gen_vector_loop_niters): Refactor. Set range information
25187 for the new vector loop bound variable.
25188 (vect_do_peeling): Ditto.
25190 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25192 * tree-affine.c (ssa.h): Include header file.
25193 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25194 has wrapping overflow behavior.
25196 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25198 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25200 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25202 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25203 (tree_to_aff_combination): ... here.
25205 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25207 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25208 reg_pressure model function.
25209 (ivopts_global_cost_for_size): Delete.
25210 (determine_set_costs, iv_ca_recount_cost): Call new model function
25211 ivopts_estimate_reg_pressure.
25213 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25215 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25216 expensive than udiv. Remove floating point cases from mod.
25218 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25220 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25221 Increase idiv cost.
25223 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25225 * config/aarch64/aarch64.md
25226 (copysignsf3): Fix mask generation.
25228 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25230 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25232 (class dump_manager): Add register_dumps method.
25233 * dumpfile.c: Include langhooks.h.
25234 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25235 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25236 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25237 (dump_manager::dump_register): Start with 512 entries instead of 32.
25238 (dump_manager::register_dumps): New method.
25239 * toplev.c (general_init): Instead of invoking register_dumps
25240 langhook, invoke register_dumps method on the dump manager.
25241 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25244 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25246 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25247 Say that # requires an associated define_split to exist, and that
25248 the define_split must be suitable for use after register allocation.
25250 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25252 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25253 (compute_outgoing_frequencies): Also initialize zero counts.
25254 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25255 around loops; preserve more of profile when nothing changes.
25257 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25259 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25261 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25262 * config/arm/arm-cpu-cdata.h: Regenerate.
25263 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25264 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25265 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25266 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25267 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25269 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25271 * config/arm/t-rmprofile: Likewise.
25272 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25274 2017-06-06 David S. Miller <davem@davemloft.net>
25277 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25278 blockage if function uses alloca.
25280 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25282 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25283 New "uid" fields to hold pretty-print IDs of group and ref.
25284 Memory references are now identified as <group_id>:<ref_id>
25285 instead of using [random] addresses.
25286 (dump_mem_details): Simplify, no functional change.
25287 (dump_mem_ref): Simplify and make output more concise.
25288 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25289 (find_or_create_group): Initialize group uid.
25290 (record_ref): Initialize ref uid. Improve debug output.
25291 (prune_group_by_reuse, should_issue_prefetch_p,)
25292 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25293 (mark_nontemporal_store, determine_loop_nest_reuse):
25294 Improve debug output.
25296 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25298 * dbgcnt.def (prefetch): New debug counter.
25299 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25300 (schedule_prefetches): Stop issueing prefetches if debug counter
25303 2017-06-06 Tom de Vries <tom@codesourcery.com>
25305 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25306 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25308 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25310 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25311 Use aarch64_reg_or_zero predicate for operand 4.
25312 (aarch64_compare_and_swap<mode> define_insn_and_split):
25313 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25314 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25316 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25318 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25319 (arm_compute_save_core_reg_mask): This.
25320 (thumb1_compute_save_reg_mask): Rename into ...
25321 (thumb1_compute_save_core_reg_mask): This.
25322 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25323 (arm_compute_frame_layout): Likewise.
25325 2017-06-06 Richard Biener <rguenther@suse.de>
25327 PR tree-optimization/80974
25328 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25329 keep or clear leaders SSA info.
25331 2017-06-06 Tom de Vries <tom@codesourcery.com>
25333 * config/nvptx/nvptx.c (split_mode_p): New function.
25334 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25336 2017-06-06 Tom de Vries <tom@codesourcery.com>
25338 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25340 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25343 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25346 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25348 * shrink-wrap.c (handle_simple_exit): Update profile.
25349 (try_shrink_wrapping): Upate profile.
25351 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25353 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25354 (tree_guess_outgoing_edge_probabilities): New.
25355 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25356 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25358 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25360 * ipa-split.c (split_function): Initialize return bb profile.
25362 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25364 * profile.c (compute_branch_probabilities): Also initialize
25365 EXIT_BLOCK profile.
25367 2017-06-06 Richard Biener <rguenther@suse.de>
25369 PR tree-optimization/80928
25370 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25371 (vect_analyze_loop_operations): Properly guard analysis for
25373 (vect_transform_loop): Likewise.
25374 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25375 (vect_model_induction_cost): Do not cost for pure SLP.
25376 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25377 of induction in inner loop vectorization.
25378 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25379 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25380 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25382 (vect_analyze_slp_cost_1): Cost induction.
25383 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25384 (vect_get_slp_vect_defs): Likewise.
25385 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25386 (vect_transform_stmt): Handle SLP reductions.
25387 * tree-vectorizer.h (vectorizable_induction): Adjust.
25389 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25391 * config/rs6000/rs6000.c (make_resolver_func): Update
25392 init_lowered_empty_function call.
25394 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25396 * doc/invoke.texi: Document the -fprofile-abs-path option.
25397 * common.opt (fprofile-abs-path): New option.
25398 * gcov-io.h (gcov_write_filename): Declare.
25399 * gcov-io.c (gcov_write_filename): New function.
25400 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25401 * profile.c (output_location): Likewise.
25403 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25405 * shring-wrap.c: Revert accidental commit.
25407 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25409 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25411 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25413 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25415 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25416 profile in callgraph edge.
25417 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25418 (apply_scale): Likewise.
25419 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25420 Also scale profile when inlining function with zero profile.
25421 (initialize_cfun): Update exit block profile even when it is zero.
25422 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25423 when profile is read.
25425 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25427 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25428 (CLONE_*): New constants to define the processors we can generate
25429 code for with the target_clone attribute.
25430 (rs6000_clone_map): New array to identify which clone processors
25431 the current program is running on.
25432 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25433 target_clone attribute.
25434 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25435 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25436 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25437 (cpu_expand_builtin): Add support for target_clone attribute.
25438 (rs6000_valid_attribute_p): Allow "default" attribute.
25439 (get_decl_name): New debug function to simplify printing the
25440 current function name in debugging statements.
25441 (rs6000_clone_priority): New functions to support the target_clone
25442 attribute, and be able to generate code to switch between ISA 2.05
25443 through ISA 3.0 (power6 through power9).
25444 (rs6000_compare_version_priority): Likewise.
25445 (rs6000_get_function_versions_dispatcher): Likewise.
25446 (make_resolver_func): Likewise.
25447 (add_condition_to_bb): Likewise.
25448 (dispatch_function_versions): Likewise.
25449 (rs6000_generate_version_dispatcher_body): Likewise.
25450 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25451 (fusion_gpr_load_p): Fix a spacing issue.
25452 * doc/extend.texi (Common Function Attributes): Document that the
25453 PowerPC supports the target_clone attribute.
25455 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25457 * config/arm/arm.h: explain F symbol found in description of ARM
25458 register allocation in its legend.
25460 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25462 * config/mips/frame-header-opt.c: Include profile-count.h.
25463 * config/riscv/riscv.c: Include profile-count.h
25465 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25467 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25469 (sm_set_flag_if_changed): Add bbs field.
25470 (execute_sm_if_changed_flag_set): Pass BBS.
25471 (execute_sm): Update.
25473 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25475 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25478 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25480 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25481 (peephole2): New peephole2 to emit the above.
25482 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25484 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25486 * config/aarch64/aarch64.c (define_peephole2 above
25487 *sub_<shift>_<mode>): New peephole.
25489 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25491 * config/i386/i386.c (make_resolver_func): Update.
25492 * Makefile.in: Add profile-count.h and profile-count.o
25493 * auto-profile.c (afdo_indirect_call): Update to new API.
25494 (afdo_set_bb_count): Update.
25495 (afdo_propagate_edge): Update.
25496 (afdo_propagate_circuit): Update.
25497 (afdo_calculate_branch_prob): Update.
25498 (afdo_annotate_cfg): Update.
25499 * basic-block.h: Include profile-count.h
25500 (struct edge_def): Turn count to profile_count.
25501 (struct basic_block_def): Likewie.
25502 (REG_BR_PROB_BASE): Move to profile-count.h
25503 (RDIV): Move to profile-count.h
25504 * bb-reorder.c (max_entry_count): Turn to profile_count.
25505 (find_traces): Update.
25506 (rotate_loop):Update.
25507 (connect_traces):Update.
25508 (sanitize_hot_paths):Update.
25509 * bt-load.c (migrate_btr_defs): Update.
25510 * cfg.c (RDIV): Remove.
25511 (init_flow): Use alloc_block.
25512 (alloc_block): Uninitialize count.
25513 (unchecked_make_edge): Uninitialize count.
25514 (check_bb_profile): Update.
25515 (dump_edge_info): Update.
25516 (dump_bb_info): Update.
25517 (update_bb_profile_for_threading): Update.
25518 (scale_bbs_frequencies_int): Update.
25519 (scale_bbs_frequencies_gcov_type): Update.
25520 (scale_bbs_frequencies_profile_count): New.
25521 * cfg.h (update_bb_profile_for_threading): Update.
25522 (scale_bbs_frequencies_profile_count): Declare.
25523 * cfgbuild.c (compute_outgoing_frequencies): Update.
25524 (find_many_sub_basic_blocks): Update.
25525 * cfgcleanup.c (try_forward_edges): Update.
25526 (try_crossjump_to_edge): Update.
25527 * cfgexpand.c (expand_gimple_tailcall): Update.
25528 (construct_exit_block): Update.
25529 * cfghooks.c (verify_flow_info): Update.
25530 (dump_bb_for_graph): Update.
25531 (split_edge): Update.
25532 (make_forwarder_block): Update.
25533 (duplicate_block): Update.
25534 (account_profile_record): Update.
25535 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25536 (get_estimated_loop_iterations): Update.
25537 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25538 (single_likely_exit): Update.
25539 * cfgloopmanip.c (scale_loop_profile): Update.
25541 (set_zero_probability): Update.
25542 (lv_adjust_loop_entry_edge): Update.
25543 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25544 (purge_dead_edges): Update.
25545 (rtl_account_profile_record): Update.
25546 * cgraph.c (cgraph_node::create): Uninitialize count.
25547 (symbol_table::create_edge): Uninitialize count.
25548 (cgraph_update_edges_for_call_stmt_node): Update.
25549 (cgraph_edge::dump_edge_flags): Update.
25550 (cgraph_node::dump): Update.
25551 (cgraph_edge::maybe_hot_p): Update.
25552 * cgraph.h: Include profile-count.h
25553 (create_clone), create_edge, create_indirect_edge): Update.
25554 (cgraph_node): Turn count to profile_count.
25555 (cgraph_edge0: Likewise.
25556 (make_speculative, clone): Update.
25557 (create_edge): Update.
25558 (init_lowered_empty_function): Update.
25559 * cgraphclones.c (cgraph_edge::clone): Update.
25560 (duplicate_thunk_for_node): Update.
25561 (cgraph_node::create_clone): Update.
25562 * cgraphunit.c (cgraph_node::analyze): Update.
25563 (cgraph_node::expand_thunk): Update.
25564 * final.c (dump_basic_block_info): Update.
25565 * gimple-streamer-in.c (input_bb): Update.
25566 * gimple-streamer-out.c (output_bb): Update.
25567 * graphite.c (print_global_statistics): Update.
25568 (print_graphite_scop_statistics): Update.
25569 * hsa-brig.c: Include basic-block.h.
25570 * hsa-dump.c: Include basic-block.h.
25571 * hsa-gen.c (T sum_slice): Update.
25572 (convert_switch_statements):Update.
25573 * hsa-regalloc.c: Include basic-block.h.
25574 * ipa-chkp.c (chkp_produce_thunks): Update.
25575 * ipa-cp.c (struct caller_statistics): Update.
25576 (init_caller_stats): Update.
25577 (gather_caller_stats): Update.
25578 (ipcp_cloning_candidate_p): Update.
25579 (good_cloning_opportunity_p): Update.
25580 (get_info_about_necessary_edges): Update.
25581 (dump_profile_updates): Update.
25582 (update_profiling_info): Update.
25583 (update_specialized_profile): Update.
25584 (perhaps_add_new_callers): Update.
25585 (decide_about_value): Update.
25586 (ipa_cp_c_finalize): Update.
25587 * ipa-devirt.c (struct odr_type_warn_count): Update.
25588 (struct decl_warn_count): Update.
25589 (struct final_warning_record): Update.
25590 (possible_polymorphic_call_targets): Update.
25591 (ipa_devirt): Update.
25592 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25593 * ipa-icf.c (sem_function::merge): Update.
25594 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
25595 * ipa-inline.c (compute_uninlined_call_time): Update.
25596 (compute_inlined_call_time): Update.
25597 (want_inline_small_function_p): Update.
25598 (want_inline_self_recursive_call_p): Update.
25599 (edge_badness): Update.
25600 (lookup_recursive_calls): Update.
25601 (recursive_inlining): Update.
25602 (inline_small_functions): Update.
25603 (dump_overall_stats): Update.
25604 (dump_inline_stats): Update.
25605 * ipa-profile.c (ipa_profile_generate_summary): Update.
25606 (ipa_propagate_frequency): Update.
25607 (ipa_profile): Update.
25608 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25609 * ipa-utils.c (ipa_merge_profiles): Update.
25610 * loop-doloop.c (doloop_modify): Update.
25611 * loop-unroll.c (report_unroll): Update.
25612 (unroll_loop_runtime_iterations): Update.
25613 * lto-cgraph.c (lto_output_edge): Update.
25614 (lto_output_node): Update.
25615 (input_node): Update.
25616 (input_edge): Update.
25617 (merge_profile_summaries): Update.
25618 * lto-streamer-in.c (input_cfg): Update.
25619 * lto-streamer-out.c (output_cfg): Update.
25620 * mcf.c (create_fixup_graph): Update.
25621 (adjust_cfg_counts): Update.
25622 (sum_edge_counts): Update.
25623 * modulo-sched.c (sms_schedule): Update.
25624 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
25625 * predict.c (maybe_hot_count_p): Update.
25626 (probably_never_executed): Update.
25627 (dump_prediction): Update.
25628 (combine_predictions_for_bb): Update.
25629 (propagate_freq): Update.
25630 (handle_missing_profiles): Update.
25631 (counts_to_freqs): Update.
25632 (rebuild_frequencies): Update.
25633 (force_edge_cold): Update.
25634 * predict.h: Include profile-count.h
25635 (maybe_hot_count_p, counts_to_freqs): UPdate.
25636 * print-rtl-function.c: Do not include cfg.h
25637 * print-rtl.c: Include basic-block.h
25638 * profile-count.c: New file.
25639 * profile-count.h: New file.
25640 * profile.c (is_edge_inconsistent): Update.
25641 (correct_negative_edge_counts): Update.
25642 (is_inconsistent): Update.
25643 (set_bb_counts): Update.
25644 (read_profile_edge_counts): Update.
25645 (compute_frequency_overlap): Update.
25646 (compute_branch_probabilities): Update; Initialize and deinitialize
25648 (branch_prob): Update.
25649 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
25650 (edge_gcov_count): New.
25651 (bb_gcov_count): New.
25652 * shrink-wrap.c (try_shrink_wrapping): Update.
25653 * tracer.c (better_p): Update.
25654 * trans-mem.c (expand_transaction): Update.
25655 (ipa_tm_insert_irr_call): Update.
25656 (ipa_tm_insert_gettmclone_call): Update.
25657 * tree-call-cdce.c: Update.
25658 * tree-cfg.c (gimple_duplicate_sese_region): Update.
25659 (gimple_duplicate_sese_tail): Update.
25660 (gimple_account_profile_record): Update.
25661 (execute_fixup_cfg): Update.
25662 * tree-inline.c (copy_bb): Update.
25663 (copy_edges_for_bb): Update.
25664 (initialize_cfun): Update.
25665 (freqs_to_counts): Update.
25666 (copy_cfg_body): Update.
25667 (expand_call_inline): Update.
25668 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25669 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
25670 (try_unroll_loop_completely): Update.
25671 (try_peel_loop): Update.
25672 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25673 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
25674 * tree-ssa-loop-split.c (connect_loops): Update.
25675 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
25676 * tree-ssa-reassoc.c (branch_fixup): Update.
25677 * tree-ssa-tail-merge.c (replace_block_by): Update.
25678 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
25679 (compute_path_counts): Update.
25680 (update_profile): Update.
25681 (recompute_probabilities): Update.
25682 (update_joiner_offpath_counts): Update.
25683 (estimated_freqs_path): Update.
25684 (freqs_to_counts_path): Update.
25685 (clear_counts_path): Update.
25686 (ssa_fix_duplicate_block_edges): Update.
25687 (duplicate_thread_path): Update.
25688 * tree-switch-conversion.c (case_bit_test_cmp): Update.
25689 (struct switch_conv_info): Update.
25690 * tree-tailcall.c (decrease_profile): Update.
25691 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25692 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25693 * value-prof.c (check_counter): Update.
25694 (gimple_divmod_fixed_value): Update.
25695 (gimple_mod_pow2): Update.
25696 (gimple_mod_subtract): Update.
25697 (gimple_ic_transform): Update.
25698 (gimple_stringop_fixed_value): Update.
25699 * value-prof.h (gimple_ic): Update.
25701 2017-06-02 Carl Love <cel@us.ibm.com>
25703 * config/rs6000/rs6000-c: Add support for built-in functions
25704 vector double vec_doublee (vector signed int);
25705 vector double vec_doublee (vector unsigned int);
25706 vector double vec_doublee (vector float);
25707 vector double vec_doubleh (vector signed int);
25708 vector double vec_doubleh (vector unsigned int);
25709 vector double vec_doubleh (vector float);
25710 vector double vec_doublel (vector signed int);
25711 vector double vec_doublel (vector unsigned int);
25712 vector double vec_doublel (vector float);
25713 vector double vec_doubleo (vector signed int);
25714 vector double vec_doubleo (vector unsigned int);
25715 vector double vec_doubleo (vector float);.
25716 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
25717 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
25719 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
25720 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
25721 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
25723 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
25724 vec_doublel, vec_doubleh.
25725 * doc/extend.texi: Update the built-in documentation file for the
25726 new built-in functions.
25728 2017-06-02 David Malcolm <dmalcolm@redhat.com>
25731 * ipa-inline-analysis.c (free_growth_caches): Set
25732 edge_removal_hook_holder to NULL after removing it.
25734 2017-06-02 Sudakshina Das <sudi.das@arm.com>
25736 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
25737 comparision with zero.
25739 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25740 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25741 for early expansion of vec_min and vec_max builtins.
25742 (builtin_function_type): Add min/max unsigned variants to those
25743 identified as having unsigned arguments.
25745 2017-06-02 Olivier Hainque <hainque@adacore.com>
25747 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
25749 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25751 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
25752 Use VALL_F16 iterator rather than VALL.
25754 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25756 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
25757 Emit CBNZ inside loop when doing a strong exchange and comparing
25758 against zero. Generate the CC flags after the loop.
25760 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
25762 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
25763 (dl_section_ref): New.
25764 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
25765 On AIX, append an expression to subtract the size of the
25766 section length to dl_section_ref.
25768 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25770 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25771 for early expansion of vector absolute builtins.
25773 2017-06-02 Richard Biener <rguenther@suse.de>
25775 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
25776 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
25778 2017-06-02 Richard Biener <rguenther@suse.de>
25780 PR tree-optimization/80948
25781 * tree-tailcall.c (find_tail_calls): Track stmts to move in
25782 stmt order as well.
25784 2017-06-02 Richard Biener <rguenther@suse.de>
25786 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
25788 * tree-vect-stmts.c (process_use): Do not mark backedge defs
25789 for inductions as relevant.
25791 2017-06-02 Richard Biener <rguenther@suse.de>
25793 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
25794 (vectorizable_induction): ... this. Remove dead code.
25796 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
25798 * builtins. (expand_builtin_alloca): Remove second parameter and
25799 infer its value from the first parameter instead.
25800 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
25802 2017-06-02 Jakub Jelinek <jakub@redhat.com>
25804 PR rtl-optimization/80903
25805 * loop-doloop.c (add_test): Unshare sequence.
25807 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
25809 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
25811 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
25813 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
25815 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
25816 xlogue_layout::get_instance, logue_layout::xlogue_layout,
25817 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
25818 (xlogue_layout::get_stub_rtx): Make static.
25819 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
25820 (xlogue_layout::compute_stub_managed_regs): Rename to...
25821 (xlogue_layout::count_stub_managed_regs): ...this.
25822 (xlogue_layout::is_stub_managed_reg): New function.
25823 (xlogue_layout::m_stub_names): Rename to...
25824 (xlogue_layout::s_stub_names): ...this, make static.
25825 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
25826 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
25827 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
25828 xlogue_layout::s_stub_names): Instantiate statics.
25829 (stub_managed_regs): Remove.
25830 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
25831 (disable_call_ms2sysv_xlogues): Rename to...
25832 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
25833 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
25835 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
25836 change after reload_completed.
25837 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
25839 (ix86_expand_prologue): Likewise.
25840 (ix86_expand_epilogue): Likewise.
25841 (ix86_expand_split_stack_prologue): Likewise.
25842 (ix86_compute_frame_layout): Remove frame parameter ...
25843 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
25844 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
25846 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
25847 (ix86_frame): Move from here ...
25848 * config/i386/i386.h (ix86_frame): ... to here.
25849 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
25850 complete ix86_frame data structure instead. Remove some_ld_name.
25852 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25854 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
25855 symbols that hold a DECL_VALUE_EXPR.
25857 2017-06-01 Martin Jambor <mjambor@suse.cz>
25859 PR tree-optimization/80898
25860 * tree-sra.c (process_subtree_disqualification): Removed.
25861 (disqualify_candidate): Do not acll
25862 process_subtree_disqualification.
25863 (subtree_mark_written_and_enqueue): New function.
25864 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
25865 RHS has been disqualified and re-queue LHS if necessary. Apart
25866 from that, ignore disqualified RHS.
25868 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25870 * config/s390/s390.c (s390_emit_epilogue): Disable early return
25871 address fetch for z10 or later.
25873 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25875 * config/arc/arc.md (tst_movb): Add guard when splitting.
25877 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25879 * config/arc/arc.c (arc_can_eliminate): Test against
25880 arc_frame_pointer_needed.
25882 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25884 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
25885 to prevent store reordering.
25886 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
25887 (type): Add block type.
25888 (stack_tie): Define special instruction to be used in
25891 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25893 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
25894 constraint. It is not valid for the pattern.
25895 (noncommutative_binary_comparison): Likewise.
25897 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25899 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
25902 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25904 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
25905 be used by the reg-alloc.
25907 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25909 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
25910 reg-alloc when having mul64 or mul32x16 instructions.
25911 (mulsidi3): Likewise.
25912 (umulsidi3): Likewise.
25913 (mulsi32x16): New pattern.
25914 (mulsi64): Likewise.
25915 (mulsidi64): Likewise.
25916 (umulsidi64): Likewise.
25917 (MUL32x16_REG): Define.
25918 (mul64_600): Use MUL32x16_REG.
25919 (mac64_600): Likewise.
25920 (umul64_600): Likewise.
25921 (umac64_600): Likewise.
25923 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25925 * config/arc/arc.md (mulsi3_700): Make it commutative.
25927 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
25929 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
25931 (*sign_extendsidi2_insn): Likewise for movstosw.
25933 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25935 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
25936 the type of the input discriminant value. Convert the
25937 discriminant value of signedness vary.
25939 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
25941 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
25942 Add to -Wall section.
25944 2017-06-01 Richard Biener <rguenther@suse.de>
25946 PR middle-end/66313
25947 * fold-const.c (fold_plusminus_mult_expr): If the factored
25948 factor may be zero use a wrapping type for the inner operation.
25949 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
25950 and handle moved defs.
25951 (process_assignment): Properly guard the unary op case. Return a
25952 tri-state indicating that moving the stmt before the call may allow
25953 to continue. Pass through to_move.
25954 (find_tail_calls): Handle moving unrelated defs before
25957 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
25960 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
25961 splitter result in the canonical way.
25963 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
25965 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
25966 also for 32bit target. Update insn attributes.
25967 (zero-extendsidi2 splitter): Allow all registers for operand 1.
25969 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
25971 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
25972 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
25973 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
25974 (_mm_maskz_min_ss): New intrinsics.
25976 2017-05-31 Martin Liska <mliska@suse.cz>
25978 * tree-vect-loop.c (vect_create_epilog_for_reduction):
25979 Change comment style to one we normally use.
25980 (vectorizable_reduction): Likewise.
25981 (vectorizable_induction): Likewise.
25982 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
25983 (vectorizable_call): Likewise.
25984 (vectorizable_simd_clone_call): Likewise.
25985 (vectorizable_conversion): Likewise.
25986 (vectorizable_assignment): Likewise.
25987 (vectorizable_shift): Likewise.
25988 (vectorizable_operation): Likewise.
25989 (vectorizable_store): Likewise.
25990 (vectorizable_load): Likewise.
25991 * tree-vectorizer.h: Likewise.
25993 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
25995 * passes.c (emergency_dump_function): New.
25996 * tree-pass.h (emergency_dump_function): Declare.
25997 * plugin.c (plugins_internal_error_function): Remove.
25998 * plugin.h (plugins_internal_error_function): Remove declaration.
25999 * toplev.c (internal_error_function): New static function. Use it...
26000 (general_init): ...here.
26002 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26004 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26005 (arc_rtx_costs): Add costs for new patterns.
26006 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26007 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26009 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26011 * tree-ssa-strlen.c (get_next_strinfo): New function.
26012 (get_stridx_plus_constant): Use it.
26013 (zero_length_string): Likewise.
26014 (adjust_related_strinfos): Likewise.
26015 (adjust_last_stmt): Likewise.
26017 2017-05-31 Richard Biener <rguenther@suse.de>
26020 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26021 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26023 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26025 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26026 loop_vinfo argument and use of dependence distance vectors.
26027 Check instead whether the two references differ only in their
26028 initial value and assume that they have the same alignment if the
26029 difference is a multiple of the vector alignment.
26030 (vect_analyze_data_refs_alignment): Update call accordingly.
26032 2017-05-31 Martin Liska <mliska@suse.cz>
26035 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26037 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26039 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26040 (create_intersect_range_checks): Move from ...
26041 * tree-data-ref.c (create_intersect_range_checks_index)
26042 (create_intersect_range_checks): ... to here.
26043 (create_runtime_alias_checks): New function factored from ...
26044 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26045 here. Call above function.
26046 * tree-data-ref.h (create_runtime_alias_checks): New function.
26048 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26050 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26051 segment length for dr_b and compute it in wide_int.
26053 2017-05-31 Richard Biener <rguenther@suse.de>
26055 PR tree-optimization/80906
26056 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26057 and pass through iv_map.
26058 (copy_bb_and_scalar_dependences): Adjust.
26059 (translate_pending_phi_nodes): Likewise.
26060 (copy_loop_close_phi_args): Handle code-generating IVs instead
26063 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26065 * diagnostic-color.c (color_dict): Add "type-diff".
26066 (parse_gcc_colors): Update comment.
26067 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26068 -fdiagnostics-show-template-tree and -fno-elide-type.
26069 (GCC_COLORS): Add type-diff to example.
26071 (-fdiagnostics-show-template-tree): New.
26072 (-fno-elide-type): New.
26073 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26074 the pp_format_decoder callback. Call any m_format_postprocessor's
26076 (pretty_printer::pretty_printer): Initialize
26077 m_format_postprocessor.
26078 (pretty_printer::~pretty_printer): Delete any
26079 m_format_postprocessor.
26080 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26081 (class format_postprocessor): New class.
26082 (struct pretty_printer::format_decoder): Document the new parameters.
26083 (struct pretty_printer::m_format_postprocessor): New field.
26084 * tree-diagnostic.c (default_tree_printer): Update for new
26085 bool and const char ** params.
26086 * tree-diagnostic.h (default_tree_printer): Likewise.
26088 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26090 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26091 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26092 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26093 rs6000_gen_cell_microcode code.
26094 (rs6000_final_prescan_insn): Delete.
26095 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26096 "warn-cell-microcode" entries.
26097 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26098 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26099 throughout. Change cc_reg_not_micro_cr0_operand to
26100 cc_reg_not_cr0_operand throughout.
26101 (*extendhi<mode>2_noload): Delete.
26102 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26103 (mwarn-cell-microcode): Delete.
26104 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26105 -mgen-cell-microcode and -mwarn-cell-microcode.
26107 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26110 * config/i386/constraints.md (Yd): New constraint.
26112 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26113 and (?Yd, r) alternatives. Update insn attributes.
26114 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26115 and (?*Yd, r) alternatives. Update insn attributes.
26116 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26118 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26120 * gimplify.c (gimplify_modify_expr): Don't create a
26121 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26124 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26126 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26128 2017-05-30 Richard Biener <rguenther@suse.de>
26130 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26131 and reduc_def fields.
26132 (STMT_VINFO_REDUC_TYPE): New define.
26133 (STMT_VINFO_REDUC_DEF): Likewise.
26134 (vect_force_simple_reduction): Adjust prototype.
26135 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26136 (vect_is_simple_reduction): Remove check_reduction argument.
26137 (vect_force_simple_reduction): Adjust and set
26138 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26139 (vectorizable_reduction): Do not re-do reduction analysis
26140 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26141 * tree-parloops.c (gather_scalar_reductions): Adjust.
26143 2017-05-30 Richard Biener <rguenther@suse.de>
26145 PR middle-end/80901
26146 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26149 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26151 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26152 Introduce unknown_misalignment parameter and remove vf.
26153 (vect_peeling_hash_get_lowest_cost):
26154 Pass unknown_misalignment parameter.
26155 (vect_enhance_data_refs_alignment):
26156 Fix unsupportable data ref treatment.
26158 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26160 * tree-vect-data-refs.c (vect_get_data_access_cost):
26161 Workaround for SLP handling.
26162 (vect_enhance_data_refs_alignment):
26163 Compute costs for doing no peeling at all, compare to the best
26164 peeling costs so far and avoid peeling if cheaper.
26166 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26168 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26169 Return peeling info and set costs to zero for unlimited cost
26171 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26172 with unknown misalignment. Compute and costs for unknown
26173 misalignment, compare them to the costs for known misalignment
26174 and choose the cheapest for peeling.
26176 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26178 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26179 (vect_get_peeling_costs_all_drs): Create function.
26180 (vect_peeling_hash_get_lowest_cost):
26181 Use vect_get_peeling_costs_all_drs.
26182 (vect_peeling_supportable): Create function.
26183 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26185 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26187 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26188 DR_HAS_NEGATIVE_STEP.
26189 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26190 (vect_enhance_data_refs_alignment): Use.
26191 (vect_duplicate_ssa_name_ptr_info): Use.
26192 * tree-vectorizer.h (dr_misalignment): Use.
26193 (known_alignment_for_access_p): Use.
26195 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26198 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26200 (has_section_name): New function.
26202 2017-05-30 Martin Liska <mliska@suse.cz>
26205 * auto-profile.c (get_function_decl_from_block): Fix
26208 2017-05-30 Richard Biener <rguenther@suse.de>
26210 PR middle-end/80876
26211 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26213 2017-05-30 Martin Liska <mliska@suse.cz>
26215 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26216 * dumpfile.h (struct dump_file_info): Remove ctors.
26218 2017-05-30 Martin Liska <mliska@suse.cz>
26220 * predict.def: Fix GNU coding style.
26222 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26224 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26225 Mark 'to' argument with ATTRIBUTE_UNUSED.
26227 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26229 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26230 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26231 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26234 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26236 * doc/install.texi (Options specification): Restore entry of
26237 --enable-sjlj-exceptions.
26239 2017-05-27 Michael Eager <eager@eagercon.com>
26242 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26244 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26246 * config/microblaze/microblaze.h
26247 (FIXED_REGISTERS): Update in macro.
26248 (CALL_USED_REGISTERS): Update in macro.
26250 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26252 * doc/install.texi: Add links to macOS binary distributions.
26254 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26258 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26260 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26262 2017-05-26 Martin Liska <mliska@suse.cz>
26264 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26266 2017-05-26 Martin Liska <mliska@suse.cz>
26268 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26269 always leading ';; '.
26270 (dump_bb_info): Likewise.
26271 (brief_dump_cfg): Likewise.
26272 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26273 * dumpfile.c: Remove usage of TDF_VERBOSE.
26274 * dumpfile.h (enum dump_kind): Likewise.
26275 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26276 * print-tree.c (debug_verbose): Remove.
26277 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26278 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26279 * tree-diagnostic.c (default_tree_printer): Replace
26280 TDF_DIAGNOSTIC with TDF_SLIM.
26282 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26284 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26285 in parameter loop, rather than loop_vinfo.
26286 (create_intersect_range_checks): Ditto.
26287 (vect_create_cond_for_alias_checks): Update call to above functions.
26289 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26291 PR tree-optimization/80815
26292 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26293 for merging runtime alias checks. Handle negative DR_STEPs.
26295 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26297 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26299 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26300 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26301 out code pruning runtime alias checks.
26302 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26303 factored out from above.
26304 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26306 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26308 (prune_runtime_alias_test_list): New decalaration.
26310 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26312 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26313 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26314 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26315 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26316 (operator==, comp_dr_with_seg_len_pair): Ditto.
26317 (vect_prune_runtime_alias_test_list): Ditto.
26319 2017-05-26 Martin Liska <mliska@suse.cz>
26322 * params.def: Bound partial-inlining-entry-probability param.
26324 2017-05-26 Marek Polacek <polacek@redhat.com>
26327 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26330 2017-05-26 Richard Biener <rguenther@suse.de>
26332 PR tree-optimization/80842
26333 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26336 2017-05-26 Richard Biener <rguenther@suse.de>
26338 PR tree-optimization/80844
26339 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26341 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26343 * doc/md.texi (Machine Constraints): Update x86 family
26344 machine constraints section to match 'config/i386/constraints.md'.
26346 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26348 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26350 2017-05-25 Nathan Sidwell <nathan@acm.org>
26352 * doc/invoke.texi (--enable-languages): Update documentation.
26354 2017-05-25 Martin Liska <mliska@suse.cz>
26356 * dumpfile.c: Add TDF_FOLDING.
26357 * dumpfile.h (enum dump_kind): Likewise.
26358 * genmatch.c (dt_simplify::gen_1): Use it.
26360 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26362 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26364 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26366 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26367 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26369 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26371 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26372 * match.pd (X == C): Rewrite it here.
26373 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26374 with_certain_nonzero_bits2): New predicates.
26375 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26377 2017-05-24 Nathan Sidwell <nathan@acm.org>
26379 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26382 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26385 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26387 * config/powerpcspe: New port. Files are copied from the rs6000
26388 port, with "rs6000" in filenames replaced by "powerpcspe".
26390 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26392 PR rtl-optimization/80754
26393 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26395 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26397 * config/sparc/sparc.md (length): Return the correct value for -mflat
26398 sibcalls to match output_sibcall.
26400 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26404 * config/rs6000/rs6000.c (struct machine_function): Add new field
26406 (rs6000_get_separate_components): Init that field, use it.
26407 (rs6000_components_for_bb): Use the field.
26409 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26411 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26413 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26415 PR middle-end/80823
26416 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26418 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26421 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26422 address against address_operand predicate.
26423 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26425 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26427 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26428 parameters passed indirectly.
26430 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26432 * config/i386/i386.md (*movdi_internal): Remove SSE4
26433 alternative 18 (?r, *v). Update insn attributes.
26434 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26435 Update insn attributes.
26436 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26437 Update insn attributes.
26438 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26439 alternative 1 (r, v). Remove isa attribute.
26440 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26441 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26442 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26444 2017-05-23 Tom de Vries <tom@codesourcery.com>
26446 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26449 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26451 * cgraphunit.c (symbol_table::process_new_functions): Update.
26452 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26453 (inline_generate_summary): Rename to ...
26454 (ipa_fn_summary_generate): ... this one.
26455 (inline_read_summary): Rename to ...
26456 (ipa_fn_summary_read): ... this one.
26457 (inline_write_summary): Rename to ...
26458 (ipa_fn_summary_write): ... this one.
26459 (inline_free_summary): Rename to ...
26460 (ipa_free_fn_summary): ... this one.
26461 (pass_data_local_fn_summary, pass_local_fn_summary,
26462 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26463 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26464 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26465 make_pass_ipa_fn_summary): New.
26466 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26467 inline_write_summary, inline_free_summary): Remove.
26468 (ipa_free_fn_summary) : New.
26469 * ipa-inline.c (ipa_inline): Update.
26470 (pass_ipa_inline): Do not generate summaries.
26471 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26473 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26474 and add pass_ipa_fn_summary.
26475 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26477 (make_pass_inline_parameters): Remove.
26479 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26481 * omp-low.c (struct omp_context): Remove "default_kind" member.
26484 * omp-offload.c (execute_oacc_device_lower): Remove the
26485 parallelism dimensions function attributes for unparallelized
26486 OpenACC kernels constructs.
26488 2017-05-23 Martin Liska <mliska@suse.cz>
26490 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26492 (cgraph_edge::make_speculative): Likewise.
26493 (cgraph_edge::resolve_speculation): Likewise.
26494 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26495 (cgraph_node::dump): Likewise.
26496 * cgraph.h: Likewise.
26497 * cgraphunit.c (analyze_functions): Likewise.
26498 (symbol_table::compile): Likewise.
26499 * ipa-cp.c (print_all_lattices): Likewise.
26500 (determine_versionability): Likewise.
26501 (initialize_node_lattices): Likewise.
26502 (ipcp_verify_propagated_values): Likewise.
26503 (estimate_local_effects): Likewise.
26504 (update_profiling_info): Likewise.
26505 (create_specialized_node): Likewise.
26506 (perhaps_add_new_callers): Likewise.
26507 (decide_about_value): Likewise.
26508 (decide_whether_version_node): Likewise.
26509 (identify_dead_nodes): Likewise.
26510 (ipcp_store_bits_results): Likewise.
26511 * ipa-devirt.c (dump_targets): Likewise.
26512 (ipa_devirt): Likewise.
26513 * ipa-icf.c (sem_item::dump): Likewise.
26514 (sem_function::equals): Likewise.
26515 (sem_variable::equals): Likewise.
26516 (sem_item_optimizer::read_section): Likewise.
26517 (sem_item_optimizer::execute): Likewise.
26518 (congruence_class::dump): Likewise.
26519 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26520 (dump_inline_summary): Likewise.
26521 (estimate_node_size_and_time): Likewise.
26522 (inline_analyze_function): Likewise.
26523 * ipa-inline-transform.c (inline_call): Likewise.
26524 * ipa-inline.c (report_inline_failed_reason): Likewise.
26525 (want_early_inline_function_p): Likewise.
26526 (edge_badness): Likewise.
26527 (update_edge_key): Likewise.
26528 (inline_small_functions): Likewise.
26529 * ipa-profile.c (ipa_profile): Likewise.
26530 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26531 (ipa_make_edge_direct_to_target): Likewise.
26532 (remove_described_reference): Likewise.
26533 (ipa_impossible_devirt_target): Likewise.
26534 (propagate_controlled_uses): Likewise.
26535 (ipa_print_node_params): Likewise.
26536 (ipcp_transform_function): Likewise.
26537 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26538 (propagate_pure_const): Likewise.
26539 * ipa-reference.c (generate_summary): Likewise.
26540 (read_write_all_from_decl): Likewise.
26541 (propagate): Likewise.
26542 (ipa_reference_read_optimization_summary): Likewise.
26543 * ipa-utils.c (ipa_merge_profiles): Likewise.
26544 * ipa.c (walk_polymorphic_call_targets): Likewise.
26545 (symbol_table::remove_unreachable_nodes): Likewise.
26546 (ipa_single_use): Likewise.
26547 * passes.c (execute_todo): Likewise.
26548 * predict.c (drop_profile): Likewise.
26549 * symtab.c (symtab_node::get_dump_name): New function.
26550 (symtab_node::dump_name): Likewise.
26551 (symtab_node::dump_asm_name): Likewise.
26552 (symtab_node::dump_references): Likewise.
26553 (symtab_node::dump_referring): Likewise.
26554 (symtab_node::dump_base): Likewise.
26555 (symtab_node::debug_symtab): Likewise.
26556 * tree-sra.c (convert_callers_for_node): Likewise.
26557 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26558 * value-prof.c (init_node_map): Likewise.
26560 2017-05-23 Martin Liska <mliska@suse.cz>
26562 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26563 and symtab_node::debug_symtab to symbol_table::debug.
26564 * cgraphunit.c (analyze_functions): Use the renamed function.
26565 (symbol_table::compile): Likewise.
26566 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26567 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26568 * passes.c (execute_todo): Likewise.
26569 * symtab.c (symbol_table::dump): New function.
26570 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26572 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26574 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26575 that nonconst implies exec.
26577 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26579 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26580 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26581 (inline_edge_summary_vec): Turn into ...
26582 (ipa_call_summaries): ... this one.
26583 (redirect_to_unreachable, edge_set_predicate,
26584 evaluate_properties_for_edge, inline_summary_alloc,
26585 reset_ipa_call_summary, reset_inline_summary,
26586 inline_summary_t::duplicate): Update.
26587 (inline_edge_duplication_hook): Turn to ...
26588 (ipa_call_summary_t::duplicate): ... this one.
26589 (inline_edge_removal_hook): Turn to ...
26590 (ipa_call_summary_t::remove): ... this one.
26591 (dump_inline_edge_summary): Turn to ...
26592 (dump_ipa_call_summary): ... this one.
26593 (estimate_function_body_sizes): Update.
26594 (inline_update_callee_summaries): Update.
26595 (remap_edge_change_prob): Update.
26596 (remap_edge_summaries): Update.
26597 (inline_merge_summary): Update.
26598 (do_estimate_edge_time): Update.
26599 (inline_generate_summary): Update.
26600 (inline_read_section): Update.
26601 (inline_read_summary): Update.
26602 (inline_free_summary): Update.
26603 * ipa-inline.c (can_inline_edge_p): Update.
26604 (compute_inlined_call_time): Update.
26605 (want_inline_small_function_p): Update.
26606 (edge_badness): Update.
26607 (early_inliner): Update.
26608 * ipa-inline.h (inline_edge_summary): Turn to ...
26609 (ipa_call_summary): ... this one.
26610 (ipa_call_summary_t): New class.
26611 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26612 (ipa_call_summaries): New.
26613 (inline_edge_summary): Remove.
26614 (estimate_edge_growth): Update.
26615 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26616 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26617 * ipa-split.c (execute_split_functions): Update.
26618 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26620 2017-05-23 Tom de Vries <tom@codesourcery.com>
26622 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26623 attributes): Document rdrand effective target.
26625 2017-05-23 Tom de Vries <tom@codesourcery.com>
26627 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26628 attributes): Sort alphabetically.
26630 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
26632 * config/avr/genmultilib.awk: Use gsub instead of gensub.
26634 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26637 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
26638 V2DF/V2DI splat into two separate patterns, one that handles
26639 registers, and the other that only handles memory. Drop support
26640 for splatting from a GPR on ISA 2.07 and then splitting the
26641 splat into direct move and splat.
26642 (vsx_splat_<mode>_reg): Likewise.
26643 (vsx_splat_<mode>_mem): Likewise.
26645 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
26647 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
26649 2017-05-22 Jakub Jelinek <jakub@redhat.com>
26651 PR middle-end/80809
26652 * omp-low.c (finish_taskreg_remap): New function.
26653 (finish_taskreg_scan): If unit size of ctx->record_type
26654 is non-constant, unshare the size expression and replace
26655 decls in it with possible outer var refs.
26657 PR middle-end/80809
26658 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
26659 GOVD_SHARED rather than GOVD_PRIVATE with it.
26660 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
26661 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
26663 PR middle-end/80853
26664 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
26665 as last argument to build_outer_var_ref for pointer bases of array
26666 section reductions.
26668 2017-05-19 Martin Sebor <msebor@redhat.com>
26670 * print-tree.c (print_node): Print DECL_READ_P flag.
26672 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26674 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
26675 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
26676 * cgraph.c: Likewise.
26677 * cgraphunit.c: Likewise.
26678 * gengtype.c: Likewise.
26679 * ipa-cp.c: Likewise.
26680 * ipa-devirt.c: Likewise.
26681 * ipa-icf.c: Likewise.
26682 * ipa-predicate.c: Likewise.
26683 * ipa-profile.c: Likewise.
26684 * ipa-prop.c: Likewise.
26685 * ipa-split.c: Likewise.
26687 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
26688 edge_predicate_pool, dump_inline_hints,
26689 inline_summary::account_size_time, redirect_to_unreachable,
26690 edge_set_predicate, set_hint_predicate,
26691 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
26692 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
26693 inline_summary_t::remove, remap_hint_predicate_after_duplication,
26694 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
26695 ipa_call_summary_t::remove, initialize_growth_caches,
26696 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
26697 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
26698 mark_modified, unmodified_parm_1, unmodified_parm,
26699 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
26700 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
26701 compute_bb_predicates, will_be_nonconstant_expr_predicate,
26702 will_be_nonconstant_predicate, record_modified_bb_info,
26703 get_minimal_bb, record_modified, param_change_prob,
26704 phi_result_unknown_predicate, predicate_for_phi_result,
26705 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
26706 estimate_function_body_sizes, compute_inline_parameters,
26707 compute_inline_parameters_for_curren, pass_data_inline_parameters,
26708 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
26709 inline_update_callee_summaries, remap_edge_change_prob,
26710 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
26711 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
26712 inline_analyze_function, inline_summary_t::insert,
26713 inline_generate_summary, read_ipa_call_summary, inline_read_section,
26714 inline_read_summary, write_ipa_call_summary, inline_write_summary,
26715 inline_free_summary): Move to ipa-fnsummary.h
26716 (predicate_t): Remove.
26717 * ipa-fnsummary.c: New file.
26718 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
26719 (enum inline_hints_vals, inline_hints, agg_position_info,
26720 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
26721 inline_summaries, ipa_call_summary, ipa_call_summary_t,
26722 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
26723 dump_inline_summary, dump_inline_hints, inline_generate_summary,
26724 inline_read_summary, inline_write_summary, inline_free_summary,
26725 inline_analyze_function, initialize_inline_failed,
26726 inline_merge_summary, inline_update_overall_summary,
26727 compute_inline_parameters): Move to ipa-fnsummary.h
26728 * ipa-fnsummary.h: New file.
26729 * ipa-inline-transform.h: Include ipa-inline.h.
26730 * ipa-inline.c: LIkewise.
26732 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26734 * ipa-inline.c (edge_badness): Use inlined_time instead of
26735 inline_summaries->get.
26737 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26739 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
26741 2017-05-22 Nathan Sidwell <nathan@acm.org>
26743 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
26744 (fdump-lang): Document 'raw' option.
26745 * dumpfile.h (TDI_tu): Delete.
26746 * dumpfile.c (dump_files): Remove translation-unit.
26747 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
26749 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
26751 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
26752 command option from $(AWK) call.
26753 * config/avr/genmultilib.awk: Simplify and rewrite so that it
26754 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
26755 [FORMAT]: Remove handling of variable.
26756 * config/avr/t-multilib: Regenerate.
26758 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26760 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
26762 (dump_inline_summary): Do not print self_time.
26763 (estimate_function_body_sizes): Do not set self_time.
26764 (compute_inline_parameters): Likewise.
26765 (inline_read_section, inline_write_summary): Do not stream self_time.
26766 * ipa-inline.h (inline_summary): Drop self_time.
26768 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26770 * ipa-inline-analysis.c (account_size_time): Rename to ...
26771 (inline_summary::account_size_time): ... this one.
26772 (reset_ipa_call_summary): Turn to ...
26773 (ipa_call_summary::reset): ... this one.
26774 (reset_inline_summary): Turn to ...
26775 (inline_summary::reset): ... this one.
26776 (inline_summary_t::remove): Update.
26777 (inline_summary_t::duplicate): Update.
26778 (ipa_call_summary_t::remove): Update.
26779 (dump_inline_summary): Update.
26780 (estimate_function_body_sizes): Update.
26781 (compute_inline_parameters): Update.
26782 (estimate_node_size_and_time): Update.
26783 (inline_merge_summary): Update.
26784 (inline_update_overall_summary): Update.
26785 (inline_read_section): Update.
26786 (inline_write_summary): Update.
26787 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
26788 add account_size_time and reset member functions.
26789 (ipa_call_summary): Add reset function.
26790 * ipa-predicate.h (predicate::operator &): Constify.
26792 2017-05-22 Richard Biener <rguenther@suse.de>
26794 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
26796 2017-05-19 Jason Merrill <jason@redhat.com>
26798 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
26800 2017-05-19 Marek Polacek <polacek@redhat.com>
26803 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
26804 TYPE_OVERFLOW_WRAPS checks.
26806 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
26808 * tree-core.h (enum omp_clause_default_kind): Add
26809 "OMP_CLAUSE_DEFAULT_PRESENT".
26810 * tree-pretty-print.c (dump_omp_clause): Handle it.
26811 * gimplify.c (enum gimplify_omp_var_data): Add
26812 "GOVD_MAP_FORCE_PRESENT".
26813 (gimplify_adjust_omp_clauses_1): Map it to
26814 "GOMP_MAP_FORCE_PRESENT".
26815 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
26817 * gimplify.c (oacc_default_clause): Clarify.
26819 2017-05-19 Nathan Sidwell <nathan@acm.org>
26821 LANG_HOOK_REGISTER_DUMPS
26822 * toplev.c (general_init): Call register dump lang hook.
26823 * doc/invoke.texi: Document -fdump-lang option family.
26824 * dumpfile.c (dump_files): Remove class dump here.
26825 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
26826 * dumpfile.h (tree_dump_index): Remove TDI_class.
26827 * langhooks-def.h (lhd_register_dumps): Declare.
26828 (LANG_HOOKS_REGISTER_DUMPS): Define.
26829 (LANG_HOOKS_INITIALIZER): Add it.
26830 * langhooks.c (lhd_register_dumps): Define.
26831 * langhooks.h (struct lang_hooks): Add register_dumps.
26833 2017-05-19 Nathan Sidwell <nathan@acm.org>
26835 * context.h (context::set_passes): New.
26836 * context.c (context::context): Do not create pass manager.
26837 * toplev.c (general_init): Create pass manager here.
26839 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
26841 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
26842 use this splitter if two add or or instructions would also work for
26843 the constant we want to generate.
26845 2017-05-19 Richard Biener <rguenther@suse.de>
26848 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
26849 predicate evaluation.
26851 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26853 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
26855 * ipa-inline.c (want_inline_small_function_p): Do not cast to
26858 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26860 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26861 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26862 (inline_edge_summary_vec): Turn into ...
26863 (ipa_call_summaries): ... this one.
26864 (redirect_to_unreachable, edge_set_predicate,
26865 evaluate_properties_for_edge, inline_summary_alloc,
26866 reset_ipa_call_summary, reset_inline_summary,
26867 inline_summary_t::duplicate): Update.
26868 (inline_edge_duplication_hook): Turn to ...
26869 (ipa_call_summary_t::duplicate): ... this one.
26870 (inline_edge_removal_hook): Turn to ...
26871 (ipa_call_summary_t::remove): ... this one.
26872 (dump_inline_edge_summary): Turn to ...
26873 (dump_ipa_call_summary): ... this one.
26874 (estimate_function_body_sizes): Update.
26875 (inline_update_callee_summaries): Update.
26876 (remap_edge_change_prob): Update.
26877 (remap_edge_summaries): Update.
26878 (inline_merge_summary): Update.
26879 (do_estimate_edge_time): Update.
26880 (inline_generate_summary): Update.
26881 (inline_read_section): Update.
26882 (inline_read_summary): Update.
26883 (inline_free_summary): Update.
26884 * ipa-inline.c (can_inline_edge_p): Update.
26885 (compute_inlined_call_time): Update.
26886 (want_inline_small_function_p): Update.
26887 (edge_badness): Update.
26888 (early_inliner): Update.
26889 * ipa-inline.h (inline_edge_summary): Turn to ...
26890 (ipa_call_summary): ... this one.
26891 (ipa_call_summary_t): New class.
26892 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26893 (ipa_call_summaries): New.
26894 (inline_edge_summary): Remove.
26895 (estimate_edge_growth): Update.
26896 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26897 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26898 * ipa-split.c (execute_split_functions): Update.
26899 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26901 2017-05-19 Richard Biener <rguenther@suse.de>
26903 PR middle-end/80764
26904 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
26906 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
26908 * config/rs6000/rs6000.c (struct machine_function): Add field
26909 fpr_is_wrapped_separately.
26910 (rs6000_get_separate_components): Use 64 components. Handle the
26911 new FPR components.
26912 (rs6000_components_for_bb): Handle the FPR components.
26913 (rs6000_emit_prologue_components): Handle the FPR components.
26914 (rs6000_emit_epilogue_components): Handle the FPR components.
26915 (rs6000_set_handled_components): Handle the FPR components.
26916 (rs6000_emit_prologue): Don't output prologue code for those FPRs
26917 that are already separately shrink-wrapped.
26918 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
26919 that are already separately shrink-wrapped.
26921 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26924 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
26927 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
26928 (define_peephole2 for Altivec d-form load): Add peepholes to catch
26929 cases where the register allocator uses a move and an offsettable
26930 memory operation to/from a FPR register on ISA 2.06/2.07.
26931 (define_peephole2 for Altivec d-form store): Likewise.
26933 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
26936 * config/i386/mmx.md (*mov<mode>_internal): Enable
26937 alternatives 11, 12, 13 and 14 also for 32bit targets.
26938 Remove alternatives 15, 16, 17 and 18.
26939 * config/i386/sse.md (vec_concatv2di): Change
26940 alternative (!x, *y) to (x, ?!*Yn).
26942 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
26944 * dumpfile.h (enum dump_kind): Remove stray comma.
26946 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26948 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
26949 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
26950 predicate::num_conditions
26951 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
26952 (CHANGED): turn into predicate::changed.
26953 (agg_position_info): Move to ipa-predicate.h
26954 (add_condition, predicate::add_clause, predicate::operator &=,
26955 predicate::or_with, predicate::evaluate, predicate::probability,
26956 dump_condition, dump_clause, predicate::dump,
26957 predicate::remap_after_duplication, predicate::remap_after_inlining,
26958 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
26959 (evaluate_conditions_for_known_args): Update.
26960 (set_cond_stmt_execution_predicate): Update.
26961 * ipa-inline.h: Include ipa-predicate.h
26962 (condition, inline_param_summary, conditions, agg_position_info,
26963 predicate): Move to ipa-predicate.h
26964 * ipa-predicate.c: New file.
26965 * ipa-predicate.h: New file.
26967 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
26969 * final.c (leaf_function_p): Check we are not in a sequence.
26971 2017-05-18 Martin Liska <mliska@suse.cz>
26973 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
26974 * dumpfile.c (dump_register): Use new enum dump_kind.
26975 (get_dump_file_name): Likewise.
26976 (dump_enable_all): Likewise.
26977 (dump_switch_p_1): Likewise.
26978 (enable_rtl_dump_file): Remove usage of TDF_RTL.
26979 * dumpfile.h (enum dump_kind): New enum type.
26980 (struct dump_file_info): Create constructor and
26981 format fields and comments.
26982 * passes.c (pass_manager::register_one_dump_file):
26984 * statistics.c (statistics_early_init): Likewise.
26985 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
26986 TDF_TREE with TDF_SLIM.
26987 (gather_memory_references_ref): Likewise.
26989 2017-05-18 Martin Liska <mliska@suse.cz>
26991 * vec.h (struct vnull): Use it.
26993 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26995 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
26996 (true_predicate, false_predicate, true_predicate_p,
26997 false_predicate_p): Remove.
26998 (single_cond_predicate, not_inlined_predicate): Turn to member function
27000 (add_condition): Update.
27001 (add_clause): Turn to...
27002 (predicate::add_clause): ... this one; update; allow passing NULL
27004 (and_predicates): Turn to ...
27005 (predicate::operator &=): ... this one.
27006 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27007 (or_predicates): Turn to ...
27008 (predicate::or_with): ... this one.
27009 (evaluate_predicate): Turn to ...
27010 (predicate::evaluate): ... this one.
27011 (predicate_probability): Turn to ...
27012 (predicate::probability): ... this one.
27013 (dump_condition): Update.
27014 (dump_predicate): Turn to ...
27015 (predicate::dump): ... this one.
27016 (account_size_time): Update.
27017 (edge_set_predicate): Update.
27018 (set_hint_predicate): UPdate.
27019 (evaluate_conditions_for_known_args): Update.
27020 (evaluate_properties_for_edge): Update.
27021 (remap_predicate_after_duplication): Turn to...
27022 (predicate::remap_after_duplication): ... this one.
27023 (remap_hint_predicate_after_duplication): Update.
27024 (inline_summary_t::duplicate): UPdate.
27025 (dump_inline_edge_summary): Update.
27026 (dump_inline_summary): Update.
27027 (set_cond_stmt_execution_predicate): Update.
27028 (set_switch_stmt_execution_predicate): Update.
27029 (compute_bb_predicates): Update.
27030 (will_be_nonconstant_expr_predicate): Update.
27031 (will_be_nonconstant_predicate): Update.
27032 (phi_result_unknown_predicate): Update.
27033 (predicate_for_phi_result): Update.
27034 (array_index_predicate): Update.
27035 (estimate_function_body_sizes): Update.
27036 (estimate_node_size_and_time): Update.
27037 (estimate_ipcp_clone_size_and_time): Update.
27038 (remap_predicate): Rename to ...
27039 (predicate::remap_after_inlining): ... this one.
27040 (remap_hint_predicate): Update.
27041 (inline_merge_summary): Update.
27042 (inline_update_overall_summary): Update.
27043 (estimate_size_after_inlining): Update.
27044 (read_predicate): Rename to ...
27045 (predicate::stream_in): ... this one.
27046 (read_inline_edge_summary): Update.
27047 (write_predicate): Rename to ...
27048 (predicate::stream_out): ... this one.
27049 (write_inline_edge_summary): Update.
27050 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27051 (clause_t): Turn to uint32_t
27052 (predicate): Turn to class; implement constructor and operators
27054 (size_time_entry): Update.
27055 (inline_summary): Update.
27056 (inline_edge_summary): Update.
27058 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27060 * fold-const.c (fold_binary_loc): Move transformation...
27061 * match.pd (C - X CMP X): ... here.
27063 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27065 * config/sparc/sparc.c (sparc_option_override): Set function
27066 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27067 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27069 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27071 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27073 2017-05-18 Marek Polacek <polacek@redhat.com>
27076 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27077 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27078 gimple_assign_load_p.
27080 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27082 PR middle-end/80692
27083 * real.c (do_compare): Give decimal_do_compare preference over
27084 comparing just the signs.
27086 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27088 * doc/md.texi (Canonicalization of Instructions): Describe the
27089 canonical form of instructions that inherently set a condition
27092 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27094 PR middle-end/80775
27095 * tree-cfg.c: Move deletion of unreachable case statements to after
27096 the merging of consecutive case labels.
27098 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27100 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27101 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27102 restoring of callee-saved registers.
27104 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27106 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27107 * config/visium/visium.c (single_set_and_flags): Likewise.
27108 * config/visium/visium.md (Substitutions): Likewise.
27110 2017-05-17 Martin Liska <mliska@suse.cz>
27112 * cfg.c: Introduce dump_flags_t type and
27113 use it instead of int type.
27115 * cfghooks.c: Likewise.
27116 * cfghooks.h (struct cfg_hooks): Likewise.
27117 * cfgrtl.c: Likewise.
27118 * cfgrtl.h: Likewise.
27119 * cgraph.c (cgraph_node::get_body): Likewise.
27120 * coretypes.h: Likewise.
27121 * domwalk.c: Likewise.
27122 * domwalk.h: Likewise.
27123 * dumpfile.c (struct dump_option_value_info): Likewise.
27124 (dump_enable_all): Likewise.
27125 (dump_switch_p_1): Likewise.
27126 (opt_info_switch_p): Likewise.
27127 * dumpfile.h (enum tree_dump_index): Likewise.
27128 (struct dump_file_info): Likewise.
27129 * genemit.c: Likewise.
27130 * generic-match-head.c: Likewise.
27131 * gengtype.c (open_base_files): Likewise.
27132 * gimple-pretty-print.c: Likewise.
27133 * gimple-pretty-print.h: Likewise.
27134 * graph.c (print_graph_cfg): Likewise.
27135 * graphite-scop-detection.c (dot_all_sese): Likewise.
27136 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27137 * loop-unroll.c (report_unroll): Likewise.
27138 * passes.c (pass_manager::register_one_dump_file): Likewise.
27139 * print-tree.c: Likewise.
27140 * statistics.c: Likewise.
27141 * tree-cfg.c: Likewise.
27142 * tree-cfg.h: Likewise.
27143 * tree-dfa.c: Likewise.
27144 * tree-dfa.h: Likewise.
27145 * tree-dump.c (dump_function): Likewise.
27146 * tree-dump.h (struct dump_info): Likewise.
27147 * tree-pretty-print.c: Likewise.
27148 * tree-pretty-print.h: Likewise.
27149 * tree-ssa-live.c: Likewise.
27150 * tree-ssa-live.h: Likewise.
27151 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27152 * tree-vect-loop.c: Likewise.
27153 * tree-vect-slp.c: Likewise.
27155 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27156 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27158 PR tree-optimization/80457
27159 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27160 of all arguments to a statement as scalar_to_vec operations.
27161 (vectorizable_call): Adjust call to vect_model_simple_cost for
27163 (vectorizable_conversion): Likewise.
27164 (vectorizable_assignment): Likewise.
27165 (vectorizable_shift): Likewise.
27166 (vectorizable_operation): Likewise.
27167 (vectorizable_comparison): Likewise.
27168 (vect_is_simple_cond): Record the def types for operands.
27169 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27170 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27171 for statement argument count.
27173 2017-05-16 Carl Love <cel@us.ibm.com>
27175 * config/rs6000/rs6000-c: Add support for built-in functions
27176 vector unsigned long long vec_bperm (vector unsigned long long,
27177 vector unsigned char)
27178 vector signed long long vec_mule (vector signed int,
27180 vector unsigned long long vec_mule (vector unsigned int,
27181 vector unsigned int)
27182 vector signed long long vec_mulo (vector signed int,
27184 vector unsigned long long vec_mulo (vector unsigned int,
27185 vector unsigned int)
27186 vector signed char vec_sldw (vector signed char,
27187 vector signed char,
27189 vector unsigned char vec_sldw (vector unsigned char,
27190 vector unsigned char,
27192 vector signed short vec_sldw (vector signed short,
27193 vector signed short,
27195 vector unsigned short vec_sldw (vector unsigned short,
27196 vector unsigned short,
27198 vector signed int vec_sldw (vector signed int,
27201 vector unsigned int vec_sldw (vector unsigned int,
27202 vector unsigned int,
27204 vector signed long long vec_sldw (vector signed long long,
27205 vector signed long long,
27207 vector unsigned long long vec_sldw (vector unsigned long long,
27208 vector unsigned long long,
27210 * config/rs6000/rs6000-c: Add support for built-in functions
27211 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27212 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27213 * doc/extend.texi: Update the built-in documentation for the
27214 new built-in functions.
27216 2017-05-16 Marek Polacek <polacek@redhat.com>
27220 * tree.c (save_expr): Don't fold the expression.
27222 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27224 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27225 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27226 and (?*y,m). Update insn attributes.
27228 2017-05-16 Martin Liska <mliska@suse.cz>
27230 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27231 flags argument of print_gimple_stmt, print_gimple_expr,
27232 print_generic_stmt and print_generic_expr.
27233 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27234 * coretypes.h: Likewise.
27235 * except.c (dump_eh_tree): Likewise.
27236 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27237 * gimple-pretty-print.h: Likewise.
27238 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27239 (backprop::push_to_worklist): Likewise.
27240 (backprop::pop_from_worklist): Likewise.
27241 (backprop::process_use): Likewise.
27242 (backprop::intersect_uses): Likewise.
27243 (note_replacement): Likewise.
27244 * gimple-ssa-store-merging.c
27245 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27246 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27247 (pass_store_merging::execute): Likewise.
27248 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27249 (ssa_base_cand_dump_callback): Likewise.
27250 (dump_incr_vec): Likewise.
27251 (replace_refs): Likewise.
27252 (replace_mult_candidate): Likewise.
27253 (create_add_on_incoming_edge): Likewise.
27254 (create_phi_basis): Likewise.
27255 (insert_initializers): Likewise.
27256 (all_phi_incrs_profitable): Likewise.
27257 (introduce_cast_before_cand): Likewise.
27258 (replace_one_candidate): Likewise.
27259 * gimplify.c (gimplify_expr): Likewise.
27260 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27261 (set_rename): Likewise.
27262 (rename_uses): Likewise.
27263 (copy_loop_phi_nodes): Likewise.
27264 (add_close_phis_to_merge_points): Likewise.
27265 (copy_loop_close_phi_args): Likewise.
27266 (copy_cond_phi_args): Likewise.
27267 (graphite_copy_stmts_from_block): Likewise.
27268 (translate_pending_phi_nodes): Likewise.
27269 * graphite-poly.c (print_pdr): Likewise.
27270 (dump_gbb_cases): Likewise.
27271 (dump_gbb_conditions): Likewise.
27272 (print_scop_params): Likewise.
27273 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27274 (build_cross_bb_scalars_use): Likewise.
27275 (gather_bbs::before_dom_children): Likewise.
27276 * hsa-dump.c (dump_hsa_immed): Likewise.
27277 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27278 (get_replacement_map): Likewise.
27279 * ipa-inline-analysis.c (dump_condition): Likewise.
27280 (estimate_function_body_sizes): Likewise.
27281 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27282 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27283 * ipa-prop.c (ipa_dump_param): Likewise.
27284 (ipa_print_node_jump_functions_for_edge): Likewise.
27285 (ipa_modify_call_arguments): Likewise.
27286 (ipa_modify_expr): Likewise.
27287 (ipa_dump_param_adjustments): Likewise.
27288 (ipa_dump_agg_replacement_values): Likewise.
27289 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27290 * ipa-pure-const.c (check_stmt): Likewise.
27291 (pass_nothrow::execute): Likewise.
27292 * ipa-split.c (execute_split_functions): Likewise.
27293 * omp-offload.c (dump_oacc_loop_part): Likewise.
27294 (dump_oacc_loop): Likewise.
27295 * trans-mem.c (tm_log_emit): Likewise.
27296 (tm_memopt_accumulate_memops): Likewise.
27297 (dump_tm_memopt_set): Likewise.
27298 (dump_tm_memopt_transform): Likewise.
27299 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27300 (print_loop): Likewise.
27301 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27302 (chkp_gather_checks_info): Likewise.
27303 (chkp_get_check_result): Likewise.
27304 (chkp_remove_check_if_pass): Likewise.
27305 (chkp_use_outer_bounds_if_possible): Likewise.
27306 (chkp_reduce_bounds_lifetime): Likewise.
27307 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27308 (chkp_mark_completed_bounds): Likewise.
27309 (chkp_register_incomplete_bounds): Likewise.
27310 (chkp_mark_invalid_bounds): Likewise.
27311 (chkp_maybe_copy_and_register_bounds): Likewise.
27312 (chkp_build_returned_bound): Likewise.
27313 (chkp_get_bound_for_parm): Likewise.
27314 (chkp_build_bndldx): Likewise.
27315 (chkp_get_bounds_by_definition): Likewise.
27316 (chkp_generate_extern_var_bounds): Likewise.
27317 (chkp_get_bounds_for_decl_addr): Likewise.
27318 * tree-chrec.c (chrec_apply): Likewise.
27319 * tree-data-ref.c (dump_data_reference): Likewise.
27320 (dump_subscript): Likewise.
27321 (dump_data_dependence_relation): Likewise.
27322 (analyze_overlapping_iterations): Likewise.
27323 * tree-inline.c (expand_call_inline): Likewise.
27324 (tree_function_versioning): Likewise.
27325 * tree-into-ssa.c (dump_defs_stack): Likewise.
27326 (dump_currdefs): Likewise.
27327 (dump_names_replaced_by): Likewise.
27328 (dump_update_ssa): Likewise.
27329 (update_ssa): Likewise.
27330 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27331 * tree-parloops.c (build_new_reduction): Likewise.
27332 (try_create_reduction_list): Likewise.
27333 (ref_conflicts_with_region): Likewise.
27334 (oacc_entry_exit_ok_1): Likewise.
27335 (oacc_entry_exit_single_gang): Likewise.
27336 * tree-pretty-print.h: Likewise.
27337 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27338 (get_scalar_evolution): Likewise.
27339 (add_to_evolution): Likewise.
27340 (get_loop_exit_condition): Likewise.
27341 (analyze_evolution_in_loop): Likewise.
27342 (analyze_initial_condition): Likewise.
27343 (analyze_scalar_evolution): Likewise.
27344 (instantiate_scev): Likewise.
27345 (number_of_latch_executions): Likewise.
27346 (gather_chrec_stats): Likewise.
27347 (final_value_replacement_loop): Likewise.
27348 (scev_const_prop): Likewise.
27349 * tree-sra.c (dump_access): Likewise.
27350 (disqualify_candidate): Likewise.
27351 (create_access): Likewise.
27352 (reject): Likewise.
27353 (maybe_add_sra_candidate): Likewise.
27354 (create_access_replacement): Likewise.
27355 (analyze_access_subtree): Likewise.
27356 (analyze_all_variable_accesses): Likewise.
27357 (sra_modify_assign): Likewise.
27358 (initialize_constant_pool_replacements): Likewise.
27359 (find_param_candidates): Likewise.
27360 (decide_one_param_reduction): Likewise.
27361 (replace_removed_params_ssa_names): Likewise.
27362 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27363 * tree-ssa-copy.c (dump_copy_of): Likewise.
27364 (copy_prop_visit_cond_stmt): Likewise.
27365 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27366 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27367 (record_equivalences_from_stmt): Likewise.
27368 * tree-ssa-dse.c (compute_trims): Likewise.
27369 (delete_dead_call): Likewise.
27370 (delete_dead_assignment): Likewise.
27371 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27372 (forward_propagate_into_cond): Likewise.
27373 (pass_forwprop::execute): Likewise.
27374 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27375 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27377 (move_computations_worker): Likewise.
27378 (execute_sm): Likewise.
27379 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27380 (remove_exits_and_undefined_stmts): Likewise.
27381 (remove_redundant_iv_tests): Likewise.
27382 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27383 (adjust_iv_update_pos): Likewise.
27384 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27385 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27386 (value_replacement): Likewise.
27387 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27388 * tree-ssa-pre.c (print_pre_expr): Likewise.
27389 (get_representative_for): Likewise.
27390 (create_expression_by_pieces): Likewise.
27391 (insert_into_preds_of_block): Likewise.
27392 (eliminate_insert): Likewise.
27393 (eliminate_dom_walker::before_dom_children): Likewise.
27394 (eliminate): Likewise.
27395 (remove_dead_inserted_code): Likewise.
27396 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27397 * tree-ssa-reassoc.c (get_rank): Likewise.
27398 (eliminate_duplicate_pair): Likewise.
27399 (eliminate_plus_minus_pair): Likewise.
27400 (eliminate_not_pairs): Likewise.
27401 (undistribute_ops_list): Likewise.
27402 (eliminate_redundant_comparison): Likewise.
27403 (update_range_test): Likewise.
27404 (optimize_range_tests_var_bound): Likewise.
27405 (optimize_vec_cond_expr): Likewise.
27406 (rewrite_expr_tree): Likewise.
27407 (rewrite_expr_tree_parallel): Likewise.
27408 (linearize_expr): Likewise.
27409 (break_up_subtract): Likewise.
27410 (linearize_expr_tree): Likewise.
27411 (attempt_builtin_powi): Likewise.
27412 (attempt_builtin_copysign): Likewise.
27413 (transform_stmt_to_copy): Likewise.
27414 (transform_stmt_to_multiply): Likewise.
27415 (dump_ops_vector): Likewise.
27416 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27417 (print_scc): Likewise.
27418 (set_ssa_val_to): Likewise.
27419 (visit_reference_op_store): Likewise.
27420 (visit_use): Likewise.
27421 (sccvn_dom_walker::before_dom_children): Likewise.
27422 (run_scc_vn): Likewise.
27423 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27425 (expr_hash_elt::print): Likewise.
27426 (const_and_copies::pop_to_marker): Likewise.
27427 (const_and_copies::record_const_or_copy_raw): Likewise.
27428 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27429 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27430 (dump_predicates): Likewise.
27431 (find_uninit_use): Likewise.
27432 (warn_uninitialized_phi): Likewise.
27433 (pass_late_warn_uninitialized::execute): Likewise.
27434 * tree-ssa.c (verify_vssa): Likewise.
27435 (verify_ssa): Likewise.
27436 (maybe_optimize_var): Likewise.
27437 * tree-vrp.c (dump_value_range): Likewise.
27438 (dump_all_value_ranges): Likewise.
27439 (dump_asserts_for): Likewise.
27440 (register_edge_assert_for_2): Likewise.
27441 (vrp_visit_cond_stmt): Likewise.
27442 (vrp_visit_switch_stmt): Likewise.
27443 (vrp_visit_stmt): Likewise.
27444 (vrp_visit_phi_node): Likewise.
27445 (simplify_cond_using_ranges_1): Likewise.
27446 (fold_predicate_in): Likewise.
27447 (evrp_dom_walker::before_dom_children): Likewise.
27448 (evrp_dom_walker::push_value_range): Likewise.
27449 (evrp_dom_walker::pop_value_range): Likewise.
27450 (execute_early_vrp): Likewise.
27452 2017-05-16 Richard Biener <rguenther@suse.de>
27454 * dwarf2out.c (loc_list_from_tree_1): Do not create
27455 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27457 2017-05-16 Richard Biener <rguenther@suse.de>
27459 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27461 (note_variable_value_in_expr): If we resolved the decl ref
27462 do not push to the stack.
27464 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27466 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27467 operations in fast-math mode.
27468 (vaddq_f16): Likewise.
27469 (vmul_f16): Likewise.
27470 (vmulq_f16): Likewise.
27471 (vsub_f16): Likewise.
27472 (vsubq_f16): Likewise.
27473 * config/arm/neon.md (add<mode>3): New.
27475 (fma:<VH:mode>3): New. Also remove outdated comment.
27478 2017-05-16 Martin Liska <mliska@suse.cz>
27482 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27483 (odr_types_equivalent_p): Likewise.
27485 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27487 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27489 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27492 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27493 non-interunit SSE move alternatives with '?'.
27494 (zero-extendsidi peephole2): New peephole to skip intermediate
27495 general register in SSE zero-extend sequence.
27497 2017-05-15 Jeff Law <law@redhat.com>
27499 * reorg.c (relax_delay_slots): Create a new variable to hold
27500 the temporary target rather than clobbering TARGET_LABEL.
27502 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27503 missing argument to extract_bit_field call.
27504 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27506 2017-05-15 Martin Liska <mliska@suse.cz>
27509 * gcc.c (process_command): Do not allow empty argument of -o option.
27511 2017-05-15 Renlin Li <renlin.li@arm.com>
27513 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27514 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27515 * config/aarch64/constraints.md (Usf): Add long call check.
27516 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27517 (call_value): Likewise.
27518 (sibcall): Likewise.
27519 (sibcall_value): Likewise.
27521 (call_value_insn): New.
27522 (sibcall_insn): Update rtx pattern.
27523 (sibcall_value_insn): Likewise.
27524 (call_internal): Remove.
27525 (call_value_internal): Likewise.
27526 (sibcall_internal): Likewise.
27527 (sibcall_value_internal): Likewise.
27528 (call_reg): Likewise.
27529 (call_symbol): Likewise.
27530 (call_value_reg): Likewise.
27531 (call_value_symbol): Likewise.
27533 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27536 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27538 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27540 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27541 compatible with CCGOCmode and with CCZmode.
27543 2017-05-14 Martin Sebor <msebor@redhat.com>
27545 PR middle-end/77671
27546 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27547 (gimple_fold_builtin_snprintf): Same.
27548 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27549 (gimple_fold_builtin_snprintf): Same.
27550 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27551 of character types.
27552 (is_call_safe): New function.
27553 (try_substitute_return_value): Call it.
27554 (try_simplify_call): New function.
27555 (pass_sprintf_length::handle_gimple_call): Call it.
27557 2017-05-14 Martin Sebor <msebor@redhat.com>
27559 PR middle-end/80669
27560 * builtins.c (expand_builtin_stpncpy): Simplify.
27562 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27564 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27565 * config/i386/i386.h
27566 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27567 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27568 (struct machine_function): Add new members call_ms2sysv,
27569 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27570 (struct machine_frame_state): New fields sp_realigned and
27571 sp_realigned_offset.
27572 * config/i386/i386.c
27573 (enum xlogue_stub): New enum.
27574 (enum xlogue_stub_sets): New enum.
27575 (class xlogue_layout): New class.
27576 (struct ix86_frame): New fields stack_realign_allocate_offset,
27577 stack_realign_offset and outlined_save_offset. Modify comments to
27578 detail stack layout when using out-of-line stubs.
27579 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27580 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27581 -mcall-ms2sysv-xlogues.
27582 (stub_managed_regs): New static variable.
27583 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27584 registers managed by out-of-line stub.
27585 (disable_call_ms2sysv_xlogues): New function.
27586 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27587 m->call_ms2sysv when appropriate and compute frame layout for
27589 (sp_valid_at, fp_valid_at): New inline functions.
27590 (choose_basereg): New function.
27591 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27593 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27594 Use align parameter of choose_baseaddr to generated aligned SSE movs
27596 (pro_epilogue_adjust_stack): Modify to track
27597 machine_frame_state::sp_realigned.
27598 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
27599 (ix86_nsaved_sseregs): Likewise.
27600 (ix86_emit_save_regs): Likewise.
27601 (ix86_emit_save_regs_using_mov): Likewise.
27602 (ix86_emit_save_sse_regs_using_mov): Likewise.
27603 (get_scratch_register_on_entry): Likewise.
27604 (gen_frame_set): New function.
27605 (gen_frame_load): Likewise.
27606 (gen_frame_store): Likewise.
27607 (emit_outlined_ms2sysv_save): Likewise.
27608 (emit_outlined_ms2sysv_restore): Likewise.
27609 (ix86_expand_prologue): Modify stack re-alignment code and call
27610 emit_outlined_ms2sysv_save when appropriate.
27611 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
27612 parameter rtx_insn *insn, which allows the function to be used to only
27613 generate the notes.
27614 (ix86_expand_epilogue): Modify validity checks of frame and stack
27615 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
27616 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
27617 * config/i386/predicates.md
27618 (save_multiple): New predicate.
27619 (restore_multiple): Likewise.
27620 * config/i386/sse.md
27621 (save_multiple<mode>): New pattern.
27622 (save_multiple_realign<mode>): Likewise.
27623 (restore_multiple<mode>): Likewise.
27624 (restore_multiple_and_return<mode>): Likewise.
27625 (restore_multiple_leave_return<mode>): Likewise.
27626 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
27628 2017-05-14 Julia Koval <julia.koval@intel.com>
27630 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
27631 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
27632 (__builtin_ia32_xsetbv): New builtins.
27633 * config/i386/i386.c (ix86_expand_special_args_builtin):
27635 (ix86_expand_builtin): Special expand for new intrinsics.
27636 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
27637 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
27638 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
27640 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27642 * cfganal.c (inverted_post_order_compute): Change argument type
27644 * cfganal.h (inverted_post_order_compute): Adjust prototype.
27645 * df-core.c (rest_of_handle_df_initialize): Adjust.
27646 (rest_of_handle_df_finish): Likewise.
27647 (df_analyze_1): Likewise.
27648 (df_analyze): Likewise.
27649 (loop_inverted_post_order_compute): Change argument to be a vec *.
27650 (df_analyze_loop): Adjust.
27651 (df_get_n_blocks): Likewise.
27652 (df_get_postorder): Likewise.
27653 * df.h (struct df_d): Change field to be a vec.
27654 * lcm.c (compute_laterin): Adjust.
27655 (compute_available): Likewise.
27656 * lra-lives.c (lra_create_live_ranges_1): Likewise.
27657 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
27658 * tree-ssa-pre.c (compute_antic): Likewise.
27660 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27662 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
27663 (depth_first_search::depth_first_search): Change structure init
27664 function to this constructor.
27665 (depth_first_search::add_bb): Rename function to this member.
27666 (depth_first_search::execute): Likewise.
27667 (flow_dfs_compute_reverse_finish): Adjust.
27669 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27671 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
27672 (longest_simple_path): Likewise.
27673 * shrink-wrap.c (spread_components): Likewise.
27674 (disqualify_problematic_components): Likewise.
27675 (emit_common_heads_for_components): Likewise.
27676 (emit_common_tails_for_components): Likewise.
27677 (insert_prologue_epilogue_for_components): Likewise.
27679 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27681 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
27684 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27686 * df-core.c (df_set_blocks): Start using auto_bitmap.
27687 (df_compact_blocks): Likewise.
27688 * df-problems.c (df_rd_confluence_n): Likewise.
27689 * df-scan.c (df_insn_rescan_all): Likewise.
27690 (df_process_deferred_rescans): Likewise.
27691 (df_update_entry_block_defs): Likewise.
27692 (df_update_exit_block_uses): Likewise.
27693 (df_entry_block_bitmap_verify): Likewise.
27694 (df_exit_block_bitmap_verify): Likewise.
27695 (df_scan_verify): Likewise.
27696 * lra-constraints.c (lra_constraints): Likewise.
27697 (undo_optional_reloads): Likewise.
27698 (lra_undo_inheritance): Likewise.
27699 * lra-remat.c (calculate_gen_cands): Likewise.
27700 (do_remat): Likewise.
27701 * lra-spills.c (assign_spill_hard_regs): Likewise.
27702 (spill_pseudos): Likewise.
27703 * tree-ssa-pre.c (bitmap_set_and): Likewise.
27704 (bitmap_set_subtract_values): Likewise.
27706 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27708 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
27709 management with auto_bitmap.
27710 (fix_inter_tick): Likewise.
27711 (fix_recovery_deps): Likewise.
27712 * ira.c (add_store_equivs): Likewise.
27713 (find_moveable_pseudos): Likewise.
27714 (split_live_ranges_for_shrink_wrap): Likewise.
27715 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
27716 (rtx_reuse_manager::seen_def_p): Likewise.
27717 (rtx_reuse_manager::set_seen_def): Likewise.
27718 * print-rtl.h (class rtx_reuse_manager): Likewise.
27720 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27722 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
27724 (migrate_btr_def): Likewise.
27725 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
27726 * df-core.c (loop_post_order_compute): Likewise.
27727 (loop_inverted_post_order_compute): Likewise.
27728 * hsa-common.h: Likewise.
27729 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
27730 * init-regs.c (initialize_uninitialized_regs): Likewise.
27731 * ipa-inline.c (resolve_noninline_speculation): Likewise.
27732 (inline_small_functions): Likewise.
27733 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
27734 * ira.c (combine_and_move_insns): Likewise.
27735 (build_insn_chain): Likewise.
27736 * loop-invariant.c (find_invariants): Likewise.
27737 * lower-subreg.c (propagate_pseudo_copies): Likewise.
27738 * predict.c (tree_predict_by_opcode): Likewise.
27739 (predict_paths_leading_to): Likewise.
27740 (predict_paths_leading_to_edge): Likewise.
27741 (estimate_loops_at_level): Likewise.
27742 (estimate_loops): Likewise.
27743 * shrink-wrap.c (try_shrink_wrapping): Likewise.
27744 (spread_components): Likewise.
27745 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
27746 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
27747 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
27748 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
27749 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
27750 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
27751 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
27752 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
27753 (mark_threaded_blocks): Likewise.
27754 (thread_through_all_blocks): Likewise.
27755 * tree-ssa.c (verify_ssa): Likewise.
27756 (execute_update_addresses_taken): Likewise.
27757 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
27759 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27761 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
27763 (post_order_compute): Likewise.
27764 (inverted_post_order_compute): Likewise.
27765 (pre_and_rev_post_order_compute_fn): Likewise.
27767 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27769 * genrecog.c (int_set::int_set): Explicitly construct our
27770 auto_vec base class.
27771 * vec.h (auto_vec::auto_vec): New constructor.
27773 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27775 * bitmap.h (class auto_bitmap): New constructor taking
27776 bitmap_obstack * argument.
27778 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27780 * bitmap.h (class auto_bitmap): Change type of m_bits to
27781 bitmap_head, and adjust ctor / dtor and member operators.
27783 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
27785 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
27786 when returned register mode doesn't match original mode.
27788 2017-05-12 Jeff Law <law@redhat.com>
27789 Jakub Jelinek <jakub@redhat.com>
27791 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
27792 we look for cc setter after the compare-elim changes.
27793 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
27794 within the vector to match what compare-elim now expects.
27795 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
27796 (xorsi3_flags, one_cmplsi2_flags): Likewise.
27798 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
27799 after the compare-elim changes.
27800 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
27801 the vector to match what compare-elim now expects.
27802 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
27803 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
27804 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
27805 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
27806 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
27808 * config/visium/visium.c (single_set_and_flags): Fix where
27809 we look for cc setter after the compare-elim changes.
27810 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
27811 with the vector to match what compare-elim now expects.
27812 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
27813 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
27814 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
27815 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
27816 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
27817 (neg<mode>2_insn_set_overflow): Likewise.
27819 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
27821 PR middle-end/79794
27822 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
27823 maybe_expand_insn call, set ops[0].target. If still set after call,
27824 set alt_rtl. Add extra arg to recursive calls.
27825 (extract_bit_field): Add alt_rtl argument. Pass to
27827 * expmed.h (extract_bit_field): Fix prototype.
27828 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
27829 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
27830 to extract_bit_field_calls.
27831 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
27832 Pass alt_rtl to extract_bit_field calls.
27833 * calls.c (store_unaligned_arguments_into_psuedos)
27834 load_register_parameters): Pass extra NULL to extract_bit_field calls.
27835 * optabs.c (maybe_legitimize_operand): Clear op->target when call
27837 * optabs.h (struct expand_operand): Add target bitfield.
27839 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27841 * compare-elim.c (try_eliminate_compare): Canonicalize
27842 operation with embedded compare to
27843 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
27844 (set (reg) (operation)].
27846 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
27848 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27851 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
27852 cost of adding a carry flag for ADC instruction.
27853 [case MINUS]: Ignore the cost of subtracting a carry flag
27854 for SBB instruction.
27856 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
27858 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
27860 * config/rs6000/bmiintrin.h: New file.
27861 * config/rs6000/bmi2intrin.h: New file.
27862 * config/rs6000/x86intrin.h: New file.
27864 2017-05-12 Jeff Law <law@redhat.com>
27866 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
27869 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
27871 PR middle-end/80707
27872 * tree-cfg.c: Remove cfg edges of unreachable case statements.
27874 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27876 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27877 early expansion of vector divide builtins.
27878 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
27879 builtins identified as having unsigned arguments.
27881 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27883 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
27884 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
27885 expansion of vector logical operations (and, andc, or, xor,
27888 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27890 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
27891 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
27893 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27895 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27896 early GIMPLE expansion of vector multiplies.
27898 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27900 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
27901 TARGET_HAVE_MOVT conditional.
27902 (movt splitter): Likewise.
27904 2017-05-12 Richard Biener <rguenther@suse.de>
27906 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
27907 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27908 Fold all stmts not inplace.
27910 2017-05-12 Richard Biener <rguenther@suse.de>
27912 PR tree-optimization/80713
27913 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
27914 inserted_exprs bit for not removed stmts.
27916 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
27918 PR middle-end/69921
27919 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
27920 parallelized" attribute for parallelized OpenACC kernels.
27921 * omp-offload.c (execute_oacc_device_lower): Use it.
27923 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
27924 Set "oacc kernels" attribute.
27925 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
27926 parameter. Adjust all users.
27927 (oacc_fn_attrib_kernels_p): Remove function.
27928 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
27929 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
27930 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
27931 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
27932 assert "oacc kernels" attribute is set.
27934 2017-05-11 Carl Love <cel@us.ibm.com>
27936 * config/rs6000/rs6000-c: Add support for built-in functions
27937 vector unsigned char vec_popcnt (vector signed char)
27938 vector unsigned char vec_popcnt (vector unsigned char)
27939 vector unsigned short vec_popcnt (vector signed short)
27940 vector unsigned short vec_popcnt (vector unsigned short)
27941 vector unsigned int vec_popcnt (vector signed int)
27942 vector unsigned int vec_popcnt (vector unsigned int)
27943 vector unsigned long long vec_popcnt (vector signed long long)
27944 vector unsigned long long vec_popcnt (vector unsigned long long)
27945 vector signed long long vec_slo (vector signed long long,
27946 vector signed char)
27947 vector signed long long vec_slo (vector signed long long,
27948 vector unsigned char)
27949 vector unsigned long long vec_slo (vector unsigned long long,
27950 vector signed char)
27951 vector unsigned long long vec_slo (vector unsigned long long,
27952 vector unsigned char)
27953 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
27954 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
27955 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
27956 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
27957 * doc/extend.texi: Update the built-in documentation file for the
27958 new built-in functions.
27960 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27962 * attribs.h (sorted_attr_string): Move machine independent
27963 functions for target clone support from the i386 port to common
27964 code. Rename ix86_function_versions to common_function_versions.
27965 Rename make_name to make_unique_name.
27966 (common_function_versions): Likewise.
27967 (make_unique_name): Likewise.
27968 (make_dispatcher_decl): Likewise.
27969 (is_function_default_version): Likewise.
27970 * attribs.c (attr_strcmp): Likewise.
27971 (sorted_attr_string): Likewise.
27972 (common_function_versions): Likewise.
27973 (make_unique_name): Likewise.
27974 (make_dispatcher_decl): Likewise.
27975 (is_function_default_version): Likewise.
27976 * config/i386/i386.c (attr_strcmp): Likewise.
27977 (sorted_attr_string): Likewise.
27978 (ix86_function_versions): Likewise.
27979 (make_name): Likewise.
27980 (make_dispatcher_decl): Likewise.
27981 (is_function_default_version): Likewise.
27982 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
27984 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27987 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
27988 Account for direct move costs for vec_construct of integer
27991 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
27994 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
27995 (UNSPEC_STX_ATOMIC): Ditto.
27996 (loaddi_via_sse): New insn.
27997 (storedi_via_sse): Ditto.
27998 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
27999 Update corresponding peephole2 patterns.
28000 (atomic_storedi_fpu): Ditto.
28002 2017-05-11 Julia Koval <julia.koval@intel.com>
28004 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28005 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28007 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28008 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28009 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28011 2017-05-11 Nathan Sidwell <nathan@acm.org>
28013 * graphite-poly.c: Include dumpfile.h.
28015 * dumpfle.h (dump_function): Declare here ...
28016 * tree-dump.h (dump_function): ... not here.
28017 * dumpfile.c: #include tree-cfg.h.
28018 (dump_function): Move here from ...
28019 * tree-dump.c (dump_function): ... here.
28020 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28021 * graphite-poly.c: Don't include tree-dump.h.
28022 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28023 * print-tree.c: Likewise.
28024 * stor-layout.c: Likewise.
28025 * tree-nested.c: Likewise.
28027 * dumpfile.c (dump_start): Use TDF_FLAGS.
28028 (dump_enable_all): Fix TDF_KIND check thinko.
28030 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28032 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28033 array entries to represent two legal parameterizations of the
28034 overloaded __builtin_cmpb function, as represented by the
28035 P6_OV_BUILTIN_CMPB constant.
28036 (altivec_resolve_overloaded_builtin): Add special case handling
28037 for the __builtin_cmpb function, as represented by the
28038 P6_OV_BUILTIN_CMPB constant.
28039 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28040 (BU_P6_64BIT_2): New macro.
28041 (BU_P6_OVERLOAD_2): New macro
28042 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28043 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28044 (CMPB): Add overload support to represent both 32-bit and 64-bit
28045 compare-bytes function.
28046 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28047 support for TARGET_CMPB.
28048 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28049 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28050 documentation of the __builtin_cmpb overloaded built-in function.
28052 2017-05-11 Richard Biener <rguenther@suse.de>
28054 PR tree-optimization/80705
28055 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28056 bases are not vectorizable.
28058 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28060 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28061 when counting register pressure.
28063 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28065 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28066 (iv_ca_more_deps): Renamed to ...
28067 (iv_ca_compare_deps): ... this.
28068 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28070 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28072 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28074 (determine_group_iv_costs): ... here.
28075 (find_inv_vars_cb): Record inv var if it's not recorded before.
28077 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28079 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28080 (get_shiftadd_cost): Ditto.
28082 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28084 * tree-ssa-address.c: Include header file.
28085 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28087 (add_to_parts): Refactor.
28088 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28089 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28092 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28094 PR tree-optimization/53090
28095 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28097 (extract_cond_operands): Detect condition with IV on both sides
28098 and return COMP_IV_EXPR_2.
28099 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28100 (rewrite_use_compare): Simplify by removing call to function
28101 extract_cond_operands.
28103 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28105 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28106 (extract_cond_operands): Detect condition comparing against non-
28107 invariant bound and return appropriate enum value.
28108 (find_interesting_uses_cond): Update use of extract_cond_operands.
28109 Handle its return value accordingly.
28110 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28112 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28114 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28115 nonlinear iv_use computation in loop invariant sensitive way.
28117 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28119 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28120 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28122 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28124 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28125 (dump_cand): Support iv_cand.inv_exprs.
28126 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28128 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28131 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28133 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28135 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28136 as local function. Include necessary header files.
28137 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28139 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28141 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28143 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28145 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28146 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28147 RSHIFT_EXPR and BIT_NOT_EXPR.
28149 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28151 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28152 (adjust_setup_cost): New parameter supporting round up adjustment.
28153 (struct address_cost_data): Delete.
28154 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28155 (split_address_cost, ptr_difference_cost): Delete.
28156 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28157 (struct ainc_cost_data): New struct.
28158 (get_address_cost_ainc): New function.
28159 (get_address_cost, get_computation_cost): Reimplement.
28160 (determine_group_iv_cost_address): Record inv_expr for all uses of
28162 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28163 (iv_ca_has_deps): Reimplemented to ...
28164 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28166 (iv_ca_extend): Call iv_ca_more_deps.
28168 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28170 * tree-ssa-address.c (struct mem_address): Move to header file.
28171 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28172 * tree-ssa-address.h (struct mem_address): Move from C file.
28173 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28175 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28177 * tree-affine.h (aff_combination_type): New interface.
28178 (aff_combination_zero_p): Remove static.
28179 (aff_combination_const_p): New interface.
28180 (aff_combination_singleton_var_p): New interfaces.
28182 2017-05-11 Richard Biener <rguenther@suse.de>
28184 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28185 Skip unreachable blocks and destinations.
28186 (eliminate): Move stmt removal and fixup ...
28187 (fini_eliminate): ... here. Skip inserted exprs.
28188 (pass_pre::execute): Move fini_pre after fini_eliminate.
28189 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28190 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28191 PRE to get rid of dead code that has invalid SSA form and
28192 split critical edges again.
28194 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28196 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28198 2017-05-11 Richard Biener <rguenther@suse.de>
28200 * passes.c (execute_function_todo): Verify loops if they are
28201 said to be up-to-date.
28202 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28203 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28205 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28208 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28209 handle calling assemble_external ourself.
28212 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28213 modes with zero size. Enhance comment.
28215 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28217 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28218 built-ins for vec_xl and vec_xst with short and char pointer
28221 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28223 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28224 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28225 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28226 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28227 (_mm_maskz_min_round_ss): New intrinsics.
28228 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28229 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28230 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28231 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28232 (__builtin_ia32_minss_mask_round): New builtins.
28233 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28234 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28235 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28237 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28238 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28240 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28243 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28245 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28246 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28247 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28248 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28249 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28250 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28251 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28252 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28253 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28254 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28255 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28256 (__builtin_ia32_mulss_mask_round): New builtins.
28257 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28258 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28259 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28261 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28262 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28264 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28267 2017-05-10 Julia Koval <julia.koval@intel.com>
28269 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28270 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28271 (_mm256_setr_m128i): New intrinsics.
28273 2017-05-10 Julia Koval <julia.koval@intel.com>
28275 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28276 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28277 (_mm_maskz_rcp14_ss): New intrinsics.
28278 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28279 (__builtin_ia32_rcp14ss_mask): New builtins.
28280 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28282 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28284 PR tree-optimization/51513
28285 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28286 (assert_unreachable_fallthru_edge_p): Use it.
28287 (group_case_labels_stmt): Likewise.
28288 * tree-cfg.h: Prototype it.
28289 * stmt.c: Include cfghooks.h and tree-cfg.h.
28290 (emit_case_dispatch_table) <gap_label>: New local variable.
28291 Use it to fill dispatch table gaps.
28292 Test for default_label before updating probabilities.
28293 (expand_case) <default_label>: Remove unneeded initialization.
28294 Test for unreachable default case statement and remove its edge.
28295 Set default_label accordingly.
28296 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28298 2017-05-10 Carl Love <cel@us.ibm.com>
28300 * config/rs6000/rs6000-c: Add support for built-in functions
28301 vector signed char vec_neg (vector signed char)
28302 vector signed short int vec_neg (vector short int)
28303 vector signed int vec_neg (vector signed int)
28304 vector signed long long vec_neg (vector signed long long)
28305 vector float vec_neg (vector float)
28306 vector double vec_neg (vector double)
28307 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28309 * config/rs6000/altivec.h: Add define for vec_neg
28310 * doc/extend.texi: Update the built-in documentation for the
28311 new built-in functions.
28313 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28315 PR tree-optimization/77644
28316 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28318 2017-05-10 Nathan Sidwell <nathan@acm.org>
28320 * dumpfile.h (TDI_lang_all): New.
28321 (TDF_KIND): New. Renumber others
28322 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28324 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28326 (get_dump_file_name): Adjust suffix generation.
28327 (dump_enable_all): Use TDF_KIND.
28328 * doc/invoke.texi (-fdump-lang-all): Document.
28330 * dumpfile.h: Tabify.
28332 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28335 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28336 Move member access before delete.
28338 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28340 * tree-inline.c (expand_call_inline): Split block at stmt
28343 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28346 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28347 are now unused after splitting mov{sf,sd}_hardfloat.
28348 (f32_lr2): Likewise.
28349 (f32_lm): Likewise.
28350 (f32_lm2): Likewise.
28351 (f32_li): Likewise.
28352 (f32_li2): Likewise.
28353 (f32_lv): Likewise.
28354 (f32_sr): Likewise.
28355 (f32_sr2): Likewise.
28356 (f32_sm): Likewise.
28357 (f32_sm2): Likewise.
28358 (f32_si): Likewise.
28359 (f32_si2): Likewise.
28360 (f32_sv): Likewise.
28361 (f32_dm): Likewise.
28362 (f32_vsx): Likewise.
28363 (f32_av): Likewise.
28364 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28365 For movsf, order stores so the VSX stores occur before the GPR
28366 store which encourages the register allocator to use a traditional
28367 FPR instead of a GPR. For movsd, order the stores so that the GPR
28368 store comes before the VSX stores to allow the power6 to work.
28369 This is due to the power6 not having a 32-bit integer store
28370 instruction from a FPR.
28371 (movsf_hardfloat): Likewise.
28372 (movsd_hardfloat): Likewise.
28374 2017-05-09 Martin Sebor <msebor@redhat.com>
28376 PR translation/80280
28377 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28380 PR translation/80280
28381 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28382 data member added in r247778.
28383 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28385 2017-05-09 Nathan Sidwell <nathan@acm.org>
28387 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28389 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28392 2017-05-09 Marek Polacek <polacek@redhat.com>
28394 * doc/invoke.texi: Fix typo.
28396 2017-05-09 Richard Biener <rguenther@suse.de>
28398 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28399 (vrp_val_is_min): Likewise.
28400 (set_value_range_to_value): Likewise.
28401 (set_value_range_to_nonnegative): Likewise.
28402 (gimple_assign_nonzero_p): Likewise.
28403 (gimple_stmt_nonzero_p): Likewise.
28404 (vrp_int_const_binop): Likewise. Remove unreachable case.
28405 (adjust_range_with_scev): Adjust comments.
28406 (compare_range_with_value): Likewise.
28407 (extract_range_from_phi_node): Likewise.
28408 (test_for_singularity): Likewise.
28410 2017-05-09 Richard Biener <rguenther@suse.de>
28412 * tree-vrp.c (get_single_symbol): Add assert that we don't
28413 get overflowed constants as invariant part.
28414 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28415 checks. Use wi::cmp instead of recursing for integer constants.
28416 (compare_values): Just ignore whether we assumed undefined
28417 overflow instead of failing the compare.
28418 (extract_range_for_var_from_comparison_expr): Add comment before the
28419 TREE_NO_WARNING sets.
28420 (test_for_singularity): Likewise.
28421 (extract_range_from_comparison): Do not disable optimization
28422 when we assumed undefined overflow.
28423 (extract_range_basic): Remove init of unused var.
28425 2017-05-09 Richard Biener <rguenther@suse.de>
28427 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28428 (extract_range_from_multiplicative_op_1): Adjust.
28429 (extract_range_from_binary_expr_1): Use int_const_binop.
28431 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28434 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28435 rs6000_store_data_bypass_p in seven define_bypass directives and
28436 in several comments.
28437 * config/rs6000/rs6000-protos.h: Add prototype for
28438 rs6000_store_data_bypass_p function.
28439 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28440 function implements slightly different (rs6000-specific) semantics
28441 than store_data_bypass_p, returning false rather than aborting
28442 with assertion error when arguments do not satisfy the
28443 requirements of store data bypass.
28444 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28445 rs6000_store_data_bypass_p.
28447 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28449 * config/xtensa/xtensa-protos.h
28450 (xtensa_initial_elimination_offset): New declaration.
28451 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28452 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28453 macro definition, add case for FRAME_POINTER_REGNUM when
28454 FRAME_GROWS_DOWNWARD.
28455 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28456 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28457 xtensa_initial_elimination_offset.
28459 2017-05-08 Nathan Sidwell <nathan@acm.org>
28461 * doc/invoke.texi: Alphabetize -fdump options.
28463 2017-05-08 Martin Sebor <msebor@redhat.com>
28465 PR translation/80280
28466 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28468 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28470 * target.def (compute_frame_layout): New optional target hook.
28471 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28472 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28473 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28475 * reload1.c (verify_initial_elim_offsets): Likewise.
28476 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28477 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28478 (arm_get_frame_offsets): Split up into this ...
28479 (arm_compute_frame_layout): ... and this function.
28481 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28483 * config/aarch64/constraints.md (Usa): New constraint.
28484 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28486 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28488 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28489 with_multilib_list after it has been checked.
28491 2017-05-08 Richard Biener <rguenther@suse.de>
28493 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28494 (bitmap_set_subtract_values): Likewise.
28496 2017-05-08 Richard Biener <rguenther@suse.de>
28498 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28499 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28501 (gimple_stmt_nonzero_warnv_p): Rename to ...
28502 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28504 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28505 (extract_range_basic): Adjust, do not disable propagation on
28506 strict overflow sensitive simplification.
28507 (vrp_visit_cond_stmt): Likewise.
28509 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28511 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28512 body size unconditionally.
28514 2017-05-07 Jeff Law <law@redhat.com>
28517 2017-05-06 Jeff Law <law@redhat.com>
28518 PR tree-optimization/78496
28519 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28522 PR tree-optimization/78496
28523 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28524 (simplify_stmt_using_ranges): Call it.
28525 (vrp_dom_walker::before_dom_children): Extract equivalences
28526 from an ASSERT_EXPR with an equality comparison against a
28529 2017-05-06 Jeff Law <law@redhat.com>
28531 PR tree-optimization/78496
28532 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28535 PR tree-optimization/78496
28536 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28537 (simplify_stmt_using_ranges): Call it.
28538 (vrp_dom_walker::before_dom_children): Extract equivalences
28539 from an ASSERT_EXPR with an equality comparison against a
28542 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28544 * lra-constraints.c (lra_copy_reg_equiv): New function.
28545 (split_reg): Use it to copy equivalence information from the
28546 original register to the spill register.
28548 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28550 PR rtl-optimization/75964
28551 * simplify-rtx.c (simplify_const_relational_operation): Remove
28552 invalid handling of comparisons of integer ABS.
28554 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28556 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28557 initialize to zero.
28558 (init_regs): Remove declaration.
28559 (function_arg_advance_32): Initialize error_p as boolean variable.
28561 2017-05-05 Nathan Sidwell <nathan@acm.org>
28563 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28564 lines. Use for (;;).
28566 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28568 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28569 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28570 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28571 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28572 VF=2 that require versioning.
28574 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28576 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28579 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28581 * diagnostic.h (diagnostic_override_option_index): Convert from
28582 macro to inline function.
28584 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28586 * diagnostic.c (last_module_changed_p): New function.
28587 (set_last_module): New function.
28588 (diagnostic_report_current_module): Convert macro usage to
28589 the above functions.
28590 * diagnostic.h (diagnostic_context::last_module): Strengthen
28591 from const line_map * to const line_map_ordinary *.
28592 (diagnostic_last_module_changed): Delete macro.
28593 (diagnostic_set_last_module): Delete macro.
28595 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28597 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
28598 with diagnostic_report_diagnostic.
28599 (diagnostic_n_impl_richloc): Likewise.
28600 * diagnostic.h (report_diagnostic): Delete macro.
28601 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
28602 with diagnostic_report_diagnostic.
28603 * substring-locations.c (format_warning_va): Likewise.
28605 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28607 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
28608 save/restor of format_spec. Move option-printing code to...
28609 (print_option_information): ...this new function, and
28610 reimplement by simply printing to the pretty_printer,
28611 rather than appending to the format string.
28613 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28615 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
28616 handling logic into...
28617 (update_effective_level_from_pragmas): ...this new function.
28619 2017-05-04 Andrew Waterman <andrew@sifive.com>
28621 * config/riscv/riscv.opt (mstrict-align): New option.
28622 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
28623 (SLOW_UNALIGNED_ACCESS): Define.
28624 (riscv_slow_unaligned_access): Declare.
28625 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
28627 (riscv_slow_unaligned_access): New variable.
28628 (rocket_tune_info): Set slow_unaligned_access to true.
28629 (optimize_size_tune_info): Set slow_unaligned_access to false.
28630 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
28631 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
28632 (riscv_option_override): Set riscv_slow_unaligned_access.
28633 * doc/invoke.texi: Add -mstrict-align to RISC-V.
28635 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
28637 * config/riscv/riscv.md: Unify indentation.
28639 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28644 * config/rs6000/rs6000.md (u code attribute): Add FIX and
28646 (extendsi<mode>2): Add support for doing sign extension via
28647 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
28648 don't have ISA 3.0 instructions.
28649 (extendsi<mode>2 splitter): Likewise.
28650 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
28651 generate the normal insns since SImode can now go in vector
28652 registers. Disallow the special UNSPECs needed for previous
28653 machines to hide SImode being used. Add new insns
28654 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
28655 (fix_trunc<mode>si2_stfiwx): Likewise.
28656 (fix_trunc<mode>si2_internal): Likewise.
28657 (fixuns_trunc<mode>si2): Likewise.
28658 (fixuns_trunc<mode>si2_stfiwx): Likewise.
28659 (fctiw<u>z_<mode>_smallint): Likewise.
28660 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
28661 of floating point to 32-bit integer from doing a direct move to
28662 the GPR registers to do a store.
28663 (fctiwz_<mode>): Break long line.
28665 2017-05-05 Bin Cheng <bin.cheng@arm.com>
28667 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
28668 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
28669 (addr_list, addr_offset_valid_p): New.
28670 (split_address_groups): Check offset validity with above function.
28671 (gt-tree-ssa-loop-ivopts.h): Include header file.
28673 2017-05-05 Nathan Sidwell <nathan@acm.org>
28675 * config.gcc (arm*-*-*): Add missing 'fi'.
28677 2017-05-05 Steve Ellcey <sellcey@cavium.com>
28679 * doc/invoke.texi (-fopt-info): Explicitly say order of options
28680 included in -fopt-info does not matter.
28681 * doc/optinfo.texi (-fopt-info): Fix description of default
28682 behavour. Explicitly say order of options included in -fopt-info
28685 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28687 * config.gcc: Allow combinations of aprofile and rmprofile values for
28688 --with-multilib-list.
28689 * config/arm/t-multilib: New file.
28690 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
28691 variables. Remove setting of ISA and floating-point ABI in
28692 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
28693 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
28694 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
28695 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
28697 * config/arm/t-rmprofile: Likewise except for the matches changes.
28698 * doc/install.texi (--with-multilib-list): Document the combination of
28699 aprofile and rmprofile values and warn about pitfalls in doing that.
28701 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28703 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
28704 (movdi_aarch64): Likewise.
28706 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28708 PR tree-optimization/80632
28709 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
28711 (build_arrays): Initialize it for virtual phis.
28712 (fix_phi_nodes): Use it for virtual phis.
28714 PR tree-optimization/80558
28715 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
28716 [x, y] op z into [x op, y op z] for op & or | if conditions
28719 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
28720 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28723 * config/arm/arm.md (use_literal_pool): Remove.
28724 (64-bit immediate split): No longer takes cost into consideration
28725 if arm_disable_literal_pool is enabled.
28726 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
28727 used when arm_disable_literal_pool is enabled.
28728 (arm_max_const_double_inline_cost): Remove use of
28729 arm_disable_literal_pool.
28730 (push_minipool_fix): Add assert.
28731 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
28732 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
28733 (no_literal_pool_sf_immediate): New.
28735 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28737 PR tree-optimization/80613
28738 * tree-ssa-dce.c (propagate_necessity): Remove cases for
28739 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
28741 2017-05-05 Richard Biener <rguenther@suse.de>
28743 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
28745 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
28747 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
28748 of this flag from insn conditions due to removal from r247495.
28750 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28752 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
28754 (arm_early_store_addr_dep_ptr): Likewise.
28755 * config/arm/aarch-common-protos.h
28756 (arm_early_load_addr_dep_ptr): Add prototype.
28757 (arm_early_store_addr_dep_ptr): Likewise.
28758 * config/arm/cortex-a53.md: Add new bypasses.
28760 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28762 * tree.c (next_type_uid): Change type to unsigned.
28763 (type_hash_canon): Decrement back next_type_uid if
28764 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
28765 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
28768 2017-05-04 Martin Sebor <msebor@redhat.com>
28770 * builtins.c: Fix a trivial typo in a comment.
28772 PR middle-end/79234
28773 * builtins.c (check_sizes): Adjust to handle reading past the end.
28774 Avoid printing excessive upper bound of ranges. Use %E to print
28775 tree nodes instead of converting them to %wu.
28776 (expand_builtin_memchr): New function.
28777 (compute_dest_size): Rename...
28778 (compute_objsize): ...to this.
28779 (expand_builtin_memcpy): Adjust.
28780 (expand_builtin_mempcpy): Adjust.
28781 (expand_builtin_strcat): Adjust.
28782 (expand_builtin_strcpy): Adjust.
28783 (check_strncat_sizes): Adjust.
28784 (expand_builtin_strncat): Adjust.
28785 (expand_builtin_strncpy): Adjust and simplify.
28786 (expand_builtin_memset): Adjust.
28787 (expand_builtin_bzero): Adjust.
28788 (expand_builtin_memcmp): Adjust.
28789 (expand_builtin): Handle memcmp.
28790 (maybe_emit_chk_warning): Check strncat just once.
28792 2017-05-04 Martin Sebor <msebor@redhat.com>
28794 PR preprocessor/79214
28795 PR middle-end/79222
28796 PR middle-end/79223
28797 * builtins.c (check_sizes): Add inlining context and issue
28798 warnings even when -Wno-system-headers is set.
28799 (check_strncat_sizes): Same.
28800 (expand_builtin_strncat): Same.
28801 (expand_builtin_memmove): New function.
28802 (expand_builtin_stpncpy): Same.
28803 (expand_builtin): Handle memmove and stpncpy.
28805 2017-05-04 Bin Cheng <bin.cheng@arm.com>
28807 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
28808 which is not used any more.
28810 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28812 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
28814 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28816 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
28817 (cortexa53_tunings): Likewise.
28818 (cortexa57_tunings): Likewise.
28819 (cortexa72_tunings): Likewise.
28820 (cortexa73_tunings): Likewise.
28822 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28824 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
28825 Set loop alignment to 8.
28827 2017-05-04 Martin Sebor <msebor@redhat.com>
28829 PR translation/80280
28830 * builtins.c (expand_builtin_object_size): Add missing quoting to
28831 %D and like directives.
28832 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
28833 (hsa_type_for_tree_type): Same.
28834 (verify_function_arguments): Same.
28835 * symtab.c (symbol_table::change_decl_assembler_name): Same.
28836 * varasm.c (get_section): Same.
28839 2017-05-04 Martin Sebor <msebor@redhat.com>
28841 PR translation/80280
28842 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
28844 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28846 * config/aarch64/aarch64.c (generic_addrcost_table):
28847 Change HI/TI mode setting.
28849 2017-05-04 Martin Jambor <mjambor@suse.cz>
28851 PR tree-optimization/80622
28852 * tree-sra.c (comes_initialized_p): New function.
28853 (build_accesses_from_assign): Only set write lazily when
28854 comes_initialized_p is false.
28855 (analyze_access_subtree): Use comes_initialized_p.
28856 (propagate_subaccesses_across_link): Assert !comes_initialized_p
28857 instead of testing for PARM_DECL.
28859 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28861 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
28862 constraint on operand 0 to allow more general addressing modes.
28863 Adjust output template.
28864 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
28866 * config/aarch64/aarch64-protos.h
28867 (aarch64_address_valid_for_prefetch_p): Declare prototype.
28868 * config/aarch64/constraints.md (Dp): New address constraint.
28869 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
28872 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
28874 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
28875 update use of estimate_ipcp_clone_size_and_time.
28876 (estimate_local_effects): Update use of
28877 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
28878 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
28879 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
28880 Return nonspecialized time.
28882 2017-05-04 Richard Biener <rguenther@suse.de>
28884 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
28885 for the last VUSE which def dominates the PHI. Directly call
28887 (get_continuation_for_phi_1): Remove.
28889 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
28891 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
28892 to explain the use of truncating division. Cap the number of
28893 iterations to the maximum given by nb_iterations_upper_bound,
28896 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28898 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
28899 * configure: Regenerate.
28900 * config.in: Regenerate.
28901 * config/i386/driver-mingw32.c: new file.
28902 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
28903 * config.host: Link driver-mingw32.o on MinGW host.
28904 * doc/install.texi: Document new --enable-mingw-wildcard configure
28907 2017-05-04 Marek Polacek <polacek@redhat.com>
28909 PR tree-optimization/80612
28910 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
28912 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28913 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
28915 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
28916 (movt splitter): Likewise.
28917 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
28918 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
28919 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
28920 block for Thumb-1 with MOVT.
28921 (thumb2_legitimate_address_p): Move code block ...
28922 (can_avoid_literal_pool_for_label_p): ... into this new function.
28923 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
28925 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
28926 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
28927 "M-profile targets with the MOVT instruction".
28929 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28931 * config/arm/arm-builtins.c (arm_init_builtins): Rename
28932 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
28933 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
28935 2017-05-04 Martin Liska <mliska@suse.cz>
28937 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
28938 variable cond_code.
28940 2017-05-04 Richard Biener <rguenther@suse.de>
28942 * tree.c (array_at_struct_end_p): Handle arrays at struct
28943 end with flexarrays more conservatively. Refactor and treat
28944 arrays of arrays or aggregates more strict. Fix
28945 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
28946 * tree.c (array_at_struct_end_p): Adjust prototype.
28947 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
28948 * gimple-fold.c (get_range_strlen): Likewise.
28949 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
28951 2017-05-04 Richard Biener <rguenther@suse.de>
28953 PR tree-optimization/31130
28954 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
28956 (supports_overflow_infinity): Likewise.
28957 (is_negative_overflow_infinity): Likewise.
28958 (is_positive_overflow_infinity): Likewise.
28959 (is_overflow_infinity): Likewise.
28960 (stmt_overflow_infinity): Likewise.
28961 (overflow_infinity_range_p): Likewise.
28962 (usable_range_p): Remove as always returning true.
28963 (make_overflow_infinity): Remove.
28964 (negative_overflow_infinity): Likewise.
28965 (positive_overflow_infinity): Likewise.
28966 (avoid_overflow_infinity): Likewise.
28967 (set_value_range): Adjust accordingly.
28968 (set_value_range_to_nonnegative): Likewise, remove now unused
28969 overflow_infinity arg.
28970 (vrp_operand_equal_p): Adjust.
28971 (update_value_range): Likewise.
28972 (range_int_cst_singleton_p): Likewise.
28973 (operand_less_p): Likewise.
28974 (compare_values_warnv): Likewise.
28975 (extract_range_for_var_from_comparison_expr): Likewise.
28976 (vrp_int_const_binop): Likewise.
28977 (zero_nonzero_bits_from_vr): Likewise.
28978 (extract_range_from_multiplicative_op_1): Likewise.
28979 (extract_range_from_binary_expr_1): Likewise.
28980 (extract_range_from_unary_expr): Likewise.
28981 (extract_range_from_comparison): Likewise.
28982 (extract_range_basic): Likewise.
28983 (adjust_range_with_scev): Likewise.
28984 (compare_ranges): Likewise.
28985 (compare_range_with_value): Likewise.
28986 (dump_value_range): Likewise.
28987 (test_for_singularity): Likewise, remove strict_overflow_p parameter
28989 (simplify_cond_using_ranges): Adjust.
28991 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28993 * brig-builtins.def: Added a builtin for class_f64.
28994 * builtin-types.def: Added a builtin type needed by class_f64.
28996 2017-05-03 Jason Merrill <jason@redhat.com>
28998 * timevar.def: Add TV_CONSTEXPR.
29000 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29002 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29004 2017-05-03 Martin Jambor <mjambor@suse.cz>
29006 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29007 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29008 * ipa-cp.c (ipcp_propagate_stage): Do not call
29009 ipa_update_after_lto_read.
29010 * ipa-inline.c (ipa_inline): Likewise.
29012 2017-05-03 Martin Jambor <mjambor@suse.cz>
29014 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29015 tag. Added a default constructor and a destructor.
29016 (ipa_edge_args_sum_t): New class;
29017 (ipa_edge_args_sum): Declare.
29018 (ipa_edge_args_vector): Remove declaration.
29019 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29020 (ipa_free_edge_args_substructures): Remove declaration.
29021 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29022 (ipa_edge_args_info_available_for_edge_p): Likewise.
29023 * ipa-prop.c (ipa_edge_args_vector): Removed.
29024 (edge_removal_hook_holder): Likewise.
29025 (edge_duplication_hook_holder): Likewise.
29026 (ipa_edge_args_sum): New variable.
29027 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29028 ipa_edge_args_vector.
29029 (ipa_free_edge_args_substructures): Likewise.
29030 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29031 ipa_edge_args_vector.
29032 (ipa_edge_removal_hook): Turned into method
29033 ipa_edge_args_sum_t::remove.
29034 (ipa_edge_duplication_hook): Turned into method
29035 ipa_edge_args_sum_t::duplicate.
29036 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29037 registering edge hooks.
29038 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29039 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29040 ipa_edge_args_sum instead of ipa_edge_args_vector.
29041 * ipa-profile.c (ipa_profile): Likewise.
29043 2017-05-03 Martin Jambor <mjambor@suse.cz>
29045 * symbol-summary.h (function_summary): New method exists.
29046 (function_summary::symtab_removal): Deallocate through release.
29047 (call_summary): New class.
29048 (gt_ggc_mx): New overload.
29049 (gt_pch_nx): Likewise.
29050 (gt_pch_nx): Likewise.
29052 2017-05-03 Jeff Law <law@redhat.com>
29054 PR tree-optimization/78496
29055 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29056 from simplify_cond_using_ranges. Split off code to walk
29057 backwards through casts into ...
29058 (simplify_cond_using_ranges_2): New function.
29059 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29060 (execute_vrp): After identifying jump threads, call
29061 simplify_cond_using_ranges_2.
29063 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29066 * ipa-inline.h (inline_summary): Add ctor.
29067 (create_ggc): Do not use ggc_cleared_alloc.
29069 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29070 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29072 * gcc.c (handle_braces): Support escaping in switch matching
29074 * doc/invoke.texi (Spec Files): Document it.
29075 Remove superfluous @code markup in items.
29077 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29079 * diagnostic-show-locus.c (struct column_range): New struct.
29080 (get_affected_columns): New function.
29081 (get_printed_columns): New function.
29082 (struct correction): New struct.
29083 (correction::ensure_capacity): New function.
29084 (correction::ensure_terminated): New function.
29085 (struct line_corrections): New struct.
29086 (line_corrections::~line_corrections): New dtor.
29087 (line_corrections::add_hint): New function.
29088 (layout::print_trailing_fixits): Reimplement in terms of the new
29090 (selftest::test_overlapped_fixit_printing): New function.
29091 (selftest::diagnostic_show_locus_c_tests): Call it.
29093 2017-05-03 Nathan Sidwell <nathan@acm.org>
29095 Canonicalize canonical type hashing
29096 * tree.h (type_hash_canon_hash): Declare.
29097 * tree.c (type_hash_list, attribute_hash_list): Move into
29098 type_hash_canon_hash.
29099 (build_type_attribute_qual_variant): Break out hash code calc into
29100 type_hash_canon_hash.
29101 (type_hash_canon_hash): New. Generic type hash computation.
29102 (build_range_type_1, build_array_type_1, build_function_type,
29103 build_method_type_directly, build_offset_type, build_complex_type,
29104 make_vector_type): Call it.
29106 2017-05-03 Richard Biener <rguenther@suse.de>
29108 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29109 When all DRs have unknown misaligned do not always peel
29110 when there is a store but apply the same costing model as if
29111 there were only loads.
29113 2017-05-03 Richard Biener <rguenther@suse.de>
29116 PR tree-optimization/80492
29117 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29118 compare_base_decls returning dont-know properly.
29120 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29122 * config/arm/iterators.md (CCSI): New mode iterator.
29123 (arch): New mode attribute.
29124 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29125 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29126 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29127 code iterator for success result mode.
29128 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29129 the corresponding new insn generators.
29131 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29134 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29135 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29137 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29139 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29140 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29141 (DDR_A): Wrap DDR argument in brackets.
29142 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29143 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29144 (DDR_REVERSED_P): Likewise.
29146 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29148 PR tree-optimization/79472
29149 * tree-switch-conversion.c (struct switch_conv_info): Add
29150 contiguous_range and default_case_nonstandard fields.
29151 (collect_switch_conv_info): Compute contiguous_range and
29152 default_case_nonstandard fields, don't clear final_bb if
29153 contiguous_range and only the default case doesn't have the required
29155 (check_all_empty_except_final): Set default_case_nonstandard instead
29156 of failing if contiguous_range and the default case doesn't have empty
29158 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29159 and only the default case doesn't have the required constants. Skip
29161 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29162 if default_case_nonstandard.
29163 (build_constructors): Build constant 1 just once. Assert that default
29164 values aren't inserted in between cases if contiguous_range. Skip
29166 (build_arrays): Skip virtual phis.
29167 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29168 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29169 Handle virtual phis.
29170 (gen_inbound_check): Handle default_case_nonstandard case.
29171 (process_switch): Adjust check_final_bb caller. Call
29172 gather_default_values with the first non-default case instead of
29173 default case if default_case_nonstandard.
29175 2017-05-02 Nathan Sidwell <nathan@acm.org>
29177 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29178 check. Fix formatting.
29180 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29182 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29183 errors when comparing specialized and unspecialized times.
29185 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29187 * diagnostic-show-locus.c
29188 (layout::should_print_annotation_line_p): Make private.
29189 (layout::print_annotation_line): Make private.
29190 (layout::annotation_line_showed_range_p): Make private.
29191 (layout::show_ruler): Make private.
29192 (layout::print_source_line): Make private. Pass in line and
29193 line_width, rather than calling location_get_source_line. Drop
29195 (layout::print_leading_fixits): New method.
29196 (layout::print_any_fixits): Rename to...
29197 (layout::print_trailing_fixits): ...this, and make private.
29198 Don't print newline fixits.
29199 (diagnostic_show_locus): Move logic for printing one row into...
29200 (layout::print_line): ...this new function. Move the
29201 location_get_source_line call and error-handling from
29202 print_source_line to here. Call print_leading_fixits, and rename
29203 print_any_fixits to print_trailing_fixits.
29204 (selftest::test_fixit_insert_containing_newline): Update now that
29205 newlines are partially supported.
29206 (selftest::test_fixit_insert_containing_newline_2): New test.
29207 (selftest::test_fixit_replace_containing_newline): Update comments.
29208 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29209 * edit-context.c (class added_line): New class.
29210 (class edited_line): Describe newline handling in comment.
29211 (edited_line::actually_edited_p): New method.
29212 (edited_line::print_content): Delete redundant decl.
29213 (edited_line::m_predecessors): New field.
29214 (edited_file::print_content): Call edited_line::print_content.
29215 (edited_file::print_diff): Update to support newlines.
29216 (edited_file::print_diff_hunk): Likewise.
29217 (edited_file::print_run_of_changed_lines): New function.
29218 (edited_file::print_diff_line): Convert to...
29219 (print_diff_line): ...this.
29220 (edited_file::get_effective_line_count): New function.
29221 (edited_line::edited_line): Initialize new field m_predecessors.
29222 (edited_line::~edited_line): Clean up m_predecessors.
29223 (edited_line::apply_fixit): Handle newlines.
29224 (edited_line::get_effective_line_count): New function.
29225 (edited_line::print_content): New function.
29226 (edited_line::print_diff_lines): New function.
29227 (selftest::test_applying_fixits_insert_containing_newline): New
29229 (selftest::test_applying_fixits_replace_containing_newline): New
29231 (selftest::insert_line): New function.
29232 (selftest::test_applying_fixits_multiple_lines): Add example of
29234 (selftest::edit_context_c_tests): Call the new tests.
29236 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29238 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29239 parameter cand. Update dump information.
29240 (get_computation_cost): Update uses.
29242 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29244 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29245 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29246 (get_computation_at, rewrite_use_address): Update use of
29247 get_computation_aff.
29249 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29251 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29252 (get_computation): Delete.
29253 (get_computation_cost): Implement like get_computation_cost_at.
29254 Use get_computation_at.
29255 (get_computation_cost_at): Delete.
29256 (rewrite_use_nonlinear_expr): Use get_computation_at.
29257 (rewrite_use_compare, remove_unused_ivs): Ditto.
29259 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29261 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29263 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29265 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29266 (ivopts_global_cost_for_size): Rename parameter and update uses.
29267 (iv_ca_recount_cost): Update uses.
29268 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29269 candidates seperately in n_invs and n_cands.
29270 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29272 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29274 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29275 (find_inv_vars_cb): New.
29276 (find_depends): Renamed to ...
29277 (find_inv_vars): ... this.
29278 (add_candidate_1, force_var_cost): Call find_inv_vars.
29279 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29281 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29283 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29284 inv_vars. Add inv_exprs.
29285 (struct iv_cand): Rename depends_on to inv_vars.
29286 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29287 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29288 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29289 (dump_cand): Dump inv_vars.
29290 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29291 (record_invariant, find_depends, add_candidate_1): Ditto.
29292 (set_group_iv_cost, force_var_cost): Ditto.
29293 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29294 (get_computation_cost_at, get_computation_cost): Ditto.
29295 (determine_group_iv_cost_generic): Ditto.
29296 (determine_group_iv_cost_address): Ditto.
29297 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29298 (determine_group_iv_costs): Ditto.
29299 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29300 (iv_ca_set_remove_invariants): Renamed to ...
29301 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29302 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29303 (iv_ca_set_add_invariants): Renamed to ...
29304 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29305 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29306 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29307 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29308 (create_new_ivs): Remove useless dump.
29310 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29312 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29314 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29315 (iv_ca_set_no_cp, create_new_iv): Ditto.
29317 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29319 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29321 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29323 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29324 function tree_check2.
29326 2017-05-02 Martin Liska <mliska@suse.cz>
29328 * doc/gcov.texi: Add missing preposition.
29329 * gcov.c (function_info::function_info): Properly fill up
29330 all member variables.
29332 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29334 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29336 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29338 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29340 2017-05-02 Martin Liska <mliska@suse.cz>
29343 * lto-streamer-in.c (lto_read_tree_1): Remove
29344 LTO_STREAMER_DEBUG.
29345 * lto-streamer.c (struct tree_hash_entry): Likewise.
29346 (struct tree_entry_hasher): Likewise.
29347 (tree_entry_hasher::hash): Likewise.
29348 (tree_entry_hasher::equal): Likewise.
29349 (lto_streamer_init): Likewise.
29350 (lto_orig_address_map): Likewise.
29351 (lto_orig_address_get): Likewise.
29352 (lto_orig_address_remove): Likewise.
29353 * lto-streamer.h: Likewise.
29354 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29355 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29357 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29359 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29360 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29361 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29362 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29363 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29364 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29365 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29366 (mm_maskz_sub_ss): New intrinsics.
29367 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29368 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29369 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29370 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29371 (__builtin_ia32_subss_mask_round): New builtins.
29372 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29373 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29374 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29376 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29377 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29379 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29382 2017-05-02 Martin Jambor <mjambor@suse.cz>
29384 PR tree-optimization/78687
29385 * tree-sra.c (access): New field parent.
29386 (process_subtree_disqualification): New function.
29387 (disqualify_candidate): Call it.
29388 (build_accesses_from_assign): Reset write flag if creating an
29390 (build_access_subtree): Fill in parent field and also prpagate
29391 down grp_write flag.
29392 (create_artificial_child_access): New parameter set_grp_write, set
29393 grp_write to its value.
29394 (propagate_subaccesses_across_link): Also propagate grp_write flag
29396 (propagate_all_subaccesses): Push the closest parent back to work
29397 queue if add_access_to_work_queue returned true.
29399 2017-05-02 Richard Biener <rguenther@suse.de>
29401 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29402 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29403 -fstrict-overflow documentation.
29404 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29405 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29406 flag_strict_overflow.
29407 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29408 * lto-opts.c (lto_write_options): Do not stream it.
29409 * lto-wrapper.c (merge_and_complain): Do not handle it.
29410 * opts.c (default_options_table): Do not set -fstrict-overflow.
29411 (finish_options): Likewise do not clear it when sanitizing.
29412 * simplify-rtx.c (simplify_const_relational_operation): Do not
29413 test flag_strict_overflow.
29415 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29417 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29418 using enabled attribute.
29419 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29420 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29421 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29422 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29423 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29424 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29425 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29426 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29427 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29428 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29430 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29432 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29434 2017-05-02 Richard Biener <rguenther@suse.de>
29436 PR tree-optimization/80591
29438 2017-04-10 Richard Biener <rguenther@suse.de>
29440 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29443 2017-05-02 Richard Biener <rguenther@suse.de>
29445 PR tree-optimization/80549
29446 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29447 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29448 headers if they do not have a preheader.
29450 2017-05-02 Martin Liska <mliska@suse.cz>
29453 * common.opt: Fix typo.
29454 * doc/invoke.texi: Likewise.
29456 2017-05-01 Jan Beulich <jbeulich@suse.com>
29458 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29459 swapping, add (x,x,m,x,n) alternative.
29461 2017-05-01 Nathan Sidwell <nathan@acm.org>
29463 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29464 unnecessary unadjusted_alignment check.
29466 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29469 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29471 * gimplify.c (gimplify_cilk_detach): New function.
29472 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29473 * tree-core.h: Document EXPR_CILK_SPAWN.
29474 * tree.h (EXPR_CILK_SPAWN): Define.
29476 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29478 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29479 to use new fixit_hint representation, using the "replace" logic.
29480 (get_line_span_for_fixit_hint): Likewise.
29481 (layout::print_any_fixits): Likewise.
29482 (selftest::test_one_liner_many_fixits): Rename to...
29483 (selftest::test_one_liner_many_fixits_1): ...this, and update
29484 comment and expected output to reflect that the multiple fix-it
29485 hints are now consolidated into one insertion.
29486 (selftest::test_one_liner_many_fixits_2): New test.
29487 (selftest::test_diagnostic_show_locus_one_liner): Update for
29489 (selftest::test_fixit_consolidation): Update for fix-it API
29491 * diagnostic.c (print_parseable_fixits): Likewise.
29492 * edit-context.c (edited_line::m_line_events): Convert from
29493 auto_vec <line_event *> to auto_vec <line_event>.
29494 (class line_event): Convert from abstract base class to a concrete
29495 class, taking over the role of replace_event.
29496 (class insert_event): Delete.
29497 (class replace_event): Rename to class line_event. Convert to
29499 (edit_context::add_fixits): Reimplement.
29500 (edit_context::apply_insert): Delete.
29501 (edit_context::apply_replace): Rename to...
29502 (edit_context::apply_fixit): ...this. Convert to half-open range.
29503 (edited_file::apply_insert): Delete.
29504 (edited_file::apply_replace): Rename to...
29505 (edited_file::apply_fixit): ...this.
29506 (edited_line::~edited_line): Drop deletion of events.
29507 (edited_line::apply_insert): Delete.
29508 (edited_line::apply_replace): Rename to...
29509 (edited_line::apply_fixit): ...this. Convert to half-open range.
29510 Update for change to type of m_line_events.
29511 * edit-context.h (edit_context::apply_insert): Delete.
29512 (edit_context::apply_replace): Rename to...
29513 (edit_context::apply_fixit): ...this.
29515 2017-05-01 Martin Sebor <msebor@redhat.com>
29517 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29520 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29523 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29524 __get_cpuid_max returns 0.
29525 (__get_cpuid_count): Ditto.
29527 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29529 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29530 replacement expression is another instance of one of its arguments.
29532 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29535 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29536 check for stack push/pop autoinc.
29537 * config/i386/i386.c (ix86_agi_dependent): Return false
29538 if the only reason why modified_in_p returned true is that
29539 addr is SP based and set_insn is a push or pop.
29541 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29543 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29546 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29549 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29550 (account_size_time): Use two predicates - exec_pred and
29552 (evaluate_conditions_for_known_args): Compute both clause and
29554 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29555 (inline_summary_t::duplicate): Update.
29556 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29558 (compute_inline_parameters): Likewise.
29559 (estimate_edge_size_and_time): Update caluclation of time.
29560 (estimate_node_size_and_time): Compute both time and nonspecialized
29562 (estimate_ipcp_clone_size_and_time): Update.
29563 (inline_merge_summary): Update.
29564 (do_estimate_edge_time): Update.
29565 (do_estimate_edge_size): Update.
29566 (do_estimate_edge_hints): Update.
29567 (inline_read_section, inline_write_summary): Stream both new predicates.
29568 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29570 (compute_inlined_call_time): Cleanup.
29571 (big_speedup_p): Update.
29572 (edge_badness): Update.
29573 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29574 (size_time_entry): Replace predicate by exec_predicate and
29575 nonconst_predicate.
29576 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29577 (estimate_edge_time): Return also nonspec_time.
29578 (reset_edge_growth_cache): Update.
29580 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29582 PR rtl-optimization/80491
29583 * ifcvt.c (noce_process_if_block): When looking for x setter
29584 with missing else_bb, don't check only the insn right before
29585 cond_earliest, but look for the last insn that x is modified in
29586 within the same bb.
29588 PR rtl-optimization/80491
29589 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29591 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29593 PR tree-optimization/80487
29594 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
29596 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29598 PR tree-optimization/79697
29599 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
29600 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
29601 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
29603 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
29604 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
29606 2017-04-28 Martin Sebor <msebor@redhat.com>
29608 PR tree-optimization/80523
29609 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
29610 (init_target_to_host_charmap, target_to_host, target_strtol10): New
29612 (maybe_warn, format_directive, parse_directive): Use new functions.
29613 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
29615 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
29617 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
29619 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
29621 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
29622 target_header_dir): Set correctly.
29623 * configure: Regenerated.
29624 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29625 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29626 instead of SYSTEM_HEADER_DIR.
29628 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
29630 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
29631 (estimate_local_effects): Likewise.
29632 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
29633 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
29634 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
29635 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
29636 do_estimate_edge_time, estimate_edge_time): Likewise.
29637 * ipa-inline-analysis.c (estimate_node_size_and_time,
29638 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
29639 (estimate_time_after_inlining): Remove.
29641 2017-04-28 Martin Liska <mliska@suse.cz>
29643 * doc/gcov.texi: Enhance documentation of gcov.
29645 2017-04-28 Martin Liska <mliska@suse.cz>
29647 * doc/gcov.texi: Sort options in alphabetic order.
29648 * doc/gcov-dump.texi: Likewise.
29649 * doc/gcov-tool.texi: Likewise.
29650 * gcov.c (print_usage): Likewise.
29651 * gcov-dump.c (print_usage): Likewise.
29652 * gcov-tool.c (print_merge_usage_message): Likewise.
29653 (print_rewrite_usage_message): Likewise.
29654 (print_overlap_usage_message): Likewise.
29656 2017-04-28 Martin Liska <mliska@suse.cz>
29658 PR gcov-profile/53915
29659 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
29661 2017-04-28 Martin Liska <mliska@suse.cz>
29663 PR gcov-profile/79891
29664 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
29665 is marked by compiler as living on a line.
29666 (get_cycles_count): Remove usage of the union.
29667 (output_intermediate_file): Likewise.
29668 (find_source): Fix GNU coding style.
29669 (accumulate_line_counts): Remove old non-all block mode.
29670 (output_lines): Remove usage of the union.
29671 * profile.c (output_location): Include all BBs, even if
29672 belonging to a same line (and file) as a previous BB.
29674 2017-04-28 Martin Liska <mliska@suse.cz>
29676 * gcov.c (process_args): Handle new argument 'w'.
29677 (read_graph_file): Assign ID to BBs.
29678 (output_branch_count): Display BB # if verbose flag is set.
29679 (output_lines): Likewise for arcs.
29680 (print_usage): Add '--verbose' option help.
29681 * doc/gcov.texi: Document --verbose (-w) option.
29683 2017-04-28 Martin Liska <mliska@suse.cz>
29685 * gcov.c (struct block_location_info): New struct.
29686 (process_file): Fill up the new structure.
29687 (read_graph_file): Replace usage of encoding by the newly added
29689 (add_line_counts): Likewise.
29690 (accumulate_line_counts): Remove usage of the union.
29691 (function_info::function_info): New function.
29692 (function_info::~function_info): Likewise.
29693 (process_file): Call delete instead of release_function.
29694 (release_function): Release the function.
29695 (release_structures): Call delete instead of release_function.
29696 (solve_flow_graph): Replace usage of num_blocks.
29697 (find_exception_blocks): Likewise.
29698 (output_lines): Fix GNU coding style.
29700 2017-04-28 Martin Liska <mliska@suse.cz>
29703 * coverage.c (coverage_remove_note_file): New function.
29704 * coverage.h: Declare the function.
29705 * toplev.c (finalize): Clean if an error has been seen.
29707 2017-04-28 Martin Liska <mliska@suse.cz>
29709 PR gcov-profile/80031
29710 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
29711 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
29712 * gcov.c (read_graph_file): Read just number of blocks.
29713 * profile.c (branch_prob): Do not stream 0 flags per a basic
29716 2017-04-28 Martin Liska <mliska@suse.cz>
29718 * gcov-dump.c (tag_*): Add new argument to declarations.
29719 (dump_gcov_file): Likewise.
29720 (tag_blocks): Add and use new argument depth.
29721 (tag_arcs): Likewise.
29722 (tag_lines): Likewise.
29723 (tag_counters): Likewise.
29724 (tag_summary): Likewise.
29725 (dump_working_sets): Use depth to do a proper indentation.
29727 2017-04-28 Jakub Jelinek <jakub@redhat.com>
29730 * cgraph.h (symtab_node::debug_symtab): No longer inline.
29731 * symtab.c (symtab_node::debug_symtab): Move definition here.
29733 2017-04-28 Richard Biener <rguenther@suse.de>
29735 * lto-streamer.h (LTO_major_version): Bump to 7.
29737 2017-04-28 Richard Biener <rguenther@suse.de>
29739 * tree-vrp.c (assert_info): New struct.
29740 (add_assert_info): New helper.
29741 (register_edge_assert_for_2): Refactor to add asserts to a vector
29743 (register_edge_assert_for_1): Likewise.
29744 (register_edge_assert_for): Likewise.
29745 (finish_register_edge_assert_for): New helper actually registering
29746 asserts where live on edge.
29747 (find_conditional_asserts): Adjust.
29748 (find_switch_asserts): Likewise.
29749 (evrp_dom_walker::try_find_new_range): Generalize.
29750 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
29752 2017-04-27 Marek Polacek <polacek@redhat.com>
29755 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
29756 arg10 and arg11 to itype.
29758 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
29760 * doc/extend.texi (Object Size Checking): Improve grammar.
29762 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
29765 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
29766 that the logic for permitting reciprocal estimates matches that
29769 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29772 * tree.c (type_cache_hasher::equal): Only compare
29773 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
29774 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
29775 non-aggregate element types.
29776 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
29777 about the flag on ARRAY_TYPEs in the comment, formatting fix.
29779 2017-04-27 Richard Biener <rguenther@suse.de>
29781 PR middle-end/80533
29782 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
29783 stripping ARRAY_REFs from MEM_EXPR make sure we're not
29784 keeping a reference to a trailing array.
29786 2017-04-27 Richard Biener <rguenther@suse.de>
29788 PR middle-end/80539
29789 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
29790 being in loop-closed SSA form conservatively.
29791 (chrec_fold_multiply_poly_poly): Likewise.
29793 2017-04-27 Tamar Christina <tamar.christina@arm.com>
29795 PR middle-end/79665
29796 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
29797 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
29799 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29802 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
29803 (aarch64_function_arg_alignment): Return unsigned int again, but still
29804 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
29805 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29806 Don't emit -Wpsabi note.
29807 (aarch64_function_arg_boundary): Likewise.
29808 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29811 2017-04-26 Nathan Sidwell <nathan@acm.org>
29813 * tree.h (crc32_unsigned_n): Declare.
29814 (crc32_unsigned, crc32_unsigned): Make inline.
29815 * tree.c (crc32_unsigned_bits): Replace with ...
29816 (crc32_unsigned_n): ... this.
29817 (crc32_unsigned, crc32_byte): Remove.
29818 (crc32_string): Remove unnecessary braces.
29820 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29822 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
29823 * ipa-inline-analysis.c (MAX_TIME): Remove.
29824 (account_size_time): Use sreal for time.
29825 (dump_inline_summary): Update.
29826 (estimate_function_body_sizes): Update.
29827 (estimate_edge_size_and_time): Update.
29828 (estimate_calls_size_and_time): Update.
29829 (estimate_node_size_and_time): Update.
29830 (inline_merge_summary): Update.
29831 (inline_update_overall_summary): Update.
29832 (estimate_time_after_inlining): Update.
29833 (inline_read_section): Update.
29834 (inline_write_summary): Update.
29835 * ipa-inline.c (compute_uninlined_call_time): Update.
29836 (compute_inlined_call_time): Update.
29837 (recursive_inlining): Update.
29838 (inline_small_functions): Update.
29839 (dump_overall_stats): Update.
29840 * ipa-inline.h: Include sreal.h.
29841 (size_time_entry): Turn time to sreal.
29842 (inline_summary): Turn self_time nad time to sreal.
29844 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29846 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
29848 (sreal::stream_out, sreal::stream_in): New.
29849 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
29851 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29853 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
29856 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
29859 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
29861 (dimode_scalar_chain::compute_convert_gain): Ditto.
29862 (dimode_scalar_chain::make_vector_copies): Ditto.
29863 (dimode_scalar_chain::convert_reg): Ditto.
29864 (dimode_scalar_chain::convert_insn): Ditto.
29865 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
29866 (VI248_AVX512BW_1): New mode iterator.
29867 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
29868 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
29871 2017-04-25 Martin Sebor <msebor@redhat.com>
29873 PR tree-optimization/80497
29874 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
29875 constants are representable in HOST_WIDE_INT.
29876 (parse_directive): Ditto.
29878 2017-04-25 Martin Sebor <msebor@redhat.com>
29881 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
29882 (new_zero_array): Adjust signature.
29883 (dom_info::dom_init): Used unsigned rather that size_t.
29884 (dom_info::dom_info): Same.
29886 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29887 Jakub Jelinek <jakub@redhat.com>
29890 * config/arm/arm.c: Include gimple.h.
29891 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29892 returns negative, increment ncrn only if it returned positive.
29893 (arm_needs_doubleword_align): Return int instead of bool,
29894 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
29895 members, but if there is any such non-FIELD_DECL
29896 > PARM_BOUNDARY aligned decl, return -1 instead of false.
29897 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29898 returns negative, increment nregs only if it returned positive.
29899 (arm_setup_incoming_varargs): Likewise.
29900 (arm_function_arg_boundary): Emit -Wpsabi note if
29901 arm_needs_doubleword_align returns negative, return
29902 DOUBLEWORD_ALIGNMENT only if it returned positive.
29904 2017-04-25 Marek Polacek <polacek@redhat.com>
29907 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29908 first argument to type.
29910 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
29913 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
29914 type checks to test for compatibility instead of equality.
29916 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29917 Jakub Jelinek <jakub@redhat.com>
29920 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
29922 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
29923 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
29924 the alignment computation, but return their maximum in warn_alignment.
29925 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29926 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
29928 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
29929 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29932 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29934 * config/arc/simdext.md (dmpyh): Fix typo.
29936 2017-04-25 Richard Biener <rguenther@suse.de>
29938 PR tree-optimization/80492
29939 * alias.c (compare_base_decls): Handle registers with asm
29940 specification conservatively.
29941 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29942 compare_base_decls returning dont-know properly.
29944 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29946 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
29947 (legitimate_offset_address_p): New function.
29948 (arc_legitimate_address_p): Use above function.
29950 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29952 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
29954 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29956 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
29957 ACCH registers whenever they are available.
29959 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29961 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
29962 double regs fix when not used.
29964 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29966 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
29968 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
29969 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
29971 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29973 * config/arc/arc.c (arc_output_addsi): Check for h-register class
29974 when emitting short ADD instructions.
29976 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29978 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
29980 (cmpsi_cc_c_insn): Likewise.
29981 (cbranchsi4_scratch): Compute proper instruction length using
29982 compact_hreg_operand.
29983 * config/arc/predicates.md (compact_hreg_operand): New predicate.
29985 2017-04-25 Richard Biener <rguenther@suse.de>
29987 PR middle-end/80509
29988 * passes.c (pass_manager::pass_manager): Initialize
29989 m_name_to_pass_map.
29991 2017-04-25 Richard Biener <rguenther@suse.de>
29993 PR tree-optimization/79201
29994 * tree-ssa-sink.c (statement_sink_location): Handle calls.
29996 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29999 * config/s390/vector.md: Split MEM->GPR vector moves for
30000 non-s_operand addresses.
30002 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30005 * config/s390/predicates.md (reload_const_wide_int_operand): New
30007 * config/s390/s390.md ("movti"): Remove d/P alternative.
30008 ("movti_bigconst"): New pattern definition.
30010 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30013 * s390-protos.h (s390_expand_cs_hqi): Removed.
30014 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30015 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30016 modes as well as CCZ1mode and CCZmode.
30017 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30018 signature of s390_emit_compare_and_swap.
30019 (s390_expand_cs_hqi): Likewise, make static.
30020 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30021 compare-and-swap, in some cases.
30022 (s390_expand_cs): Wrapper function.
30023 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30025 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30026 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30027 patterns for small and large integers. Forbid symref memory operands.
30028 Move expander to s390.c. Require cc register.
30029 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30030 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30031 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30032 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30033 symref memory operands. Remove CC mode and call s390_match_ccmode
30035 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30037 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30039 * config/s390/s390.md (define_peephole2): New peephole to help
30040 combining the load-and-test pattern with volatile memory.
30042 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30044 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30045 with CCZmode for TARGET_Z196.
30047 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30049 PR rtl-optimization/80501
30050 * combine.c (make_compound_operation_int): Set subreg_code to SET
30051 even for AND with mask of the sign bit of mode.
30053 PR rtl-optimization/80500
30054 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30055 sum's initial value.
30057 2017-04-25 Julian Brown <julian@codesourcery.com>
30058 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30060 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30062 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30064 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30066 2017-04-25 Julian Brown <julian@codesourcery.com>
30067 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30069 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30070 (thunderx2t99_sha): New Reservation.
30072 2017-04-25 Julian Brown <julian@codesourcery.com>
30073 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30075 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30076 type for 1-element load.
30078 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30080 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30082 2017-04-24 Martin Jambor <mjambor@suse.cz>
30084 PR tree-optimization/80293
30085 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30086 char arrays not totally scalarizable if it is false.
30087 (analyze_all_variable_accesses): Pass correct value in the new
30088 parameter. Add a statistics counter.
30090 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30092 PR middle-end/79931
30093 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30095 2017-04-24 Richard Biener <rguenther@suse.de>
30097 PR tree-optimization/80494
30098 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30099 out for complex types.
30101 2017-04-24 Richard Biener <rguenther@suse.de>
30103 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30104 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30105 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30106 (DFS): Adjust and never fail.
30107 (sccvn_dom_walker::fail): Remove.
30108 (sccvn_dom_walker::before_dom_children): Adjust.
30109 (run_scc_vn): Likewise and never fail.
30110 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30111 (pass_fre::execute): Likewise.
30113 2017-04-24 Richard Biener <rguenther@suse.de>
30115 PR tree-optimization/79725
30116 * tree-ssa-sink.c (statement_sink_location): Return whether
30117 failure reason was zero uses. Move that check later.
30118 (sink_code_in_bb): Deal with zero uses by removing the stmt
30121 2017-04-24 Richard Biener <rguenther@suse.de>
30124 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30125 pointer-based references.
30127 2017-04-24 Richard Biener <rguenther@suse.de>
30130 * pass_manager.h (pass_manager::operator new): Remove.
30131 (pass_manager::operator delete): Likewise.
30132 * passes.c (pass_manager::operator new): Remove.
30133 (pass_manager::operator delete): Likewise.
30134 (pass_manager::pass_manager): Zero individual pass members.
30136 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30139 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30140 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30141 Check "XEXP (src, 1)" operand here.
30142 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30143 Check "XEXP (src, 1)" operand here.
30144 (dimode_scalar_chain::make_vector_copies): Detect count register
30145 of a shift instruction. Zero extend count register from QImode
30146 to DImode to satisfy vector shift pattern count operand predicate.
30147 Substitute vector shift count operand with a DImode copy.
30148 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30151 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30153 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30154 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30155 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30156 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30157 (UNSPEC_NOREX_MEM): Remove definition.
30159 2017-04-21 Richard Biener <rguenther@suse.de>
30161 PR tree-optimization/79547
30162 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30163 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30164 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30165 without any constraints.
30167 2017-04-21 Richard Biener <rguenther@suse.de>
30169 PR tree-optimization/78847
30170 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30172 2017-04-21 Richard Biener <rguenther@suse.de>
30174 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30175 (build_distinct_type_copy): Likewise.
30176 (build_variant_type_copy): Likewise.
30177 * tree.c (build_qualified_type): Pass down mem-stat info.
30178 (build_distinct_type_copy): Likewise.
30179 (build_variant_type_copy): Likewise.
30181 2017-04-21 Richard Biener <rguenther@suse.de>
30183 PR tree-optimization/80237
30184 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30186 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30187 for a simplified result.
30189 2016-04-21 Richard Biener <rguenther@suse.de>
30191 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30192 sth as strict as a simple_iv but a chrec without symbols and an
30193 operand defined in the loop we are peeling (and not some subloop).
30194 (propagate_constants_for_unrolling): Propagate all constants.
30196 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30199 * config/i386/i386.c (print_reg): Remove assert for disalowed
30200 regno values, call output_operand_lossage instead.
30202 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30205 * config/i386/constraints.md (Yc): New register constraint.
30206 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30207 Use Yc constraint for alternative 2 of operand 0. Remove
30208 preferred_for_speed attribute.
30210 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30212 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30213 lastprivate clauses in SIMT case.
30215 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30217 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30219 2017-04-20 Richard Biener <rguenther@suse.de>
30221 PR tree-optimization/57796
30222 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30223 as N scalar stores.
30224 (vect_model_load_cost): Cost gathers as N scalar loads.
30226 2017-04-20 Richard Biener <rguenther@suse.de>
30228 * ggc-page.c (ggc_allocated_p): Rename to ...
30229 (safe_lookup_page_table_entry): ... this and return the lookup
30231 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30233 2017-04-20 Richard Biener <rguenther@suse.de>
30235 PR tree-optimization/80453
30236 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30237 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30238 from the conditions.
30239 (vn_phi_eq): Pass them down.
30240 (vn_phi_lookup): Record them.
30241 (vn_phi_insert): Likewise.
30243 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30245 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30246 uninitialized variable warning to avoid buffer overrun.
30248 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30251 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30252 is suppressed for '{ 0 }' in C.
30254 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30256 * BASE-VER: Set to 8.0.0.
30258 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30260 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30261 priority .init_array and .fini_array section with SECTION_NOTYPE
30264 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30266 PR middle-end/80423
30267 * tree.h (build_array_type): Add typeless_storage default argument.
30268 * tree.c (type_cache_hasher::equal): Also compare
30269 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30270 (build_array_type): Add typeless_storage argument, set
30271 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30273 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30274 (build_array_type): Likewise. Add typeless_storage argument.
30276 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30277 Jakub Jelinek <jakub@redhat.com>
30279 PR tree-optimization/80426
30280 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30281 operation on symbolic operands, also compute the overflow for the
30282 invariant part when the operation degenerates into a negation.
30284 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30287 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30288 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30291 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30293 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30296 * config/avr/avr.c (tree.h): Include it.
30297 (cgraph.h): Include it.
30298 (avr_encode_section_info): Don't warn for uninitialized progmem
30299 variable if it's just an alias.
30301 2017-04-19 Richard Biener <rguenther@suse.de>
30304 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30305 when needed by AutoPGO.
30307 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30310 * doc/lto.texi: Remove an extra 'that'.
30312 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30314 PR rtl-optimization/80429
30315 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30316 are only used in debug insns.
30318 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30319 Vladimir Makarov <vmakarov@redhat.com>
30321 * config/sparc/predicates.md (input_operand): Add comment. Return
30322 true for any memory operand when LRA is in progress.
30323 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30325 2017-04-18 Jeff Law <law@redhat.com>
30328 * mips.md ({return,simple_return}_internal): Do not overwrite
30331 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30333 PR tree-optimization/80443
30334 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30335 instead of adding 1, subtract -1 and similarly instead of subtracting
30338 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30340 PR rtl-optimization/80357
30341 * haifa-sched.c (tmp_bitmap): New variable.
30342 (model_recompute): Handle duplicate use records.
30343 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30344 (free_global_sched_pressure_data): Free it.
30346 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30349 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30350 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30351 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30352 instead of SYSTEM_HEADER_DIR.
30354 2017-04-18 Jeff Law <law@redhat.com>
30356 PR middle-end/80422
30357 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30358 predecessors after walking up the insn chain.
30360 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30363 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30364 sizetype type into debug info.
30366 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30369 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30370 unneeded test for TARGET_UPPER_REGS_SF.
30371 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30373 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30376 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30377 instead of gsi_after_labels.
30379 2017-04-18 Jeff Law <law@redhat.com>
30381 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30385 2017-04-13 Jeff Law <law@redhat.com>
30386 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30387 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30389 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30392 * config/avr/avr.c (intl.h): Include it.
30393 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30395 2017-04-18 Martin Liska <mliska@suse.cz>
30397 PR gcov-profile/78783
30398 * gcov-tool.c (gcov_output_files): Validate that destination
30399 file is either removed by the tool or by a user.
30401 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30402 Guy Benyei <guybe@mellanox.com>
30404 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30405 block, and do not negate it, the stored id is already negative.
30407 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30409 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30411 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30414 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30415 masks of options that should be turned off if the VSX vector
30416 options are turned off.
30417 (OTHER_P8_VECTOR_MASKS): Likewise.
30418 (OTHER_VSX_VECTOR_MASKS): Likewise.
30419 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30420 rs6000_disable_incompatible_switches to validate no type switches
30422 (rs6000_incompatible_switch): New function to disallow turning on
30423 other vector options if -mno-vsx, -mno-power8-vector, or
30424 -mno-power9-vector are specified.
30426 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30428 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30430 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30432 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30433 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30434 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30435 (ARG_POINTER_CFA_OFFSET): Likewise.
30437 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30439 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30440 conditions to take advantage of various optimizations.
30442 2017-04-13 Jeff Law <law@redhat.com>
30444 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30445 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30446 (zero_extendsidi2_dext): Likewise.
30448 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30451 * fold-const.c (fold_ternary_loc): Revert
30452 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30455 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30457 PR rtl-optimization/80343
30458 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30459 new scratch pseudo.
30461 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30464 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30465 to ubsan_encode_value.
30467 2017-04-13 Jeff Law <law@redhat.com>
30469 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30470 appearing in DEBUG_INSNs.
30472 2017-04-13 Martin Liska <mliska@suse.cz>
30474 PR gcov-profile/80413
30475 * gcov-io.c (gcov_write_string): Copy to buffer just when
30476 allocated size is greater than zero.
30478 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30481 * dwarf2out.c (decls_for_scope): Ignore declarations of
30482 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30484 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30487 * ipa-visibility.c (non_local_p): Fix typos.
30488 (localize_node): When localizing symbol in same comdat group,
30489 dissolve the group only when we know external symbols are going
30491 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30493 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30495 PR tree-optimization/79390
30496 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30497 order does not result in usable sequence, retry with reversed operand
30503 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30504 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30505 op0 instead of fold_convert_loc (loc, type, arg0).
30507 2017-04-12 Jeff Law <law@redhat.com>
30509 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30510 has a delay slot in the generated code.
30512 * config/cris/cris.md (cris_preferred_reload_class): Return
30513 GENNONACR_REGS rather than GENERAL_REGS.
30515 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30518 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30519 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30520 signedness of the result type.
30522 2017-04-12 Richard Biener <rguenther@suse.de>
30523 Jeff Law <law@redhat.com>
30525 PR tree-optimization/80359
30526 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30527 trim stores to TARGET_MEM_REFs.
30529 2017-04-12 Richard Biener <rguenther@suse.de>
30531 PR tree-optimization/79390
30532 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30533 threading case even more.
30535 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30538 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30539 for quad_address_p for TImode, instead of just not indexed_address.
30541 2017-04-12 Richard Biener <rguenther@suse.de>
30542 Bernd Edlinger <bernd.edlinger@hotmail.de>
30544 PR middle-end/79671
30545 * alias.c (component_uses_parent_alias_set_from): Handle
30546 TYPE_TYPELESS_STORAGE.
30547 (get_alias_set): Likewise.
30548 * tree-core.h (tree_type_common): Add typeless_storage flag.
30549 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30550 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30551 for types containing members with TYPE_TYPELESS_STORAGE.
30552 (place_field): Likewise.
30553 (layout_type): Likewise for ARRAY_TYPE.
30554 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30555 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30556 TYPE_TYPELESS_STORAGE.
30557 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30559 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30562 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30563 first argument to type.
30565 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30569 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30570 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30571 (rs6000_expand_binop_builtin): Likewise.
30572 (rs6000_expand_ternop_builtin): Likewise; also add missing
30573 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30574 vshasigma built-ins.
30575 * doc/extend.texi: Document that vec_xxpermdi's third argument
30576 must be a constant.
30578 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30580 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30581 Use shift_const cost parameter when calculating gain of STV shifts.
30583 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30585 PR rtl-optimization/70478
30586 * lra-constraints.c (process_alt_operands): Check memory for
30587 disfavoring memory insn operand.
30589 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30591 PR middle-end/80100
30592 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30593 left shift in unsigned HOST_WIDE_INT type.
30595 PR rtl-optimization/80385
30596 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
30597 (not (neg X)) into (plus X -1) for complex or non-integral modes.
30600 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
30601 if they have any depend clauses.
30603 2017-04-11 Martin Liska <mliska@suse.cz>
30606 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
30607 * ipa-split.c (split_function): Create a local comdat symbol
30608 if caller is in a comdat group.
30610 2017-04-11 Martin Liska <mliska@suse.cz>
30613 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
30616 2017-04-11 Martin Sebor <msebor@redhat.com>
30618 PR middle-end/80364
30619 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
30620 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
30622 (directive::set_width, directive::set_precision, format_character):
30624 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
30627 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
30630 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
30631 conflict, set target->arch_name instead of target->cpu_name.
30633 2017-04-11 Richard Biener <rguenther@suse.de>
30635 PR tree-optimization/80374
30636 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
30637 build_zero_cst, remove fold_convertible_p check again.
30639 2017-04-11 Martin Liska <mliska@suse.cz>
30642 * ubsan.c (instrument_object_size): Do not instrument register
30645 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30648 * config/i386/i386-builtin-types.def
30649 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
30650 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
30651 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
30652 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
30653 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
30654 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
30655 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
30656 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
30657 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
30658 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
30659 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
30660 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
30661 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
30662 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
30663 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
30664 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
30665 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
30666 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
30667 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
30668 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
30669 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
30670 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
30671 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
30672 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
30673 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
30674 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
30675 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
30676 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
30677 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
30678 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
30679 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
30680 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
30681 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
30682 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
30683 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
30684 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
30685 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
30686 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
30687 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
30688 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
30689 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
30690 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
30691 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
30692 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
30693 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
30694 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
30696 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
30697 flag to second_arg_count, handle 4 argument function type _COUNT
30698 aliases, handle second_arg_count on second argument rather than last.
30700 2017-04-10 Jeff Law <law@redhat.com>
30702 PR tree-optimization/80374
30703 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
30704 record anything if we can not convert integer_zero_node to the
30707 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30710 * config/rs6000/rs6000.c (rs6000_option_override_internal):
30711 Enhance special handling given to the TARGET_P9_MINMAX option in
30712 relation to certain other options.
30714 2017-04-10 Bin Cheng <bin.cheng@arm.com>
30716 PR tree-optimization/80153
30717 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
30718 remove POINTER_PLUS_EXPR's base part directly, rather than through
30721 2017-04-10 Richard Biener <rguenther@suse.de>
30722 Bin Cheng <bin.cheng@arm.com>
30724 PR tree-optimization/80153
30725 * tree-affine.c (aff_combination_to_tree): Get base pointer from
30726 the first element of pointer type aff_tree. Build result expr in
30728 (add_elt_to_tree): Convert to type unconditionally. Remove other
30729 fold_convert calls.
30730 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
30731 (rewrite_use_nonlinear_expr): Check invariant using iv information.
30733 2017-04-10 Richard Biener <rguenther@suse.de>
30735 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30738 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
30740 PR rtl-optimization/70478
30741 * lra-constraints.c (curr_small_class_check): New.
30742 (update_and_check_small_class_inputs): New.
30743 (process_alt_operands): Update curr_small_class_check. Disfavor
30744 alternative insn memory operands. Check available regs for small
30747 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30750 * config/mips/mips.opt (-mvirt): Update description.
30751 * doc/invoke.texi (-mvirt): Likewise.
30753 2017-04-10 Richard Biener <rguenther@suse.de>
30755 PR middle-end/80362
30756 * fold-const.c (fold_binary_loc): Look at unstripped ops when
30757 looking for NEGATE_EXPR in -A / -B to A / B folding.
30759 2017-04-10 Martin Liska <mliska@suse.cz>
30761 PR gcov-profile/80224
30762 * gcov.c (print_usage): Fix usage string.
30763 (get_gcov_intermediate_filename): Remove.
30764 (output_gcov_file): Use both for normal and intermediate format.
30765 (generate_results): Do not initialize special file for
30766 intermediate format.
30768 2017-04-10 Richard Biener <rguenther@suse.de>
30770 PR tree-optimization/80304
30771 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
30774 2017-04-10 Nathan Sidwell <nathan@acm.org>
30777 * config/rs6000/rs6000.c (rs6000_vector_type): New.
30778 (rs6000_init_builtins): Use it.
30780 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30782 * config/arm/arm.md (<mrc>): Add mode to SET source.
30783 (<mrrc>): Likewise.
30785 2017-04-10 Richard Biener <rguenther@suse.de>
30787 PR middle-end/80344
30788 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
30790 2017-04-10 Jakub Jelinek <jakub@redhat.com>
30793 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
30794 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
30795 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
30796 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
30797 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
30798 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
30799 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
30800 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
30801 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
30802 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
30803 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
30804 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
30805 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
30806 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
30807 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
30808 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
30809 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
30810 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
30811 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
30812 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
30813 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
30814 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
30815 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
30817 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
30819 PR rtl-optimization/70478
30820 * lra-constraints.c: Reverse the last patch.
30822 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
30824 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
30825 Add comment for WCHAR_T.
30827 2017-04-08 Martin Liska <mliska@suse.cz>
30830 2017-04-07 Martin Liska <mliska@suse.cz>
30833 * ipa-split.c (split_function): Add function part to a same comdat
30836 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30839 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
30841 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
30843 * rs6000/rs6000.c (vec_load_pendulum): Rename...
30844 (vec_pairing): ...to this.
30845 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
30846 (rs6000_sched_init): Adjust for name change.
30847 (struct rs6000_sched_context): Likewise.
30848 (rs6000_init_sched_context): Likewise.
30849 (rs6000_set_sched_context): Likewise.
30851 2017-04-07 Jakub Jelinek <jakub@redhat.com>
30857 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
30859 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
30860 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
30861 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
30863 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
30865 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
30867 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30869 PR rtl-optimization/70703
30870 * ira-color.c (update_conflict_hard_regno_costs): Use
30871 int64_t instead of HOST_WIDE_INT.
30873 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30875 PR rtl-optimization/70478
30876 * lra-constraints.c (process_alt_operands): Disfavor alternative
30877 insn memory operands.
30879 2017-04-07 Jeff Law <law@redhat.com>
30881 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
30882 CALL and NOTE_INSN_CALL_ARG_LOCATION.
30884 2017-04-07 Martin Liska <mliska@suse.cz>
30887 * config/aarch64/aarch64.c (aarch64_process_target_attr):
30888 Show error message instead of an ICE.
30890 2017-04-07 Martin Liska <mliska@suse.cz>
30893 * ipa-split.c (split_function): Add function part to a same comdat
30896 2017-04-07 Richard Biener <rguenther@suse.de>
30898 PR middle-end/80341
30899 * tree.c (get_unwidened): Also handle ! for_type case for
30901 * convert.c (do_narrow): Split out from ...
30902 (convert_to_integer_1): ... here. Do not pass final truncation
30903 type to get_unwidened for TRUNC_DIV_EXPR.
30905 2017-04-07 Richard Biener <rguenther@suse.de>
30907 * tree-affine.c (wide_int_ext_for_comb): Take type rather
30909 (aff_combination_const): Adjust.
30910 (aff_combination_scale): Likewise.
30911 (aff_combination_add_elt): Likewise.
30912 (aff_combination_add_cst): Likewise.
30913 (aff_combination_convert): Likewise.
30914 (add_elt_to_tree): Likewise. Remove unused argument.
30915 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
30917 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
30919 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
30921 * config/arm/arm.c (arm_default_short_enums): Use
30922 ARM_DEFAULT_SHORT_ENUMS.
30923 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
30925 2017-04-06 Jakub Jelinek <jakub@redhat.com>
30928 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
30929 members with redundant out-of-class redeclaration.
30931 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30934 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
30935 * config/i386/i386.md (*zero_extendsidi2):
30936 Add (?*x,*x) and (?*v,*v) alternatives.
30938 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30941 * config/i386/i386.c (ix86_expand_builtin)
30942 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
30943 mode from insn data. Convert operands to insn operand mode.
30944 Copy operands that don't satisfy insn predicate to a register.
30946 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
30948 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
30951 2017-04-06 Richard Biener <rguenther@suse.de>
30953 PR tree-optimization/80334
30954 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
30955 preserve alignment of accesses.
30957 2017-04-06 Richard Biener <rguenther@suse.de>
30959 PR tree-optimization/80262
30960 * tree-sra.c (build_ref_for_offset): Preserve address-space
30962 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
30963 Drop useless address-space information on MEM_REF offsets.
30965 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
30967 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
30969 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30971 PR rtl-optimization/70703
30972 * ira-color.c (update_conflict_hard_regno_costs): Use
30973 HOST_WIDE_INT instead of long.
30975 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
30978 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
30979 not defined for x86_64 target. Add -mmmx target option when __SSE2__
30981 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
30982 for x86_64 target. Handle -m3dnowa option.
30984 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30986 PR rtl-optimization/70703
30987 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
30988 (update_conflict_hard_regno_costs): Use long instead of unsigned
30989 arithmetic for cost calculation.
30991 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30992 Bernd Edlinger <bernd.edlinger@hotmail.de>
30995 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
30998 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31001 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31002 ptr_mode with Pmode throughout.
31003 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31004 into probe_stack_range and use DImode.
31006 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31009 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31010 call_eh_return is true.
31012 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31014 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31015 Initialize last_match_fntype_index.
31017 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31020 * tree-nvr.c: Include internal-fn.h.
31021 (pass_return_slot::execute): Ignore internal calls without
31024 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31025 Richard Biener <rguenther@suse.de>
31028 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31029 captures used multiple times, except for the last use.
31030 * generic-match-head.c: Include gimplify.h.
31032 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31034 PR tree-optimization/79390
31035 * target.h (struct noce_if_info): Declare.
31036 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31037 * target.def (noce_conversion_profitable_p): New target hook.
31038 * ifcvt.h (struct noce_if_info): New type, moved from ...
31039 * ifcvt.c (struct noce_if_info): ... here.
31040 (noce_conversion_profitable_p): Renamed to ...
31041 (default_noce_conversion_profitable_p): ... this. No longer
31043 (noce_try_store_flag_constants, noce_try_addcc,
31044 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31045 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31046 instead of noce_conversion_profitable_p.
31047 * config/i386/i386.c: Include ifcvt.h.
31048 (ix86_option_override_internal): Don't override
31049 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31050 (ix86_noce_conversion_profitable_p): New function.
31051 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31052 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31053 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31054 * doc/tm.texi: Regenerated.
31056 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31058 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31061 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31064 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31065 instructions for small multiply cores.
31067 2017-04-04 Jeff Law <law@redhat.com>
31069 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31071 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31072 that are not set by the loop over the elements.
31074 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31077 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31078 int mode, convert_modes it to mode as unsigned, otherwise use
31079 lowpart_subreg to mode rather than SImode.
31080 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31081 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31082 Use DImode instead of SImode for the shift count operand.
31083 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31086 2017-04-04 Richard Biener <rguenther@suse.de>
31088 PR middle-end/80281
31089 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31090 arithmetic done for the negate or the plus. Simplify.
31091 (A - (-B) -> A + B): Likewise.
31092 * fold-const.c (split_tree): Make sure to not negate pointers.
31094 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31096 PR rtl-optimization/60818
31097 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31098 a compare of comparisons with the thing compared if this results
31099 in a different machine mode.
31101 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31103 * alias.c (base_alias_check): Fix typo in comment.
31104 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31105 * cgraphunit.c (symbol_table::compile): Likewise.
31106 * collect2.c (maybe_run_lto_and_relink): Likewise.
31107 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31108 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31109 * config/avr/avr.c (avr_map_op_t): Likewise.
31110 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31111 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31112 * config/epiphany/epiphany.md (movcc): Likewise.
31113 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31114 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31116 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31117 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31118 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31119 * config/sh/sh.c (sh_rtx_costs): Likewise.
31120 * fold-const.c (fold_truth_andor): Likewise.
31121 * genautomata.c (collapse_flag): Likewise.
31122 * gengtype.h (struct type::u::s): Likewise.
31123 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31124 * input.c (FORMAT_AMOUNT): Likewise.
31125 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31126 (known_aggs_to_agg_replacement_list): Likewise.
31127 * ipa-inline-analysis.c: Likewise.
31128 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31129 * ipa-polymorphic-call.c
31130 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31131 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31132 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31134 * modulo-sched.c (apply_reg_moves): Likewise.
31135 * omp-expand.c (build_omp_regions_1): Likewise.
31136 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31137 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31138 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31139 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31140 * value-prof.c: Likewise.
31141 * var-tracking.c (val_reset): Likewise.
31143 2017-04-03 Richard Biener <rguenther@suse.de>
31145 PR tree-optimization/80275
31146 * fold-const.c (split_address_to_core_and_offset): Handle
31149 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31151 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31152 descriptors is at least equal to that of functions.
31154 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31156 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31158 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31161 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31162 (mov<IMOD4:mode>): New expander.
31163 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31165 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31167 PR rtl-optimization/79405
31168 * fwprop.c (propagations_left): New variable.
31169 (forward_propagate_into): Decrement it.
31170 (fwprop_init): Initialize it.
31171 (fw_prop): If the variable has reached zero, stop propagating.
31172 (fwprop_addr): Ditto.
31174 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31177 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31178 a FUNCTION_DECL, pass it as decl instead of origin to
31181 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31183 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31186 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31189 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31190 TARGET_VSX_SMALL_INTEGER.
31192 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31194 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31195 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31197 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31199 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31200 extraction from odd-numbered MSA register.
31202 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31204 PR middle-end/80173
31205 * expmed.c (store_bit_field_1): Don't attempt to create
31206 a word subreg out of hard registers wider than word if they
31207 have HARD_REGNO_NREGS of 1 for their mode.
31209 PR middle-end/80163
31210 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31211 conversions to integer types wider than word and pointer.
31214 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31215 (rtx_equal_for_cselib_p): Pass 0 to it.
31216 * cselib.c (cselib_hasher::equal): Likewise.
31217 (rtx_equal_for_cselib_1): Add depth argument. If depth
31218 is 128, don't look up VALUE locs and punt. Increment
31219 depth in recursive calls when walking VALUE locs.
31221 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31223 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31224 (make_gcov_file_name): Use the canonical path name for generating
31226 (read_line): Fix handling of files with ascii null bytes.
31228 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31230 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31231 to initialise a vector register instead
31232 of using a const_int.
31234 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31236 PR translation/80189
31237 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31238 diagnostic messages.
31240 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31243 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31244 (dfp_diex_<mode>): Update mode of operand 1.
31245 * doc/extend.texi (dxex, dxexq): Document change to return type.
31246 (diex, diexq): Document change to argument type.
31248 2017-03-30 Martin Jambor <mjambor@suse.cz>
31251 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31252 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31253 it reflects the signature changes performed at the callee side.
31254 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31255 to cgraph_build_function_type_skip_args.
31256 (build_function_decl_skip_args): Adjust call to the above function.
31258 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31261 * config/i386/sse.md
31262 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31263 register as dest whenever it is a MEM not rtx_equal_p to the
31264 corresponding dup operand, and when forcing into reg move the
31265 reg into the memory afterwards.
31266 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31267 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31268 for the force_reg mode.
31269 (avx512vl_vextractf128<mode>): Use register as dest either
31270 always when a MEM, or when it is a MEM not rtx_equal_p to the
31271 corresponding dup operand, or even not when it is a CONST_VECTOR
31272 depending on the mode and lo vs. hi.
31273 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31275 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31276 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31277 Likewise. Require that operands[2] is even.
31278 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31279 Remove extraneous parens. Require that operands[2] is a multiple
31281 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31282 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31283 disallow memory then.
31285 2017-03-30 Richard Biener <rguenther@suse.de>
31287 PR tree-optimization/77498
31288 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31289 to non-constants over backedges.
31291 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31293 PR rtl-optimization/80233
31294 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31295 as last_combined_insn. Do not test for BARRIER_P separately.
31297 2017-03-29 Andreas Schwab <schwab@suse.de>
31300 * calls.c (prepare_call_address): Convert funexp to Pmode before
31301 copying to temp reg.
31303 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31305 PR tree-optimization/80158
31306 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31307 Handle possible future case of more than one alternate
31309 (replace_rhs_if_not_dup): Likewise.
31310 (replace_one_candidate): Likewise.
31312 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31314 PR rtl-optimization/80193
31315 * ira.c (ira): Do not check allocation for LRA.
31317 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31319 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31320 (nvptx_output_simt_exit): Declare.
31321 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31322 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31323 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31324 (nvptx_declare_function_name): ...here. Emit declaration of local
31325 memory space buffer for omp_simt_enter insn.
31326 (nvptx_output_unisimt_switch): New.
31327 (nvptx_output_softstack_switch): New.
31328 (nvptx_output_simt_enter): New.
31329 (nvptx_output_simt_exit): New.
31330 * config/nvptx/nvptx.h (struct machine_function): New fields
31331 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31332 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31333 (UNSPECV_SIMT_EXIT): Ditto.
31334 (omp_simt_enter_insn): New insn.
31335 (omp_simt_enter): New expansion.
31336 (omp_simt_exit): New insn.
31337 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31339 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31340 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31341 (expand_GOMP_SIMT_EXIT): New.
31342 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31343 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31344 (GOMP_SIMT_EXIT): Ditto.
31345 * target-insns.def (omp_simt_enter): New insn.
31346 (omp_simt_exit): Ditto.
31347 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31349 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31350 (lower_rec_input_clauses): Likewise.
31351 (lower_lastprivate_clauses): Handle SIMT privatization.
31353 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31354 (ompdevlow_adjust_simt_enter): New.
31355 (find_simtpriv_var_op): New.
31356 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31357 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31359 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31360 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31361 (copy_decl_for_dup_finish): Ditto.
31363 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31365 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31368 * config/i386/i386.c (ix86_option_override_internal): Always
31369 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31371 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31373 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31375 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31377 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31378 mark new edge's irreducible flag accordign to it.
31379 (vect_do_peeling): Check loop preheader edge's irreducible flag
31380 and pass it to function slpeel_add_loop_guard.
31382 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31384 PR tree-optimization/80218
31385 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31386 Update block frequencies and counts.
31388 2017-03-28 Richard Biener <rguenther@suse.de>
31390 PR tree-optimization/78644
31391 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31392 of a simplification result we may not use it at all.
31394 2017-03-28 Richard Biener <rguenther@suse.de>
31397 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31398 without arguments, generate default definition of a SSA name.
31400 2017-03-28 Richard Biener <rguenther@suse.de>
31402 PR middle-end/80222
31403 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31404 TYPE_REF_CAN_ALIAS_ALL references.
31405 * fold-const.c (fold_indirect_ref_1): Likewise.
31407 2017-03-28 Martin Liska <mliska@suse.cz>
31410 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31411 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31413 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31414 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31416 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31417 (EXTRA_SPECS): Define.
31418 (SUBTARGET_EXTRA_SPECS): Likewise.
31419 (SUBTARGET_CPP_SPEC): Likewise.
31420 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31421 SUBTARGET_EXTRA_SPECS.
31422 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31424 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31426 * config/arc/simdext.md (vst64_insn): Update pattern.
31427 (vld32wh_insn): Likewise.
31428 (vld32wl_insn): Likewise.
31429 (vld64_insn): Likewise.
31430 (vld32_insn): Likewise.
31432 2017-03-28 Marek Polacek <polacek@redhat.com>
31435 * fold-const.c (fold_comparison): Use protected_set_expr_location
31436 instead of SET_EXPR_LOCATION.
31438 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31440 * tree.c (add_expr): Avoid name lookup warning.
31442 2017-03-27 Jeff Law <law@redhat.com>
31444 PR tree-optimization/80216
31445 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31446 function name. Limit recursion depth.
31447 (record_temporary_equivalences): Corresponding changes.
31449 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31451 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31454 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31457 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31459 * cfgcleanup.c (reg_note_cfa_p): New array.
31460 (insns_have_identical_cfa_notes): New function.
31461 (old_insns_match_p): Don't cross-jump in between /f
31462 and non-/f instructions. If both i1 and i2 are frame related,
31463 verify all CFA notes, their order and content.
31465 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31468 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31469 HImode and SImode with zero extend to DImode to one insn.
31470 (bswap<mode>2_extenddi): Likewise.
31471 (bswapsi2_extenddi): Likewise.
31472 (bswaphi2_extendsi): Likewise.
31473 (bswaphi2): Combine bswap HImode and SImode into one insn.
31474 Separate memory insns from swapping register.
31475 (bswapsi2): Likewise.
31476 (bswap<mode>2): Likewise.
31477 (bswaphi2_internal): Delete, no longer used.
31478 (bswapsi2_internal): Likewise.
31479 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31480 store, and gpr<-gpr swap insns.
31481 (bswap<mode>2_store): Likewise.
31482 (bswaphi2_reg): Register only splitter, combine with the splitter.
31483 (bswaphi2 splitter): Likewise.
31484 (bswapsi2_reg): Likewise.
31485 (bswapsi2 splitter): Likewise.
31486 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31487 the insns into load, store, and register/register insns.
31488 (bswapdi2_ldbrx): Likewise.
31489 (bswapdi2_load): Likewise.
31490 (bswapdi2_store): Likewise.
31491 (bswapdi2_reg): Likewise.
31493 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31495 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31496 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31498 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31501 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31503 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31504 special handling for target option conflicts between dform
31505 options (-mpower9-dform, -mpower9-dform-vector,
31506 -mpower9-dform-scalar) and -mno-direct-move.
31508 2017-03-27 Richard Biener <rguenther@suse.de>
31510 PR tree-optimization/80181
31511 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31513 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31515 * config/arc/predicates.md (move_double_src_operand): Replace the
31516 call to move_double_src_operand with a call to address_operand.
31518 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31520 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31521 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31522 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31524 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31526 * config/arc/predicates.md (long_immediate_loadstore_operand):
31527 Consider scaled addresses cases.
31529 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31531 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31532 restored when in interrupt.
31533 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31534 doesn't have delay slot.
31536 2017-03-27 Richard Biener <rguenther@suse.de>
31539 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31540 inlined thunk clones.
31542 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31545 * asan.c (instrument_derefs): Copy over last operand from
31546 original COMPONENT_REF to the new COMPONENT_REF with
31547 DECL_BIT_FIELD_REPRESENTATIVE.
31548 * ubsan.c (instrument_object_size): Likewise.
31550 2017-03-27 Richard Biener <rguenther@suse.de>
31552 PR tree-optimization/80170
31553 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31554 sure DR/SCEV didnt fold in constants we do not see when looking
31555 at the reference base alignment.
31557 2017-03-27 Richard Biener <rguenther@suse.de>
31559 PR middle-end/80171
31560 * gimple-fold.c (fold_ctor_reference): Properly guard against
31561 NULL return value from canonicalize_constructor_val.
31563 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31566 * config/i386/i386.c (ix86_expand_builtin)
31567 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31568 flags reg setting and flags reg using instructions.
31569 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31570 clobbering instructions to zero extend op2.
31572 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31574 * doc/install.texi (Configuration) <--with-aix-soname>:
31575 Update link to AIX ld.
31577 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31579 PR rtl-optimization/80160
31580 PR rtl-optimization/80159
31581 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31582 reg_alternate_class into account.
31584 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31587 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31588 to consider in curr_insn_transform.
31590 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31592 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31593 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31594 and emit_mode_inner.
31596 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31598 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
31599 argument to the overloaded builtin variants. Use the new flag to
31600 deprecate certain builtin variants.
31601 * config/s390/s390-builtin-types.def: Add new builtin types.
31602 * config/s390/s390-builtins.h: Support new flags field for
31603 overloaded builtins.
31604 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
31605 (s390_macro_to_expand): Enable vector float data type.
31606 (s390_cpu_cpp_builtins_internal): Indicate support of the new
31607 builtins by incrementing the __VEC__ version number.
31608 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
31610 (s390_resolve_overloaded_builtin): Emit error messages depending
31611 on the builtin flags.
31612 * config/s390/s390.c (s390_expand_builtin): Support additional
31613 flags argument. Change error message to match the messages
31614 emitted in s390-c.c.
31615 * config/s390/s390.md: New UNSPEC_* constants.
31616 (op_type): Add new instruction types.
31617 * config/s390/vecintrin.h: Add new builtins and test data class
31619 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
31620 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
31621 (VEC_INEXACT, VEC_NOINEXACT): New constants.
31622 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
31623 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
31624 ("vec_mergel<mode>"): V_HW -> VEC_HW.
31626 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
31627 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
31628 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
31629 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
31631 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
31632 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
31633 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
31634 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
31636 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
31637 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
31638 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
31639 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
31640 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
31641 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
31642 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
31644 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
31645 ("vec_scatter_element<V_HW_4:mode>_DI")
31646 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
31647 ("vec_fpint<mode>", "vflls")
31648 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
31649 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
31650 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
31651 ("*vec_cmphe<mode>_cc"): ... these.
31653 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
31654 mode constant instead of magic value.
31656 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31658 * config/s390/s390.c (s390_expand_vec_compare): Support other
31659 vector floating point modes than just V2DF.
31660 (s390_expand_vcond): Likewise.
31661 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
31662 (s390_cannot_change_mode_class): Prevent mode changes between TF
31663 and V1TF in vector registers.
31664 * config/s390/s390.md (DF, SF): New mode attributes.
31665 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
31666 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
31667 SFmode support for VRs.
31668 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
31670 (VFT, VF_HW): New mode iterators.
31671 (vw, sdx): New mode attributes.
31672 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
31673 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
31674 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
31675 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
31676 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
31677 also the new vector floating point modes. Renaming to ...
31679 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
31680 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
31681 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
31682 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
31683 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
31684 ("vec_unordered<mode>"): ... these.
31686 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
31687 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
31688 ("*vec_extendv2df"): New insn definitions.
31690 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31692 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
31693 ("mulditi3_2", "*muldi3_sign"): New patterns.
31694 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
31695 rename the pattern definition.
31697 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31699 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
31701 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
31703 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31705 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
31706 instruction if possible.
31707 * config/s390/vector.md (vec_halfnumelts): New mode
31709 ("*vec_vllezlf<mode>"): New pattern.
31711 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31713 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
31714 ("popcountv4si2", "popcountv2di2"): Rename to ...
31715 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
31716 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
31718 ("popcount<mode>2_vxe"): New pattern.
31720 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31722 * common/config/s390/s390-common.c (processor_flags_table): Add
31724 * config.gcc: Add arch12.
31725 * config/s390/driver-native.c (s390_host_detect_local_cpu):
31726 Default to arch12 for unknown CPU model numbers.
31727 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
31728 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
31729 PROCESSOR_max sanity check.
31730 * config/s390/s390-opts.h (enum processor_type): Add
31732 * config/s390/s390.c (processor_table): Add arch12.
31733 (s390_expand_builtin): Add check for B_VXE flag.
31734 (s390_issue_rate): Add PROCESSOR_ARCH12.
31735 (s390_get_sched_attrmask): Likewise.
31736 (s390_get_unit_mask): Likewise.
31737 (s390_sched_score): Enable z13 scheduling for arch12.
31738 (s390_sched_reorder): Likewise.
31739 (s390_sched_variable_issue): Likewise.
31740 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
31742 (s390_tune_attr): Use z13 scheduling also for arch12.
31743 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
31744 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
31745 (TARGET_VXE_P): New macros.
31746 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
31747 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
31748 * config/s390/s390.opt: Add arch12 as processor_type.
31750 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31752 * config/s390/s390.md
31753 ("fixuns_truncdddi2", "fixuns_trunctddi2")
31754 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
31755 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
31757 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
31758 Rename expanders to ...
31760 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
31761 ("fixuns_truncdddi2_emu"): ... these.
31763 ("fixuns_trunc<mode>si2_emu"): New expander.
31765 ("*fixuns_truncdfdi2_z13"): Rename to ...
31766 ("*fixuns_truncdfdi2_vx"): ... this.
31768 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31770 * config/s390/2964.md: Remove the single element vector compare
31771 instructions which are no longer used.
31772 * config/s390/s390.c (s390_select_ccmode): Remove handling of
31774 (s390_canonicalize_comparison): Remove handling of DFmode
31776 (s390_expand_vec_compare_scalar): Remove function.
31777 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
31778 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
31780 ("*cmp<mode>_ccs"): Add wfcdb instruction.
31782 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31784 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
31786 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
31787 will anyway by matched by mov<mode>_64dfp.
31789 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31791 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
31792 vlef/vstef. Add missing operand to vleif.
31794 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31796 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
31797 pair for all vector types with 64 bit elements.
31798 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
31799 * config/s390/vector.md (V_HW_64): ... here.
31800 (V_128_NOSINGLE): New mode iterator.
31801 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
31802 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
31803 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
31804 ("*vec_load_pairv2di"): Change to ...
31805 ("*vec_load_pair<mode>"): ... this one.
31807 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31809 * config/s390/constraints.md: Add comments.
31810 (jKK): Reject element sizes > 8 bytes.
31811 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
31813 * config/s390/s390.md: Add the s_operand checks formerly in
31814 s390_split_ok_p to various splitters where they are still
31816 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
31817 for 128 bit vectors. Plus two splitters.
31819 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31821 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
31824 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31827 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
31828 error if the boundary argument is not constant.
31830 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31832 PR rtl-optimization/80112
31833 * loop-doloop.c (doloop_condition_get): Don't check condition
31834 if cmp isn't SET with IF_THEN_ELSE src.
31836 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31838 PR tree-optimization/80158
31839 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
31840 replacing a candidate statement, also replace it for the
31841 candidate's alternate interpretation.
31842 (replace_rhs_if_not_dup): Likewise.
31843 (replace_one_candidate): Likewise.
31845 2017-03-24 Richard Biener <rguenther@suse.de>
31847 PR tree-optimization/80167
31848 * graphite-isl-ast-to-gimple.c
31849 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
31851 (translate_isl_ast_to_gimple::get_rename): Likewise.
31853 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31855 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
31856 handling of certain combinations of target options, including the
31857 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
31858 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
31860 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31863 * config/arm/arm.md (*load_multiple): Add reload_completed to
31864 matching condition.
31866 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31867 Richard Biener <rguenth@suse.de>
31869 PR tree-optimization/79908
31870 PR tree-optimization/80136
31871 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31872 been cast away, gimplify_and_add suffices.
31874 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
31876 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
31878 2017-03-23 Richard Biener <rguenther@suse.de>
31880 PR tree-optimization/80032
31881 * gimplify.c (gimple_push_cleanup): Forced unconditional
31882 cleanups still have to go to the conditional_cleanups
31885 2017-03-22 Jakub Jelinek <jakub@redhat.com>
31887 PR tree-optimization/80072
31888 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
31890 (next_operand_entry_id): Change type to unsigned int.
31891 (sort_by_operand_rank): Make sure to return the right return value
31892 even if unsigned fields are bigger than INT_MAX.
31893 (struct oecount): Change cnt and id type to unsigned int.
31894 (oecount_hasher::equal): Formatting fix.
31895 (oecount_cmp): Make sure to return the right return value
31896 even if unsigned fields are bigger than INT_MAX.
31897 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
31900 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
31901 TREE_READONLY on result if writing it more than once.
31904 * doc/invoke.texi (-fsanitize=thread): Document that with
31905 -fnon-call-exceptions atomics are not able to throw
31909 * tsan.c: Include tree-eh.h.
31910 (instrument_builtin_call): Call maybe_clean_eh_stmt or
31911 maybe_clean_or_replace_eh_stmt where needed.
31912 (instrument_memory_accesses): Add cfg_changed argument.
31913 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
31914 if it returned true.
31915 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
31917 PR rtl-optimization/63191
31918 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
31919 wrapper function, moved the whole old content into ...
31920 (ix86_delegitimize_address_1): ... this. New inline function.
31921 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
31922 true as last argument instead of ix86_delegitimize_address.
31924 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31926 * config/aarch64/aarch64.c (generic_branch_cost): Copy
31927 cortexa57_branch_cost.
31929 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31931 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
31933 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31936 * doc/md.texi (Constraints): Document wA constraint.
31937 * config/rs6000/constraints.md (wA): New.
31938 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
31939 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
31940 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
31941 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
31943 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
31946 * gimplify.c (is_oacc_declared): New function.
31947 (oacc_default_clause): Use it to set default flags for acc declared
31948 variables inside parallel regions.
31949 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
31950 declared variables.
31951 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
31952 declare attribute to any decl as necessary.
31954 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
31957 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
31958 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
31959 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
31960 (arm_arch_lpae): This.
31961 * config/arm/arm.c (arm_arch7ve): Rename into ...
31962 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
31963 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
31966 2017-03-22 Martin Liska <mliska@suse.cz>
31969 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
31970 error message instead of an ICE.
31972 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31974 * doc/extend.texi (6.11 Additional Floating Types): Revise.
31976 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
31978 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31980 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31983 2017-03-21 Martin Sebor <msebor@redhat.com>
31985 * doc/extend.texi: Use "cannot" instead of "can't."
31986 * doc/hostconfig.texi: Same.
31987 * doc/install.texi: Same.
31988 * doc/invoke.texi: Same.
31989 * doc/loop.texi: Same.
31990 * doc/md.texi: Same.
31991 * doc/objc.texi: Same.
31992 * doc/rtl.texi: Same.
31993 * doc/tm.texi: Same.
31994 * doc/tm.texi.in: Same.
31995 * doc/trouble.texi: Same.
31997 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32000 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32001 (collect_checksum_attributes): Set it.
32002 (die_checksum_ordered): Use it.
32004 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32006 PR tree-optimization/79908
32007 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32008 change: For a VA_ARG whose LHS has been cast away, use
32009 force_gimple_operand to construct the side effects.
32011 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32013 PR translation/80001
32014 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32015 more amenable to translation.
32016 (oacc_loop_auto_partitions): Likewise.
32018 2017-03-21 Marek Polacek <polacek@redhat.com>
32019 Martin Sebor <msebor@redhat.com>
32021 PR tree-optimization/80109
32022 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32023 on INTEGRAL_TYPE_P.
32025 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32026 Segher Boessenkool <segher@kernel.crashing.org>
32029 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32030 check reg_used_between_p between insn and one of succ or succ2
32031 depending on if succ is artificial insn not inserted into insn
32034 2017-03-21 Martin Liska <mliska@suse.cz>
32036 PR gcov-profile/80081
32037 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32038 * doc/gcc.texi: Include gcov-dump stuff.
32039 * doc/gcov-dump.texi: New file.
32041 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32043 PR rtl-optimization/79150
32044 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32045 conditional jump, if the jump is the last insn of the loop.
32047 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32048 Richard Biener <rguenth@suse.de>
32050 PR tree-optimization/79908
32051 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32052 been cast away, use force_gimple_operand to construct the side
32055 2017-03-21 Martin Liska <mliska@suse.cz>
32057 PR libfortran/79956
32058 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32061 2017-03-21 Brad Spengler <spender@grsecurity.net>
32064 * plugin.c (htab_hash_plugin): New function.
32065 (add_new_plugin): Use it and adjust.
32066 (parse_plugin_arg_opt): Adjust.
32067 (init_one_plugin): Likewise.
32069 2017-03-21 Richard Biener <rguenther@suse.de>
32071 PR tree-optimization/80032
32072 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32073 if set force the cleanup to happen unconditionally.
32074 (gimplify_target_expr): Push inserted clobbers with force_uncond
32075 to avoid them being removed by control-dependent DCE.
32077 2017-03-21 Richard Biener <rguenther@suse.de>
32079 PR tree-optimization/80122
32080 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32081 va_arg_pack_len when the inlined call stmt requires pack
32083 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32085 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32088 * tsan.c (instrument_builtin_call): If the memory model argument
32089 is not a constant, assume it is valid.
32092 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32095 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32097 PR rtl-optimization/79910
32098 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32099 if its dest is used by an insn before I2 (other than the combined
32100 insns themselves, which are properly handled already).
32102 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32105 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32107 * combine.c (record_used_regs): New static function.
32108 (try_combine): Handle situations where there is an additional
32109 instruction between I2 and I3 which needs to have a LOG_LINK
32113 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32115 * combine.c (try_combine): Delete redundant i1 test. Call
32116 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32118 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32121 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32122 alternatives 13/14.
32124 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32126 PR tree-optimization/80054
32127 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32128 the optimization if a PHI or any of its arguments is not dominated
32129 by the candidate's basis. Use gphi* rather than gimple* as
32131 (replace_profitable_candidates): Clean up a gimple* variable that
32132 should be a gphi* variable.
32134 2017-03-20 Martin Sebor <msebor@redhat.com>
32137 * doc/extend.texi (attribute constructor): Document present limitation.
32139 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32142 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32143 __POWER9_VECTOR__ #ifdef control, change template definition to
32144 use Power9-specific built-in function.
32145 (vec_any_eq): Likewise.
32146 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32147 to control outcomes from this test.
32148 (vector_ae_<mode>p): For VEC_F modes, likewise.
32150 2017-03-20 Ian Lance Taylor <iant@google.com>
32152 * config/i386/i386.c (ix86_function_regparm): Save an extra
32153 register for -fsplit-stack with DECL_STATIC_CHAIN.
32155 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32158 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32159 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32161 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32163 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32165 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32168 2017-03-20 Marek Polacek <polacek@redhat.com>
32171 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32173 2017-03-20 Richard Biener <rguenther@suse.de>
32175 PR tree-optimization/80113
32176 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32177 allocate extra SSA name for PHI def.
32178 (add_close_phis_to_outer_loops): Likewise.
32179 (add_close_phis_to_merge_points): Likewise.
32180 (copy_loop_close_phi_args): Likewise.
32181 (copy_cond_phi_nodes): Likewise.
32183 2017-03-20 Martin Liska <mliska@suse.cz>
32185 PR middle-end/79753
32186 * tree-chkp.c (chkp_build_returned_bound): Do not build
32187 returned bounds for a LHS that's not a BOUNDED_P type.
32189 2017-03-20 Martin Liska <mliska@suse.cz>
32193 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32194 COMPLEX_CST and VECTOR_CST.
32196 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32199 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32200 target operand. A new splitter adds the clobber statement in case
32201 the target operand is dead anyway.
32203 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32205 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32206 to age-old versions of binutils and glibc.
32208 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32210 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32212 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32214 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32216 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32218 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32219 requirement for binutils 2.13.
32221 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32223 * combine.c (try_combine): Delete redundant i1 test. Call
32224 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32226 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32228 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32229 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32231 <riscv64-*-elf>: Re-arrange section
32232 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32233 <riscv32-*-linux>: Likewise.
32234 <riscv64-*-elf>: Likewise
32235 <riscv64-*-linux>: Likewise.
32237 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32240 * aarch64.opt(verbose-cost-dump): Fix typo.
32242 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32245 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32246 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32248 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32250 * reload.c (find_reloads): When reloading a nonoffsettable address,
32251 use RELOAD_OTHER for it and its address reloads.
32253 PR rtl-optimization/79910
32254 * combine.c (record_used_regs): New static function.
32255 (try_combine): Handle situations where there is an additional
32256 instruction between I2 and I3 which needs to have a LOG_LINK
32259 2017-03-17 Jeff Law <law@redhat.com>
32261 PR tree-optimization/71437
32262 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32263 conditional in the hash table first.
32264 (vrp_dom_walker::before_dom_children): Extract condition from
32265 ASSERT_EXPR. Record condition, its inverion and any implied
32266 conditions as well.
32268 2017-03-17 Marek Polacek <polacek@redhat.com>
32269 Markus Trippelsdorf <markus@trippelsdorf.de>
32271 PR tree-optimization/80079
32272 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32275 2017-03-17 Richard Biener <rguenther@suse.de>
32277 PR middle-end/80075
32278 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32279 Properly verify the LHS before the RHS possibly claims to be
32281 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32284 2017-03-17 Martin Jambor <mjambor@suse.cz>
32286 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32287 (List of -O2 options): Likewise.
32288 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32291 2017-03-17 Tom de Vries <tom@codesourcery.com>
32293 * gcov-dump.c (print_usage): Print bug_report_url.
32295 2017-03-17 Richard Biener <rguenther@suse.de>
32297 PR middle-end/80050
32298 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32299 (parser::peek): Likewise.
32301 2017-03-17 Richard Biener <rguenther@suse.de>
32303 PR tree-optimization/80048
32304 * sese.c (free_sese_info): Properly release rename_map and
32305 copied_bb_map elements.
32307 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32309 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32310 Add linked-list forward and backlinks. Insert on
32311 construction, remove on destruction.
32312 (class pass_store_merging): Add m_stores_head field.
32313 (pass_store_merging::terminate_and_process_all_chains):
32314 Iterate over m_stores_head list.
32315 (pass_store_merging::terminate_all_aliasing_chains):
32317 (pass_store_merging::execute): Check for debug stmts first.
32318 Push new chains onto the m_stores_head stack.
32320 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32323 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32324 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32325 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32327 2017-03-16 Jeff Law <law@redhat.com>
32329 PR tree-optimization/71437
32330 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32331 member function. Implementation moved into after_dom_children
32332 member function and into the threader's thread_outgoing_edges
32334 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32335 some code into new thread_outgoing_edges.
32336 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32337 definition. Simplify marker handling (do it here). Assume we always
32338 have the available expression and the const/copies tables.
32339 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32341 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32342 * tree-vrp.c (equiv_stack): No longer file scoped.
32343 (vrp_dom_walker): New class.
32344 (vrp_dom_walker::before_dom_children): New member function.
32345 (vrp_dom_walker::after_dom_children): Likewise.
32346 (identify_jump_threads): Setup domwalker. Use it rather than
32347 walking edges in a random order by hand. Simplify setup/finalization.
32348 (finalize_jump_threads): Remove.
32349 (vrp_finalize): Do not call identify_jump_threads here.
32350 (execute_vrp): Do it here instead and call thread_through_all_blocks
32353 PR tree-optimization/71437
32354 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32356 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32358 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32359 (dom_opt_dom_walker::thread_across_edge): Remove
32360 handle_dominating_asserts argument. All callers changed.
32361 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32362 changes. Remove calls to lhs_of_dominating_assert. Other
32363 uses of handle_dominating_asserts turn into unconditional code
32364 (simplify_control_stmt_condition_1): Likewise.
32365 (simplify_control_stmt_condition): Likewise.
32366 (thread_through_normal_block, thread_across_edge): Likewise.
32367 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32368 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32369 object if it is not an SSA_NAME.
32370 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32371 before calling into the VRP specific simplifiers.
32372 (identify_jump_threads): Remove handle_dominating_asserts
32375 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32378 * tree-diagnostic.c (default_tree_printer): No longer static.
32379 * tree-diagnostic.h (default_tree_printer): New prototype.
32381 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32383 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32384 Change ins into fmov.
32386 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32388 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32389 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32390 Use h_con constraint for operand 1.
32391 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32392 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32394 2017-03-15 Jeff Law <law@redhat.com>
32396 PR tree-optimization/71437
32397 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32398 (record_temporary_equivalences): Use it.
32400 PR tree-optimization/71437
32401 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32402 tree-ssa-scopedtables.
32403 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32404 (record_conditions, record_cond, vuse_eq): Likewise.
32405 (record_edge_info): Adjust to API tweak of record_conditions.
32406 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32407 (record_temporary_equivalences, optimize_stmt): Likewise.
32408 (eliminate_redundant_computations): Likewise.
32409 (record_equivalences_from_stmt): Likewise.
32410 * tree-ssa-scopedtables.c: Include options.h and params.h.
32411 (vuse_eq): New function, moved from tree-ssa-dom.c
32412 (build_and_record_new_cond): Likewise.
32413 (record_conditions): Likewise. Accept vector of conditions rather
32414 than edge_equivalence structure for first argument.
32415 for the first argument.
32416 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32417 from tree-ssa-dom.c.
32418 (avail_exprs_stack::record_cond): Likewise.
32419 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32420 from tree-ssa-dom.c.
32421 (avail_exprs_stack): Add new member functions lookup_avail_expr
32423 (record_conditions): Declare.
32425 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32428 * lra-constraints.c (process_alt_operands): Increase reject for
32429 reloading an input/output operand.
32431 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32434 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32435 insns to convert from signed/unsigned char/short to IEEE 128-bit
32437 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32439 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32442 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32443 subreg of inner mode for values already in registers.
32445 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32447 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32448 iteration reg is used after the loop.
32450 2017-03-14 Martin Sebor <msebor@redhat.com>
32452 PR tree-optimization/79800
32453 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32454 precision in negative-positive range.
32455 (format_floating): Call non-const overload with adjusted precision.
32457 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32460 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32463 2017-03-14 Martin Sebor <msebor@redhat.com>
32465 PR middle-end/80020
32466 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32467 * builtins.def (aligned_alloc): Use it.
32470 * Makefile.in (GTFILES): Add calls.c.
32471 * calls.c: Include "gt-calls.h".
32473 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32475 PR rtl-optimization/79728
32476 * regs.h (struct target_regs): New field
32477 x_contains_allocatable_regs_of_mode.
32478 (contains_allocatable_regs_of_mode): New macro.
32479 * reginfo.c (init_reg_sets_1): Initialize it, and change
32480 contains_reg_of_mode so it includes global regs as well.
32481 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32482 rather than contains_regs_of_mode.
32484 2017-03-14 Martin Liska <mliska@suse.cz>
32486 * doc/invoke.texi: Document options that can't be combined with
32487 -fcheck-pointer-bounds.
32489 2017-03-14 Martin Liska <mliska@suse.cz>
32491 PR middle-end/79831
32492 * doc/invoke.texi (-Wchkp): Document the option.
32494 2017-03-14 Martin Liska <mliska@suse.cz>
32496 * Makefile.in: Install gcov-dump.
32498 2017-03-14 Martin Liska <mliska@suse.cz>
32500 * multiple_target.c (expand_target_clones): Bail out for
32501 an invalid attribute.
32503 2017-03-14 Richard Biener <rguenther@suse.de>
32505 * alias.c (struct alias_set_entry): Pack properly.
32506 * cfgloop.h (struct loop): Likewise.
32507 * cse.c (struct set): Likewise.
32508 * ipa-utils.c (struct searchc_env): Likewise.
32509 * loop-invariant.c (struct invariant): Likewise.
32510 * lra-remat.c (struct cand): Likewise.
32511 * recog.c (struct change_t): Likewise.
32512 * rtl.h (struct address_info): Likewise.
32513 * symbol-summary.h (function_summary): Likewise.
32514 * tree-loop-distribution.c (struct partition): Likewise.
32515 * tree-object-size.c (struct object_size_info): Likewise.
32516 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32517 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32518 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32519 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32520 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32521 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32522 (struct _stmt_vec_info): Likewise.
32524 2017-03-14 Martin Liska <mliska@suse.cz>
32527 * multiple_target.c (create_dispatcher_calls): Check that
32528 a target can create a function dispatcher.
32530 2017-03-14 Martin Liska <mliska@suse.cz>
32533 * multiple_target.c (expand_target_clones): Drop local.local
32534 flag for default implementation.
32536 2017-03-14 Richard Biener <rguenther@suse.de>
32538 PR tree-optimization/80030
32539 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32541 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32543 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32544 gcc_fallthrough() instead of __attribute__((fallthrough));
32546 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32548 * doc/gcc.texi: Remove "up" link to (DIR).
32549 * doc/gccint.texi: Ditto.
32551 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32553 * doc/install.texi (Specific) <avr>: Remove reference to
32556 2017-03-13 Jeff Law <law@redhat.com>
32558 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32559 attribute rather than comments.
32561 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32562 match_scratch operand is highest.
32564 2017-03-13 Martin Liska <mliska@suse.cz>
32566 PR middle-end/78339
32567 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32568 is a CHKP clone, use original declaration.
32570 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32572 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32573 (arc_conditional_register_usage): Use a different allocation order
32574 when optimizing for size.
32575 * common/config/arc/arc-common.c (arc_option_optimization_table):
32576 Section anchors default on when optimizing for size.
32578 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32580 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32582 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32584 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32585 * config/arc/arc.md (cpu_facility): Add cd variant.
32586 (*movqi_insn): Add code density variant.
32587 (*movhi_insn): Likewise.
32588 (*movqi_insn): Likewise.
32589 (*addsi3_mixed): Likewise.
32590 (subsi3_insn): Likewise.
32592 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32594 * config/arc/arc.md (movsi_cond_exec): Update constraint.
32596 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32598 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
32599 expressions with MINUS and UNARY ops.
32601 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32604 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
32606 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
32607 between vec_select and vector argument.
32608 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
32609 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
32610 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
32611 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
32612 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
32613 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
32615 2017-03-13 Richard Biener <rguenther@suse.de>
32618 * params.def (vect-max-peeling-for-alignment): Fix typo.
32620 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32622 * doc/install.texi (Specific) <mips-*-*>: Remove description of
32623 issue that only occurred with binutils below 2.18.
32625 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32627 * doc/install.texi (Specific) <cris-axis-elf>: No longer
32628 refer to binutils 2.11/2.12 minimum.
32630 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32632 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
32633 ftp.kernel.org and simplify binutils requirement.
32635 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
32637 * doc/invoke.texi (Warning Options): Fix spelling of link-time
32639 (Optimize Options): Ditto. Also remove redundancy.
32641 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32643 PR translation/79848
32644 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
32646 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
32647 to G_ to avoid double translation.
32649 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32651 PR translation/79923
32652 * auto-profile.c (get_combined_location): Convert leading
32653 character of diagnostics to lower case and remove trailing period.
32654 (read_profile): Likewise for various diagnostics.
32655 * config/arm/arm.c (arm_option_override): Remove trailing period
32656 from various diagnostics.
32657 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
32658 (msp430_expand_delay_cycles): Likewise.
32660 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32663 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
32664 full command-line argument, rather than just "str".
32665 (aarch64_validate_march): Likewise.
32666 (aarch64_validate_mtune): Likewise.
32668 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
32670 PR rtl-optimization/78911
32671 * lra-assigns.c (must_not_spill_p): New function.
32672 (spill_for): Use it.
32674 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32676 PR tree-optimization/79981
32677 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
32678 ATOMIC_COMPARE_EXCHANGE ifn result.
32679 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
32680 IFN_ATOMIC_COMPARE_EXCHANGE.
32682 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32685 * opts.c (parse_sanitizer_options): Add missing question mark to
32686 "did you mean" message.
32688 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32690 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
32692 (VMULEUH_UNS): Likewise.
32693 (VMULOUB_UNS): Likewise.
32694 (VMULOUH_UNS): Likewise.
32695 * config/rs6000/rs6000.c (builtin_function_type): Remove
32696 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
32698 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32701 * read-rtl-function.c (function_reader::read_rtx_operand): Update
32702 x with result of extra_parsing_for_operand_code_0.
32703 (function_reader::extra_parsing_for_operand_code_0): Convert
32704 return type from void to rtx, returning x. When reading
32705 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
32706 larger size containing struct block_symbol.
32708 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
32710 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
32711 -mfloat128-hardware without -m64.
32713 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
32716 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
32717 entries to the case statement that marks unsigned arguments to
32718 overloaded functions.
32720 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32722 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
32723 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
32725 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
32728 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
32729 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
32731 2017-03-10 Martin Liska <mliska@suse.cz>
32735 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
32736 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
32739 2017-03-10 Olivier Hainque <hainque@adacore.com>
32741 * tree-switch-conversion (array_value_type): Start by resetting
32742 candidate type to it's main variant.
32744 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32746 PR rtl-optimization/79909
32747 * combine.c (try_combine): Use simplify_replace_rtx on individual
32748 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
32749 of the whole CALL_INSN_FUNCTION_USAGE.
32751 PR tree-optimization/79972
32752 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
32753 get_range_info on SSA_NAMEs. Formatting fixes.
32755 2017-03-10 Richard Biener <rguenther@suse.de>
32756 Jakub Jelinek <jakub@redhat.com>
32758 PR tree-optimization/77975
32759 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
32760 edge to be constant.
32761 (get_val_for): For constant x return it. Formatting fix.
32762 (loop_niter_by_eval): Avoid pointless looping if the next iteration
32763 would use the same bases as the current one.
32765 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32767 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
32768 instead of vec_select for V1TImode.
32769 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
32771 (VSX_LE_128): Add V1TI to this mode iterator.
32772 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
32773 (*vsx_le_perm_store_<mode>): Likewise.
32774 (pre-reload splitter for VSX stores): Likewise.
32775 (post-reload splitter for VSX stores): Likewise.
32776 (*vsx_xxpermdi2_le_<mode>): Likewise.
32777 (*vsx_lxvd2x2_le_<mode>): Likewise.
32778 (*vsx_stxvd2x2_le_<mode>): Likewise.
32780 2017-03-09 Michael Eager <eager@eagercon.com>
32782 Correct failures with --enable-checking=yes,rtl.
32784 * config/microblaze/microblaze.c (microblaze_expand_shift):
32785 Replace GET_CODE test with CONST_INT_P and INTVAL test with
32786 test for const0_rtx.
32787 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
32788 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
32790 2017-03-09 Richard Biener <rguenther@suse.de>
32792 PR tree-optimization/79977
32793 * graphite-scop-detection.c (scop_detection::merge_sese):
32794 Handle the case of extra exits to blocks dominating the entry.
32796 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
32798 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
32801 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
32803 PR rtl-optimization/79949
32804 * lra-constraints.c (process_alt_operands): Check memory when
32805 trying to predict a cycle. Print about the overall increase.
32807 2017-03-09 Richard Biener <rguenther@suse.de>
32809 PR middle-end/79971
32810 * gimple-expr.c (useless_type_conversion_p): Preserve
32811 TYPE_SATURATING for fixed-point types.
32813 2017-03-09 Richard Biener <rguenther@suse.de>
32816 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
32817 alignment of BLKmode params.
32819 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32822 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
32823 (VALL_NO_V2Q): Likewise.
32825 * config/aarch64/aarch64-simd.md
32826 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
32828 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
32829 VALL_NO_V2Q mode iterator.
32830 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
32832 2017-03-09 Martin Liska <mliska@suse.cz>
32834 PR tree-optimization/79631
32835 * tree-chkp-opt.c (chkp_is_constant_addr): Call
32836 tree_int_cst_sign_bit just for INTEGER constants.
32838 2017-03-09 Martin Liska <mliska@suse.cz>
32842 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
32845 2017-03-09 Marek Polacek <polacek@redhat.com>
32848 * tree.c (inchash::add_expr): Handle TREE_VEC.
32850 2017-03-09 Martin Liska <mliska@suse.cz>
32853 (chkp_narrow_size_and_offset): New function.
32854 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
32855 (void chkp_parse_bit_field_ref): New function.
32856 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
32857 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
32859 2017-03-09 Martin Liska <mliska@suse.cz>
32862 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
32863 (chkp_find_bounds_1): Remove gcc_unreachable.
32865 2017-03-09 Jakub Jelinek <jakub@redhat.com>
32868 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
32869 BUILT_IN_SYNC*, determine the access type from the size suffix and
32870 always build a MEM_REF with that type. Handle forgotten
32871 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
32874 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
32875 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
32876 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
32877 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
32878 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
32879 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
32880 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
32881 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
32882 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
32883 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
32884 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
32885 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
32886 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
32887 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
32888 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
32889 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
32890 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
32891 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
32892 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
32893 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
32894 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
32895 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
32896 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
32897 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
32898 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
32899 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
32900 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
32901 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
32902 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
32903 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
32904 definitions outside of __OPTIMIZE__ guarded section.
32907 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
32908 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
32909 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
32910 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
32913 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32915 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
32916 ("vfenez<mode>"): Add missing constraints.
32918 2017-03-08 Martin Sebor <msebor@redhat.com>
32921 * config/nds32/nds32.c (nds32_option_override):
32922 Fix misspelled diagnostic.
32924 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32927 * gimplify.c (gimplify_omp_for): Replace index var in outer
32928 taskloop statement with an artificial variable and add
32929 OMP_CLAUSE_PRIVATE clause for it.
32931 2017-03-08 Richard Biener <rguenther@suse.de>
32933 PR tree-optimization/79955
32934 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
32935 for accesses that are completely outside of the variable.
32937 2017-03-08 Andrew Haley <aph@redhat.com>
32939 PR tree-optimization/79943
32940 * tree-ssa-loop-split.c (compute_new_first_bound): When
32941 calculating the new upper bound, (END-BEG) should be added, not
32944 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32946 * config/avr/avr.md (setmemhi): Make sure match_dup
32947 operand number comes before match_scratch.
32949 2017-03-08 Richard Biener <rguenther@suse.de>
32951 PR tree-optimization/79920
32952 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
32953 with ncopies == 1 to ...
32954 (vect_transform_slp_perm_load): ... here. Properly compute
32955 all element loads by iterating VF times over the group. Do
32956 not handle ncopies (computed in a broken way) in
32957 vect_create_mask_and_perm.
32959 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32962 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
32963 is a uniform vector, use uniform_vector_p return value instead of
32964 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
32966 2017-03-07 Marek Polacek <polacek@redhat.com>
32968 PR middle-end/79809
32969 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
32970 (alloca_call_type): Likewise.
32972 2017-03-07 Martin Liska <mliska@suse.cz>
32974 * gcov.c (process_args): Put comment to correct location.
32976 2017-03-07 Martin Liska <mliska@suse.cz>
32978 PR middle-end/68270
32979 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
32980 Use array_at_struct_end_p instead of DECL_CHAIN (field).
32981 (chkp_narrow_bounds_for_field): Likewise.
32982 (chkp_parse_array_and_component_ref): Pass one more argument to
32985 2017-03-07 Richard Biener <rguenther@suse.de>
32987 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
32990 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
32992 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
32993 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
32995 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32998 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
32999 to end of description.
33000 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33002 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33004 PR rtl-optimization/79901
33005 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33007 (*avx512f_<code><mode>3<mask_name>): ... this.
33008 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33009 iterator instead of VI8_AVX2_AVX512BW.
33011 PR rtl-optimization/79901
33012 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33013 min/max expander, expand it using expand_vec_cond_expr.
33016 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33019 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33022 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33023 to void * for PCH reasons.
33024 * dwarf2out.c (output_loc_operands, output_die): Cast
33025 v.val_vec.array to unsigned char *.
33027 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33030 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33033 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33035 PR rtl-optimization/79571
33036 * lra-constraints.c (process_alt_operands): Calculate static
33037 reject and subtract it from overall when only addresses will be
33040 2017-03-06 Julia Koval <julia.koval@intel.com>
33043 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33044 incoming stack boundary to 128 for 64-bit targets.
33046 2017-03-06 Richard Biener <rguenther@suse.de>
33048 PR tree-optimization/79894
33049 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33050 to NULL after folding it.
33052 2017-03-06 Richard Biener <rguenther@suse.de>
33054 PR tree-optimization/79824
33055 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33056 check disabling peeling for gaps.
33058 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33060 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33061 attributes): Document gettimeofday.
33063 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33065 * config/s390/s390.c (s390_option_override_internal): Set
33066 PARAM_MIN_VECT_LOOP_BOUND
33068 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33070 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33071 * config/s390/s390.md: Likewise.
33073 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33076 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33077 (<avx2_avx512>_perm<mode>): Rename to ...
33078 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33080 (<avx512>_perm<mode>_mask): Rename to ...
33081 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33083 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33084 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33085 instead of VI8F_256_512.
33086 (avx512f_perm<mode>): New define_expand.
33087 (avx512f_perm<mode>_mask): Likewise.
33088 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33089 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33091 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33093 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33094 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33096 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33098 2017-03-06 Martin Liska <mliska@suse.cz>
33101 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33102 when having a SSA NAME w/o VAR_DECL assigned to it.
33104 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33106 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33107 msa_dpsub_<su>_d): Fix MODE for vec_select.
33109 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33111 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33113 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33115 2017-03-06 Richard Biener <rguenther@suse.de>
33117 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33118 * plugin.c (register_plugin_info): Likewise.
33119 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33121 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33123 * config/i386/sse.md (sse_storehps, sse_storelps,
33124 avx_<castmode><avxsizesuffix>_<castmode>,
33125 avx512f_<castmode><avxsizesuffix>_<castmode>,
33126 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33127 in condition that at least one operand is not a MEM.
33129 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33131 PR middle-end/79805
33132 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33133 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33135 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33136 gimple_call_nothrow_p flag based on whether original builtin can throw.
33137 If it can, emit following stmts on the fallthrough edge.
33138 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33139 don't create new bb if inserting just debug stmts on the edge, try to
33140 insert them on the fallthru bb or just reset debug stmts.
33142 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33145 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33146 restore recog_data (including the operand rtxes inside it) around
33147 the call to get_insn_template.
33149 2017-03-03 Martin Sebor <msebor@redhat.com>
33151 PR tree-optimization/79699
33152 * context.c (context::~context): Free MPFR caches to avoid
33153 a memory leak on program exit.
33155 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33157 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33158 Use wide_int::ulow () instead of .elt (0).
33160 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33162 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33163 (*pushxf): Limit oF constraint to 32bit targets and add oC
33164 constraint for 64bit targets.
33165 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33166 (*pushdf): Change rmF constraint to rmC.
33168 2017-03-03 Martin Liska <mliska@suse.cz>
33170 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33171 Remove unused variable.
33173 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33176 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33177 is a memory operand, increase num_memory.
33178 (ix86_expand_args_builtin): Likewise.
33180 2017-03-03 Jan Hubicka <jh@suse.cz>
33183 * ipa-devirt.c (maybe_record_node): Properly handle
33184 __cxa_pure_virtual visibility.
33186 2017-03-03 Martin Liska <mliska@suse.cz>
33188 PR tree-optimization/79803
33189 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33191 (pass_loop_prefetch::execute): Disabled optimization if an
33192 assumption about L1 cache size is not met.
33194 2017-03-03 Martin Liska <mliska@suse.cz>
33196 PR rtl-optimization/79574
33197 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33198 (hash_scan_set): Likewise.
33199 (dump_hash_table): Likewise.
33200 (hoist_code): Likewise.
33202 2017-03-03 Richard Biener <rguenther@suse.de>
33204 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33205 in place of uhigh ().
33206 (fixed_convert_from_real): Likewise.
33208 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33211 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33213 2017-03-03 Richard Biener <rguenther@suse.de>
33215 PR middle-end/79818
33216 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33217 TYPE_OVERFLOW_UNDEFINED check.
33219 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33221 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33223 (vector_ae_<mode>_p): Likewise.
33224 (vector_nez_<mode>_p): Likewise.
33225 (vector_ne_v2di_p): Likewise.
33226 (vector_ae_v2di_p): Likewise.
33227 (vector_ne_<mode>_p): Likewise.
33228 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33230 (vsx_tsqrt<mode>2_fe): Likewise.
33232 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33235 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33237 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33239 PR rtl-optimization/79780
33240 * cprop.c (one_cprop_pass): When second and further conditional trap
33241 in a single basic block is turned into an unconditional trap, turn it
33242 into a deleted note to avoid RTL verification failures.
33244 2017-03-02 Richard Biener <rguenther@suse.de>
33246 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33248 2017-03-02 Richard Biener <rguenther@suse.de>
33250 PR tree-optimization/79345
33252 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33253 param and abort the walk, returning -1 if it is hit.
33254 (walk_aliased_vdefs): Take a limit param and pass it on.
33255 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33256 defaulting to 0 and return a signed int.
33257 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33258 (check_defs): New helper.
33259 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33260 about uninitialized memory.
33261 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33262 bogus uninitialized warning.
33263 (fixed_convert_from_real): Likewise.
33265 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33267 PR tree-optimization/66768
33268 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33269 iv_use if base object can't be determined.
33271 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33273 PR tree-optimization/79345
33274 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33275 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33276 (get_pattern_stats): Initialize it.
33277 * genemit.c (gen_expand): Verify match_scratch numbers come after
33278 match_operand/match_dup numbers.
33279 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33280 match_scratch numbers.
33281 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33283 * config/s390/s390.md (trunctdsd2): Likewise.
33285 2017-03-02 Richard Biener <rguenther@suse.de>
33287 * wide-int.h (wide_int_storage::operator=): Implement in terms
33290 2017-03-02 Richard Biener <rguenther@suse.de>
33292 PR tree-optimization/79777
33293 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33294 the to insert expression to sth existing.
33296 2017-03-01 Martin Sebor <msebor@redhat.com>
33298 PR middle-end/79692
33299 * gimple-ssa-sprintf.c
33300 (directive::known_width_and_precision): New function.
33301 (format_integer): Use it.
33302 (get_mpfr_format_length): Consider the full range of precision
33303 when computing %g output with the # flag. Set the likely byte
33304 count to 3 rather than 1 when precision is indeterminate.
33305 (format_floating): Correct the lower bound of precision.
33307 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33309 * doc/invoke.texi: Document default code model for 64-bit Linux.
33311 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33314 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33315 udiv rather than div since input pattern is unsigned.
33317 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33319 * config/i386/i386.c (print_reg): Warn for values of
33320 unsupported size in integer register.
33322 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33325 * config/rs6000/predicates.md (current_file_function_operand): Do
33326 not allow self calls to be local if the function is replaceable.
33328 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33331 * config/rs6000/altivec.h (vec_ctz and others): Change the
33332 preprocessor macro that controls conditional compilation from
33333 _ARCH_PWR9 to __POWER9_VECTOR__.
33334 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33335 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33336 control (instead of _ARCH_PWR9 control) so that template
33337 definition uses power9-specific function.
33338 (vec_any_eq): Likewise.
33339 (vec_all_ne): Change macro definition to use a power9-specific
33340 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33341 _ARCH_PWR9 control).
33342 (vec_any_eq) Likewise.
33343 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33344 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33345 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33346 support for xvcmpnedp instruction.
33347 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33348 macro expansion so that Power9 implementation of vec_all_ne does
33349 not use the AltiVec predicate framework.
33350 (VCMPNEH_P): Likewise.
33351 (VCMPNEW_P): Likewise.
33352 (VCMPNED_P): Likewise.
33353 (VCMPNEFP_P): Likewise.
33354 (VCMPNEDP_P): Likewise.
33355 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33356 implementation of vec_any_eq to not use AltiVec predicate
33358 (VCMPAEH_P): Likewise.
33359 (VCMPAEW_P): Likewise.
33360 (VCMPAED_P): Likewise.
33361 (VCMPAEFP_P): Likewise.
33362 (VCMPAEDP_P): Likewise.
33363 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33364 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33365 not use the AltiVec predicate framework.
33366 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33367 of vec_any_eq to not use AltiVec predicate framework.
33368 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33369 support for predefined __POWER9_VECTOR__ macro to indicate that
33370 Power9 instruction selection is enabled.
33371 (altivec_overloaded_builtins): Remove extraneous
33372 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33373 function argument types RS6000_BTI_bool_V16QI and
33374 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33375 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33376 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33377 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33378 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33379 Power9 for implementations of vec_cmpne. Change the signature for
33380 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33381 (representing vec_all_ne) to remove the previously described first
33382 argument of type RS6000_BTI_INTSI, as this was an artifact of
33383 reliance on the AltiVec predicate framework, which is no longer
33384 used in the implementation of these functions. Add
33385 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33386 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33387 since, unlike the AltiVec predicate framework implementation, we
33388 do not share function descriptors between vec_alle and vec_anyeq.
33389 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33390 set of modes that receive special treatment even when
33391 TARGET_P9_VECTOR is true. The special treatment emits code that
33392 does not depend on Power9 instructions.
33393 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33394 define_expand to not rely on AltiVec predicate framework.
33395 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33397 (vector_ne_v2di_p): Change this define_expand to not rely on
33398 AltiVec predicate framework.
33399 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33401 (vector_ne_<mode>_p): Change this define_expand to not rely on
33402 AltiVec predicate framework.
33403 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33405 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33406 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33407 define_insn pattern.
33408 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33409 define_insn pattern because the xvcmpne<VSs>. instruction is not
33411 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33412 instruction is not supported.
33414 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33416 * config/nvptx/nvptx.c: Include intl.h.
33418 2017-03-01 Martin Jambor <mjambor@suse.cz>
33421 * ipa-prop.h (ipa_bits): Removed field known.
33422 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33423 to pointers. Adjusted their comments to warn about their sharing.
33424 (ipcp_transformation_summary): Change bits to a vector of pointers.
33425 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33426 (ipa_get_ipa_bits_for_value): Declare.
33427 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33428 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33429 (ipa_bits_hash_table): Likewise.
33430 (ipa_vr_ggc_hash_traits): Likewise.
33431 (ipa_vr_hash_table): Likewise.
33432 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33433 being pointers and vr_known being removed.
33434 (ipa_set_jf_unknown): Likewise.
33435 (ipa_get_ipa_bits_for_value): New function.
33436 (ipa_set_jfunc_bits): Likewise.
33437 (ipa_get_value_range): New overloaded functions.
33438 (ipa_set_jfunc_vr): Likewise.
33439 (ipa_compute_jump_functions_for_edge): Use the above functions to
33440 construct bits and vr parts of jump functions.
33441 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33442 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33444 (ipcp_grow_transformations_if_necessary): Also allocate
33445 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33447 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33448 them. Fix too long lines.
33449 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33450 vr_known being removed.
33451 (ipa_read_jump_function): Use new setter functions to construct bits
33452 and vr parts of jump functions or set them to NULL.
33453 (write_ipcp_transformation_info): Adjust for bits being pointers.
33454 (read_ipcp_transformation_info): Likewise.
33455 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33457 Include gt-ipa-prop.h.
33458 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33460 (ipcp_store_bits_results): Likewise.
33461 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33462 Do not write to existing jump functions but use a temporary instead.
33464 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33467 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33468 attempt to use its first operand as BIT_FIELD_REF base.
33470 2017-03-01 Richard Biener <rguenther@suse.de>
33472 PR middle-end/79721
33473 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33474 interpolating formula in wrapping arithmetic.
33475 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33477 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33479 PR tree-optimization/79734
33480 * tree-vect-generic.c (expand_vector_condition): Optimize
33481 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33482 Handle VEC_COND_EXPR where comparison has different inner width from
33483 type's inner width.
33485 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33487 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33488 markup, and similar issues. Remove @opindex entries for things
33489 that aren't options. Add missing -mmpy-option entries.
33491 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33493 PR tree-optimization/79737
33494 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33495 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33496 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33497 instead of byte_size. Formatting fix.
33498 (shift_bytes_in_array_right): Formatting fix.
33500 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33503 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33504 condition on optimize for the leaf function test.
33506 2017-02-28 Martin Liska <mliska@suse.cz>
33509 * read-rtl-function.c (function_reader::handle_unknown_directive):
33510 Bail out when one uses -flto.
33512 2017-02-28 Martin Liska <mliska@suse.cz>
33514 * common.opt: Replace space with tabular for options of <number>
33516 * config/i386/i386.opt: Show <number> value for
33517 -mlarge-data-threshold.
33518 * opts.c (print_filtered_help): Do not display number in hexadecimal
33521 2017-02-28 Martin Liska <mliska@suse.cz>
33523 * common.opt: Fix --help=option -Q for options which are of
33526 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33528 * config/i386/i386.c (print_reg): Error out for values
33529 of 8-bit size in invalid integer register.
33531 2017-02-28 Martin Sebor <msebor@redhat.com>
33533 PR tree-optimization/79691
33534 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33536 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33539 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33540 gcc_unreachable with output_operand_lossage.
33542 2017-02-28 Richard Biener <rguenther@suse.de>
33544 PR tree-optimization/79740
33545 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33547 (visit_nary_op): Insert the nary into the hashtable if we
33548 pattern-matched sth.
33549 * tree-ssa-pre.c (eliminate_insert): Robustify.
33551 2017-02-28 Richard Biener <rguenther@suse.de>
33553 PR middle-end/79731
33554 * fold-const.c (decode_field_reference): Reject out-of-bound
33557 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33559 * config/i386/i386.c: Include intl.h.
33560 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33561 instead of just cond ? "..." : "...".
33562 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33563 * coverage.c (read_counts_file): Likewise.
33564 * omp-offload.c: Include intl.h.
33565 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33566 of just cond ? "..." : "...".
33567 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33568 of just cond ? "..." : "...".
33570 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33573 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33575 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33576 'tune for' CPU name.
33577 * config/arm/arm-cpu-data.h: Regenerated.
33579 2017-02-28 Richard Biener <rguenther@suse.de>
33581 PR tree-optimization/79732
33582 * tree-inline.c (expand_call_inline): Do not shadow var.
33584 2017-02-28 Richard Biener <rguenther@suse.de>
33586 PR tree-optimization/79723
33587 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33588 address-space properly.
33590 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33592 * doc/optinfo.texi (Optimization groups): Fix option used for
33594 * doc/invoke.texi (-fopt-info): Document "omp".
33595 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
33596 (OPTGROUP_ALL): Add OPTGROUP_OMP.
33597 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
33598 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
33599 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
33601 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
33603 * dumpfile.c (optgroup_options): Instead of "openmp", associate
33604 OPTGROUP_OMP with "omp".
33606 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
33609 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
33610 for arithmetic shift of unsigned V2DI.
33612 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
33614 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
33615 arc/linux.h headers.
33616 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
33617 (LINK_SPEC): Likewise.
33618 (ARC_TLS_EXTRA_START_SPEC): Likewise.
33619 (EXTRA_SPECS): Likewise.
33620 (STARTFILE_SPEC): Likewise.
33621 (ENDFILE_SPEC): Likewise.
33622 (LIB_SPEC): Likewise.
33623 (TARGET_SDATA_DEFAULT): Likewise.
33624 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
33625 (MULTILIB_DEFAULTS): Likewise.
33626 (DWARF2_UNWIND_INFO): Likewise.
33627 * config/arc/big.h: New file.
33628 * config/arc/elf.h: Likewise.
33629 * config/arc/linux.h: Likewise.
33630 * config/arc/t-uClibc: Remove.
33632 2017-02-27 Bin Cheng <bin.cheng@arm.com>
33634 PR tree-optimization/77536
33635 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
33636 (tree_transform_and_unroll_loop): Use above function to compute the
33637 estimated niter of unrolled loop and use it when scaling profile.
33638 Also use count info rather than frequency if it's non-zero.
33639 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
33640 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
33641 (vect_transform_loop): Call above function.
33643 2017-02-27 Richard Biener <rguenther@suse.de>
33645 PR tree-optimization/45397
33646 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
33647 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
33648 (visit_nary_op): Add pattern matching for CSEing sign-changed
33649 or truncated operations with wider ones.
33651 2017-02-27 Richard Biener <rguenther@suse.de>
33653 PR tree-optimization/79690
33654 * tree-vect-stmts.c (vectorizable_store): Use vector type
33655 built from the DR with address-space.
33657 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
33659 * doc/invoke.texi (Optimize Options): Refine the description
33660 of asan-use-after-return.
33662 2017-02-25 Alan Modra <amodra@gmail.com>
33664 PR rtl-optimization/79584
33665 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
33666 base, not ad->base_term, the reg within base. Remove assertion
33667 that ad->base == ad->base_term. Replace gen_int_mode using
33668 bogus mode with const0_rtx.
33670 2017-02-25 Jakub Jelinek <jakub@redhat.com>
33672 PR middle-end/79396
33673 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
33674 FMA_EXPR like tcc_binary or tcc_unary.
33676 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
33679 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
33681 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
33682 (output_loc_operands): Handle DW_OP_call_ref and
33683 DW_OP_GNU_variable_value.
33684 (struct variable_value_struct): New type.
33685 (struct variable_value_hasher): Likewise.
33686 (variable_value_hash): New variable.
33687 (string_types): Remove.
33688 (copy_loc_descr): New function.
33689 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
33690 (prepend_loc_descr_to_each): New function.
33691 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
33692 instead of add_loc_descr_to_each if the first argument is single
33693 location list and the second has multiple.
33694 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
33695 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
33696 when looking for variable value which doesn't have other location info.
33697 (loc_list_from_tree): Formatting fix.
33698 (gen_array_type_die): Simplify DW_AT_string_length handling.
33699 (adjust_string_types): Remove.
33700 (gen_subprogram_die): Don't call adjust_string_types nor test/set
33701 string_types. Call resolve_variable_values.
33702 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
33703 (resolve_addr_in_expr): Likewise. Add A argument.
33704 (copy_deref_exprloc): Remove deref argument. Adjust for the
33705 original expression being DW_OP_GNU_variable_value with optionally
33706 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
33707 optionally after it.
33708 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
33709 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
33710 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
33711 (variable_value_hasher::hash, variable_value_hasher::equal): New
33713 (resolve_variable_value_in_expr, resolve_variable_value,
33714 resolve_variable_values, note_variable_value_in_expr,
33715 note_variable_value): New functions.
33716 (dwarf2out_early_finish): Call note_variable_value on all toplevel
33719 2017-02-24 Jakub Jelinek <jakub@redhat.com>
33722 * opts.h (handle_generated_option): Add GENERATED_P argument.
33723 * opts-common.c (handle_option): Adjust function comment.
33724 (handle_generated_option): Add GENERATED_P argument, pass it to
33726 (control_warning_option): Pass false to handle_generated_option
33728 * opts.c (maybe_default_option): Pass true to handle_generated_option
33730 * optc-gen.awk: Likewise.
33732 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33734 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
33735 a REG, look at the REG it is a SUBREG of.
33736 (splitter for cmpeqsi_t): Ditto.
33738 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33740 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
33741 the special USEs with the pattern of the insn, not the insn itself.
33743 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
33746 * doc/invoke.texi: Document -mload-store-pairs.
33748 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33749 Sandra Loosemore <sandra@codesourcery.com>
33751 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
33752 argument isn't a CONST_INT.
33753 (nios2_alternate_compare_const): Assert op is a CONST_INT.
33754 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
33755 (nios2_validate_compare): Bypass alternate compare logic if *op2
33756 is not a CONST_INT.
33757 (ldstwm_operation_p): Return false if first_base is not a REG or
33758 if first_offset is not a CONST_INT.
33760 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33762 * config/cris/cris.md: Use correct operand in a define_peephole2.
33764 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33766 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
33768 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33770 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
33771 this_insn if it is an INSN or JUMP_INSN.
33772 (force_offsettable): Look at base, not at addr.
33773 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
33774 on things that aren't necessarily CONST_INTs.
33776 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
33778 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
33779 -mfpmath=sse is the default also for x86-32 targets with SSE2
33780 instruction set when @option{-ffast-math} is enabled
33782 2017-02-24 Jeff Law <law@redhat.com>
33784 PR rtl-optimizatoin/79286
33785 * ira.c (update_equiv_regs): Drop may_trap_p exception to
33788 2017-02-24 Richard Biener <rguenther@suse.de>
33790 PR tree-optimization/79389
33791 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
33794 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
33796 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
33797 function comment to reflect reality.
33798 (loop_exits_before_overflow): Fix typo in function description.
33800 2017-02-24 Richard Biener <rguenther@suse.de>
33802 PR tree-optimization/79389
33803 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
33804 properly that a threading opportunity exists. Detect conditional
33805 copy/constant propagation opportunities.
33807 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
33809 * config/visium/visium.md (type): Add trap.
33810 (b): New mode attribute.
33811 (*btst): Rename into...
33812 (*btst<mode>): ...this and adjust.
33813 (*cbranchsi4_btst_insn): Rename into...
33814 (*cbranch<mode>4_btst_insn): ...this and adjust.
33815 (trap): New define_insn.
33817 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33819 PR tree-optimization/79389
33820 * ifcvt.c (struct noce_if_info): Add rev_cond field.
33821 (noce_reversed_cond_code): New function.
33822 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
33823 reversed_comparison_code. Formatting fix.
33824 (noce_try_store_flag): Test rev_cond != NULL in addition to
33825 reversed_comparison_code.
33826 (noce_try_store_flag_constants): Likewise.
33827 (noce_try_store_flag_mask): Likewise.
33828 (noce_try_addcc): Use rev_cond if non-NULL instead of
33829 reversed_comparison_code.
33830 (noce_try_cmove_arith): Likewise. Formatting fixes.
33831 (noce_try_minmax, noce_try_abs): Clear rev_cond.
33832 (noce_find_if_block): Initialize rev_cond.
33833 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
33834 instead of false as last argument never attempt to reverse it
33837 2017-02-23 Bin Cheng <bin.cheng@arm.com>
33839 PR tree-optimization/79663
33840 * tree-predcom.c (combine_chains): Process refs in reverse order
33841 only for ZERO length chains, and add explaining comment.
33843 2017-02-23 Jeff Law <law@redhat.com>
33845 PR tree-optimization/79578
33846 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
33847 in call to operand_equal_p.
33849 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
33852 * config/i386/cpuid.h: Fix another undefined behavior.
33854 2017-02-23 Richard Biener <rguenther@suse.de>
33856 PR tree-optimization/79683
33857 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
33858 vector types for data-refs.
33860 2017-02-23 Martin Liska <mliska@suse.cz>
33862 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
33864 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33866 PR middle-end/79665
33867 * internal-fn.c (get_range_pos_neg): Moved to ...
33868 * tree.c (get_range_pos_neg): ... here. No longer static.
33869 * tree.h (get_range_pos_neg): New prototype.
33870 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
33871 are known to be in between 0 and signed maximum inclusive, try to
33872 expand both unsigned and signed divmod and use the cheaper one from
33875 2017-02-22 Jeff Law <law@redhat.com>
33877 PR tree-optimization/79578
33878 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
33879 to compare base operands.
33881 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
33884 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
33885 gpc_reg_operand instead of fpr_reg_operand.
33887 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
33889 * config/mips/mips.c (mips_return_in_memory): Force FP
33890 vector types to be returned in memory for o32 ABI.
33892 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33894 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
33895 instead of DW_TAG_member for static data member declarations and don't
33896 set no_linkage_name for static inline data members.
33897 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
33900 2017-02-22 Martin Liska <mliska@suse.cz>
33902 * doc/invoke.texi: Replace inequality signs with square brackets
33905 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
33908 * lra-constraints.c (simplify_operand_subreg): Handle
33909 WORD_REGISTER_OPERATIONS targets.
33911 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33914 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
33915 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
33916 elimination by swapping fld*.
33918 2017-02-22 Richard Biener <rguenther@suse.de>
33920 PR tree-optimization/79673
33921 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
33922 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
33923 irrelevant address-space qualifiers and avoiding a
33924 ADDR_SPACE_CONVERT_EXPR from fold_convert.
33926 2017-02-22 Richard Biener <rguenther@suse.de>
33928 PR tree-optimization/79666
33929 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
33930 to not symbolically negate if that may introduce undefined
33933 2017-02-22 Martin Liska <mliska@suse.cz>
33936 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
33937 * data-streamer-out.c (streamer_write_gcov_count_stream):
33939 * value-prof.c (stream_out_histogram_value): Make assert more
33940 precise based on type of counter.
33942 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
33945 * config/i386/i386.md (standard_x87sse_constant_load splitter):
33946 Use nonimmediate_operand instead of memory_operand for operand 1.
33947 (float-extend standard_x87sse_constant_load splitter): Ditto.
33949 2017-02-21 Jeff Law <law@redhat.com>
33951 PR tree-optimization/79621
33952 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
33953 blocks with edges to themselves.
33955 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33958 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
33959 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
33960 Use gimple_call_builtin_p.
33963 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
33964 on temporarily removed DEBUG_INSNs.
33966 PR tree-optimization/79649
33967 * tree-loop-distribution.c (classify_partition): Give up on
33968 non-generic address space loads/stores.
33970 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
33972 * doc/loop.texi (Loop manipulation): Remove nonexistent
33973 tree_ssa_loop_version from the documentation.
33974 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
33976 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33979 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
33980 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
33981 * config/rs6000/rs6000.c: Include except.h.
33982 (rs6000_expand_split_stack_prologue): Call
33983 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
33985 2017-02-21 Martin Jambor <mjambor@suse.cz>
33988 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
33989 have been analyzed.
33991 2017-02-21 Martin Jambor <mjambor@suse.cz>
33993 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
33994 for backward compatibility only.
33995 * doc/invoke.texi (Option Summary): Remove all references to
33996 -fipa-cp-alignment.
33998 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34002 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34004 * lra-constraints.c (curr_insn_transform): Handle
34005 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34007 2017-02-21 Martin Liska <mliska@suse.cz>
34009 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34011 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34013 PR translation/79638
34014 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34016 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34019 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34020 (arm_function_ok_for_sibcall): Return false for an indirect call by
34021 descriptor if all the argument registers are used.
34022 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34023 alignment of the function.
34025 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34027 PR tree-optimization/61441
34028 * simplify-rtx.c (simplify_const_unary_operation): For
34029 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34030 the sNaN unmodified.
34032 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34034 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34035 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34036 instead of SYSTEM_HEADER_DIR.
34038 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34039 Martin Liška <mliska@suse.cz>
34041 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34042 Fix typos and grammar, use active voice, and clarify.
34044 2017-02-20 Marek Polacek <polacek@redhat.com>
34046 PR middle-end/79537
34047 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34050 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34052 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34055 * config/i386/i386.c (ix86_expand_builtin): Handle
34056 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34057 ix86_builtins_isa[fcode].isa as a requirement of those
34058 flags and any other flag in the bitmask.
34059 (ix86_init_mmx_sse_builtins): Use 0 instead of
34060 ~OPTION_MASK_ISA_64BIT as mask.
34061 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34062 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34063 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34064 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34066 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34069 * lra-constraints.c (split_reg): Check requested split mode
34070 is supported by the register.
34072 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34074 * lra-constraints.c (simplify_operand_subreg): Remove early
34077 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34080 * lra-constraints.c (curr_insn_transform): Tighten condition
34081 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34083 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34086 * lra-constraints.c (curr_insn_transform): Handle
34087 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34089 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34092 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34094 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34096 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34099 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34102 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34104 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34107 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34109 * final.c (last_columnnum, override_columnnum): New variables.
34110 (final_start_function): Set last_columnnum, pass it to begin_prologue
34111 hook and pass 0 to dwarf2out_begin_prologue.
34112 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34113 to source_line debug hook.
34114 (notice_source_line): Compute last_columnnum and for debug_column_info
34115 return true on column changes.
34116 * debug.h (struct gcc_debug_hooks): Add column argument to
34117 source_line and begin_prologue hooks.
34118 (debug_nothing_int_charstar_int_bool): Remove prototype.
34119 (debug_nothing_int_int_charstar,
34120 debug_nothing_int_int_charstar_int_bool): New prototypes.
34121 (dwarf2out_begin_prologue): Add column argument.
34122 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34123 begin_prologue hooks.
34124 (debug_nothing_int_charstar_int_bool): Remove.
34125 (debug_nothing_int_int_charstar,
34126 debug_nothing_int_int_charstar_int_bool): New functions.
34127 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34128 through to dwarf2out_source_line.
34129 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34130 (dwarf2out_source_line): Add column argument, emit it if requested.
34131 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34133 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34134 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34135 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34136 through to dwarf2out_begin_prologue.
34137 (vmsdbgout_source_line): Add column argument, pass it through to
34138 dwarf2out_source_line.
34139 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34140 dbxout_source_line caller.
34141 (dbxout_source_line): Add column argument.
34143 * common.opt (gno-column-info, gcolumn-info): New options.
34144 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34145 (check_die): Also test for multiple DW_AT_decl_column attributes.
34146 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34147 DW_AT_decl_column if requested.
34148 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34150 (gen_variable_die): Likewise.
34151 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34152 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34155 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34156 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34157 (ix86_handle_option): Handle OPT_m3dnowa.
34158 * doc/invoke.texi (-m3dnowa): Document.
34159 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34160 -m3dnowa instead of -m3dnow -march=athlon.
34163 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34164 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34166 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34169 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34170 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34171 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34172 generator for vsx_xxpermdi_<mode>_be.
34173 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34174 force big-endian semantics.
34175 (vsx_xxpermdi_<mode>_be): New define_expand with same
34176 implementation as previous version of vsx_xxpermdi_<mode>.
34178 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34180 PR tree-optimization/79327
34181 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34182 variable, its initialization and use.
34184 2017-02-17 Julia Koval <julia.koval@intel.com>
34186 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34187 (OPTION_MASK_ISA_PKU_UNSET): New.
34188 (ix86_handle_option): Handle -mrdpid.
34189 * config/i386/cpuid.h (bit_RDPID): New.
34190 * config/i386/driver-i386.c (host_detect_local_cpu):
34191 Detect RDPID feature.
34192 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34193 * config/i386/i386-c.c (ix86_target_macros_internal):
34195 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34196 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34197 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34198 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34199 * config/i386/i386.md (define_insn "rdpid"): New.
34200 * config/i386/i386.opt Add -mrdpid.
34201 * config/i386/immintrin.h (_rdpid_u32): New.
34203 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34205 PR rtl-optimization/79541
34206 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34207 instead of transforming it into USE.
34209 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34211 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34212 If HONOR_SNANS (SFmode) force the input to a register.
34213 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34214 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34215 an frsp or similar insn.
34217 2017-02-17 Martin Liska <mliska@suse.cz>
34219 PR rtl-optimization/79577
34220 * params.def (selsched-max-sched-times): Increase minimum to 1.
34222 2017-02-17 Martin Liska <mliska@suse.cz>
34224 PR rtl-optimization/79574
34225 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34227 2017-02-17 Martin Liska <mliska@suse.cz>
34229 PR tree-optimization/79529
34230 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34231 ssa_defined_default_def_p to handle cases which are implicitly
34233 * tree-ssa.c (ssa_defined_default_def_p): New function.
34234 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34235 which are implicitly defined.
34236 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34238 2017-02-17 Richard Biener <rguenther@suse.de>
34240 PR middle-end/79576
34241 * params.def (max-ssa-name-query-depth): Limit to 10.
34243 2017-02-17 Richard Biener <rguenther@suse.de>
34245 PR tree-optimization/79552
34246 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34249 2017-02-17 Richard Biener <rguenther@suse.de>
34252 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34254 2017-02-17 Marek Polacek <polacek@redhat.com>
34256 PR middle-end/79536
34257 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34258 (fold_negate_expr): New wrapper.
34260 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34262 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34263 Correct terminology and de-emphasize pre-standard behavior.
34265 2017-02-16 Alan Modra <amodra@gmail.com>
34267 PR rtl-optimization/79286
34268 * ira.c (def_dominates_uses): New function.
34269 (update_equiv_regs): Don't create an equivalence for insns that
34270 may trap where the register def does not dominate the use.
34272 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34274 PR rtl-optimization/78127
34275 * lra.c (lra): Call lra_eliminate before finish the loop after
34278 2017-02-16 Richard Biener <rguenther@suse.de>
34280 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34282 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34283 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34284 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34285 (isl_val_int_from_wi): New function.
34286 (extract_affine_gmp): Rename to ...
34287 (extract_affine_wi): ... this, take a widest_int.
34288 (extract_affine_int): Just wrap extract_affine_wi.
34289 (add_param_constraints): Use isl_val_int_from_wi.
34290 (add_loop_constraints): Likewise, and extract_affine_wi.
34292 2017-02-15 Jeff Law <law@redhat.com>
34294 PR middle-end/79521
34295 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34296 ira_init_register_move_cost_if_necessary.
34298 2017-02-15 Martin Sebor <msebor@redhat.com>
34300 PR middle-end/32003
34301 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34302 removed in a prior commit.
34304 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34306 PR tree-optimization/79347
34307 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34308 counters during peeling.
34310 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34312 * Makefile.in (site.exp): Remove "set ISLVER".
34314 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34317 * real.c (real_from_integer): Call real_convert even for decimal.
34319 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34322 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34324 2017-02-14 Andrew Pinski <apinski@cavium.com>
34326 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34327 cores and change the partno/implementer to be correct.
34328 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34329 the 'B" as the implementer.
34330 * config/aarch64/aarch64-tune.md: Regenerate.
34332 2017-02-14 Carl Love <cel@us.ibm.com>
34334 * config/rs6000/rs6000.c: Add case statement entry to make the
34335 xvcvuxdsp built-in argument unsigned.
34336 * config/rs6000/vsx.md: Fix the source and return operand types so they
34337 match the instruction definitions from the ISA document. Fix typo
34338 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34341 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34344 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34345 member early_clobber_alts.
34346 * lra-lives.c (reg_early_clobber_p): New.
34347 (process_bb_lives): Use it.
34348 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34349 (debug_operand_data): Initialize early_clobber_alts.
34350 (setup_operand_alternative): Set up early_clobber_alts.
34351 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34352 alternatives to new_insn_reg.
34353 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34355 (lra_update_insn_regno_info): Pass the new arg.
34357 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34359 PR middle-end/79505
34360 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34361 (new_oacc_loop_raw): Don't clear already cleared fields.
34364 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34365 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34366 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34368 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34371 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34373 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34376 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34377 the extra instruction to the right place to store 128-bit constant
34380 2017-02-14 Martin Sebor <msebor@redhat.com>
34382 PR middle-end/79448
34383 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34384 warning for strings of unknown length.
34386 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34388 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34390 2017-02-14 Jeff Law <law@redhat.com>
34393 * ira-costs.c (scan_one_insn): Initialize register move costs
34394 for pseudos seen in USE/CLOBBER insns.
34396 PR tree-optimization/79095
34397 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34398 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34399 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34400 if the operands are known to be not equal, then the resulting range
34402 (intersect_ranges): If the new range is ~[0,0] and the old range is
34403 wide, then prefer ~[0,0].
34404 * tree-vrp.c (overflow_comparison_p_1): New function.
34405 (overflow_comparison_p): New function.
34406 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34407 if NAME is used in an overflow test.
34408 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34409 overflow check that can be expressed as an equality test, then adjust
34410 ops to be that equality test.
34412 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34414 * config/s390/s390-builtin-types.def: Remove flags argument.
34415 * config/s390/s390.c (s390_init_builtins): Likewise.
34417 2017-02-14 Martin Liska <mliska@suse.cz>
34419 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34420 vector. Fix trailing white spaces.
34422 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34424 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34427 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34429 PR rtl-optimization/68664
34430 * config/arm/arm.c (arm_sched_can_speculate_insn):
34431 New function. Declare prototype.
34432 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34434 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34436 PR rtl-optimization/68664
34437 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34439 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34441 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34443 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34444 max skip bytes for function, loop and jump.
34446 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34448 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34449 ABS_EXPR for gimple dump.
34451 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34454 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34456 PR tree-optimization/79408
34457 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34458 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34459 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34460 also if rhs1 is INTEGER_CST.
34462 2017-02-14 Richard Biener <rguenther@suse.de>
34464 PR middle-end/79432
34465 * tree-into-ssa.c (insert_phi_nodes): When the function can
34466 have abnormal edges rewrite SSA names with broken use-def
34467 dominance out of SSA and register them for PHI insertion.
34469 2017-02-13 Martin Sebor <msebor@redhat.com>
34471 PR middle-end/79496
34472 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34473 clearing info.nowrite flag when snprintf size argument is a range.
34475 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34477 * cprop.c (cprop_jump): Add missing space in string literal.
34478 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34479 (get_constraint_for_component_ref): Likewise.
34480 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34481 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34482 * lra-constraints.c (process_alt_operands): Likewise.
34483 * ipa-inline.c (inline_small_functions): Likewise.
34484 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34485 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34486 * trans-mem.c (diagnose_tm_1_op): Likewise.
34487 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34488 (grid_parallel_clauses_gridifiable): Likewise.
34490 * config/nvptx/mkoffload.c (process): Add space in between
34493 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34494 "MOD4_SSE_REGS" and "ALL_REGS".
34496 * spellcheck.c (test_data): Add , in between "foo" and "food".
34498 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34501 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34502 boundary crossing check and subsequent code generation agree.
34504 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34506 * config/aarch64/aarch64.c (has_memory_op): Delete.
34507 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34510 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34512 PR rtl-optimization/79388
34513 PR rtl-optimization/79450
34514 * combine.c (distribute_notes): When removing TEM_INSN for which
34515 corresponding dest has last value recorded, invalidate that last
34518 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34520 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34521 of explicit '@'. Add missing assembly comment marker on branch costs
34524 2017-02-13 Nathan Sidwell <nathan@acm.org>
34526 * gengtype-lex.l (<in_struct>): Add '/'.
34528 2017-02-13 Martin Liska <mliska@suse.cz>
34531 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34533 2017-02-13 Richard Biener <rguenther@suse.de>
34535 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34537 * configure: Re-generate.
34538 * config.in: Likewise.
34539 * graphite-dependences.c: Simplify as if
34540 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34541 * graphite-isl-ast-to-gimple.c: Likewise.
34542 * graphite-optimize-isl.c: Likewise.
34543 * graphite-poly.c: Likewise.
34544 * graphite-sese-to-poly.c: Likewise.
34545 * graphite.h: Likewise.
34546 * toplev.c: Include isl/version.h and use isl_version () for
34547 printing the ISL version.
34548 * doc/install.texi: Update ISL requirement.
34550 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34552 * doc/standards.texi (Standards): Update reference to
34555 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34557 * doc/extend.texi (Named Address Spaces): sourceware.org now
34559 * doc/install.texi (Binaries): Ditto.
34562 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34564 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34565 standard terminology "stringize"/"stringizing" throughout.
34566 * doc/cppinternals.texi: Likewise.
34568 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34570 * doc/extend.texi: Fix some spelling mistakes and typos.
34571 * doc/invoke.texi: Likewise.
34573 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34576 * params.def (inline-min-speedup) Change from 10 to 8.
34578 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34580 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34583 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34586 * ipa-inline-analysis.c (get_minimal_bb): New function.
34587 (record_modified): Use it.
34588 (remap_edge_change_prob): Handle also ancestor functions.
34590 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34592 * doc/contrib.texi (Contributors): Remove broken link into
34593 the Mauve CVS repository.
34595 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34597 PR middle-end/79454
34598 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
34599 result computation whenever lhs doesn't have vector mode, not
34600 just when it has BLKmode.
34602 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34604 * doc/makefile.texi (profiledbootstrap): Refer to the
34605 installation instructions only in textual form.
34607 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34610 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
34612 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34614 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
34615 (Specific): Update mingw-w64 reference.
34617 (Specific): Remove broken link to Renesas RX processor.
34619 2017-02-10 Richard Biener <rguenther@suse.de>
34621 * toplev.c (process_options): Do not mention obsolete graphite
34622 options when printing sorry message about missing graphite support.
34623 Mention -floop-nest-optimize.
34625 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
34627 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
34628 (vtst_p16): Likewise.
34629 (vtstq_p8): Likewise.
34630 (vtstq_p16): Likewise.
34632 (vtstq_p64): Likewise.
34633 * config/arm/arm_neon.h (vgetq_lane_p64): New.
34634 (vset_lane_p64): New.
34635 (vsetq_lane_p64): New.
34637 2017-02-10 Jakub Jelinek <jakub@redhat.com>
34639 PR tree-optimization/79411
34640 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
34641 stmt operands are SSA_NAMEs used in abnormal phis.
34642 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
34645 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34648 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
34651 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34653 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
34655 2017-02-09 Jakub Jelinek <jakub@redhat.com>
34657 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
34661 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
34662 not arbitrary TREE_CONSTANT.
34665 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
34666 "omp declare target link" attribute unless is_global_var.
34667 * omp-offload.c (find_link_var_op): Likewise.
34669 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
34670 Chung-Lin Tang <cltang@codesourcery.com>
34672 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
34674 (gimplify_adjust_omp_clauses): Don't delete TILE.
34675 (gimplify_omp_for): Deal with TILE.
34676 * internal-fn.c (expand_GOACC_TILE): New function.
34677 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
34679 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
34680 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
34682 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
34683 avoid DIV for outermost collapse var.
34684 (expand_oacc_for): Insert tile element loop as needed. Adjust.
34685 Remove out of date comments, fix whitespace.
34686 * omp-general.c (omp_extract_for_data): Deal with tiling.
34687 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
34688 adjust OLF_DIM_BASE value.
34689 (struct omp_for_data): Add tiling field.
34690 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
34691 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
34692 for auto loops. Remove default auto determining, moved to
34693 oacc_loop_fixed_partitions.
34694 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
34695 stmts, add e_mask field.
34696 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
34697 (oacc_thread_numbers): Use oacc_dim_call.
34698 (oacc_xform_tile): New.
34699 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
34700 (finish_oacc_loop): Adjust for ifns vector.
34701 (oacc_loop_discover_walk): Append loop abstraction sites to list,
34702 add case for GOACC_TILE fns.
34703 (oacc_loop_xform_loop): Delete.
34704 (oacc_loop_process): Iterate over call list directly, and add
34705 handling for GOACC_TILE fns.
34706 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
34708 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
34709 vector partitioning to outer loops. Assign 2 partitions to loops
34710 when available. Add TILE handling.
34711 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
34712 (execite_oacc_device_lower): Process GOACC_TILE fns,
34713 ignore unknown specs.
34714 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
34715 * tree.c (omp_clause_num_ops): Adjust TILE ops.
34716 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
34718 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
34720 * configure.ac (ACX_BUGURL): Update.
34721 * configure: Regenerate.
34723 2017-02-09 Richard Biener <rguenther@suse.de>
34725 PR tree-optimization/69823
34726 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
34727 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
34729 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34731 * config/arc/arc-c.def: Add __NPS400__ definition.
34732 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
34733 (TARGET_NPS400): Define.
34735 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34737 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
34739 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
34740 pointer, arch_info.
34741 (arc_cpu_types): Fill the arch_info field with a pointer into the
34742 arc_arch_types table.
34743 (arc_selected_cpu): Declare.
34744 * config/arc/arc.c (arc_selected_cpu): Make global.
34745 (arc_selected_arch): Delete.
34746 (arc_base_cpu): Delete.
34747 (arc_override_options): Remove references to deleted variables,
34748 update access to arch information.
34749 (ARC_OPT): Update access to arch information.
34750 (ARC_OPTX): Likewise.
34751 * config/arc/arc.h (arc_base_cpu): Remove declaration.
34752 (TARGET_ARC600): Update access to arch information.
34753 (TARGET_ARC601): Likewise.
34754 (TARGET_ARC700): Likewise.
34755 (TARGET_EM): Likewise.
34756 (TARGET_HS): Likewise.
34757 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
34760 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
34763 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
34764 condition/operands for integer GE/LE/GEU/LEU operations.
34766 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
34768 PR translation/79397
34769 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
34772 2017-02-08 Martin Jambor <mjambor@suse.cz>
34775 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
34776 whether allocation happened.
34777 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
34778 nothing was allocated.
34780 2017-02-08 Jakub Jelinek <jakub@redhat.com>
34782 PR tree-optimization/79408
34783 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
34784 constant, but SSA_NAME with a known integer range, use the minimum
34785 of that range instead of op1 to determine if modulo can be replaced
34786 with its first operand.
34788 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34790 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
34792 2017-02-08 Richard Biener <rguenther@suse.de>
34794 PR tree-optimization/71824
34795 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34796 Check all loops contained in the merged region.
34798 2017-02-07 Andrew Pinski <apinski@cavium.com>
34800 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
34802 2017-02-07 Andrew Pinski <apinski@cavium.com>
34804 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
34805 (thunderxt88): Likewise.
34806 (thunderxt81): Disable LSE and change v8.1 to v8.
34807 (thunderxt83): Likewise.
34809 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34810 Richard Biener <rguenther@suse.de>
34812 PR middle-end/79399
34813 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
34814 type from int to size_t.
34815 * ira-costs.c (struct_costs_size): Change type from int to size_t.
34817 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34819 PR rtl-optimization/79386
34820 * cprop.c (bypass_conditional_jumps): Initialize
34821 bypass_last_basic_block already before splitting bbs after
34822 unconditional traps...
34823 (bypass_conditional_jumps): ... rather than here.
34826 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
34827 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
34828 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
34829 fix -masm=intel patterns.
34831 2017-02-07 Richard Biener <rguenther@suse.de>
34833 PR tree-optimization/79256
34834 PR middle-end/79278
34835 * builtins.c (get_object_alignment_2): Use min_align_of_type
34836 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
34837 and ADJUST_FIELD_ALIGN.
34839 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
34841 * doc/tm.texi: Regenerate.
34842 * stor-layout.c (layout_decl): Adjust.
34843 (update_alignment_for_field): Likewise.
34844 (place_field): Likewise.
34845 (min_align_of_type): Likewise.
34846 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
34847 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
34848 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
34849 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
34850 * config/frv/frv.c (frv_adjust_field_align): Likewise.
34851 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
34852 * config/i386/i386.c (x86_field_alignment): Likewise.
34853 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
34854 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
34855 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
34856 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
34857 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
34858 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
34862 2017-01-30 Richard Biener <rguenther@suse.de>
34864 PR tree-optimization/79256
34865 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34866 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34869 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
34871 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
34872 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
34873 builtins to SImode and emit a zero-extend, if necessary.
34875 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34877 * docs/invoke.texi (RISC-V Options): Alphabetize.
34879 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34881 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
34884 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34886 * config/riscv/riscv.c: New file.
34887 * common/config/riscv/riscv-common.c: Likewise.
34888 * config.gcc: Likewise.
34889 * config/riscv/constraints.md: Likewise.
34890 * config/riscv/elf.h: Likewise.
34891 * config/riscv/generic.md: Likewise.
34892 * config/riscv/linux.h: Likewise.
34893 * config/riscv/multilib-generator: Likewise.
34894 * config/riscv/peephole.md: Likewise.
34895 * config/riscv/pic.md: Likewise.
34896 * config/riscv/predicates.md: Likewise.
34897 * config/riscv/riscv-builtins.c: Likewise.
34898 * config/riscv/riscv-c.c: Likewise.
34899 * config/riscv/riscv-ftypes.def: Likewise.
34900 * config/riscv/riscv-modes.def: Likewise.
34901 * config/riscv/riscv-opts.h: Likewise.
34902 * config/riscv/riscv-protos.h: Likewise.
34903 * config/riscv/riscv.h: Likewise.
34904 * config/riscv/riscv.md: Likewise.
34905 * config/riscv/riscv.opt: Likewise.
34906 * config/riscv/sync.md: Likewise.
34907 * config/riscv/t-elf-multilib: Likewise.
34908 * config/riscv/t-linux: Likewise.
34909 * config/riscv/t-linux-multilib: Likewise.
34910 * config/riscv/t-riscv: Likewise.
34911 * configure.ac: Likewise.
34912 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
34913 Waterman as RISC-V maintainers.
34914 * doc/install.texi: Add RISC-V entries.
34915 * doc/invoke.texi: Add RISC-V options section.
34916 * doc/md.texi: Add RISC-V constraints section.
34917 * configure: Regenerated.
34919 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
34922 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
34923 false values to be constant vectors with all 0 or all 1 bits set.
34924 (vcondu<mode><mode>): Likewise.
34925 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
34927 (fpmask_comparison_operator): Update comment.
34928 (vecint_comparison_operator): New predicate.
34929 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
34930 vector conditionals when the true and false values are constant
34931 vectors with all 0 bits or all 1 bits set.
34933 2017-02-06 Martin Sebor <msebor@redhat.com>
34935 PR tree-optimization/79376
34936 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
34938 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
34940 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
34941 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
34942 to simplify split condition.
34944 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34946 * omp-expand.c (oxpand_omp_atomic_fetch_op,
34947 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
34950 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
34952 PR rtl-optimization/68664
34953 * target.def (can_speculate_insn): New hook.
34954 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
34955 * doc/tm.texi: Regenerate.
34956 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
34957 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
34958 (rs6000_sched_can_speculate_insn): New function.
34960 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34962 PR tree-optimization/79284
34963 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
34964 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
34965 vectorizable_mask_load_store, vectorizable_operation,
34966 vect_is_simple_cond, get_same_sized_vectype): Use it instead
34967 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
34968 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
34969 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
34970 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
34971 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
34972 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
34973 is_gimple_assign (stmt). Replace another such test with
34974 is_gimple_assign (stmt).
34976 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34979 * config/avr/avr.c (rtl-iter.h): Include it.
34980 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
34981 (avr_legitimate_combined_insn): ...and implementation.
34983 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
34985 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
34986 * config/s390/s390.c (s390_const_operand_ok)
34987 (s390_canonicalize_comparison, s390_extract_part)
34988 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
34989 (s390_contiguous_bitmask_p, s390_rtx_costs)
34990 (legitimize_pic_address): Likewise.
34991 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
34992 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
34993 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
34994 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
34995 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
34997 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34999 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35000 REGNO($0) == REGNO($1).
35002 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35004 * config/s390/linux.h(SIZE_TYPE): Add comment.
35006 2017-02-06 Julian Brown <julian@codesourcery.com>
35007 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35008 Virendra Pathak <virendra.pathak@broadcom.com>
35010 * config/aarch64/aarch64-cores.def: Change the scheduler
35012 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35013 * config/aarch64/thunderx2t99.md: New file.
35015 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35017 * doc/standards.texi (Go Language): Update link to language
35020 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35022 * tree-eh.c (lower_resx): Sanitize profile.
35023 (cleanup_empty_eh_move_lp): Likewise.
35025 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35028 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35030 * cfgloopmanip.h (loop_version): Update prototype.
35031 * modulo-sched.c (sms_schedule): Update call of loop_version.
35032 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35033 * tree-parloops.c (gen_parallel_loop): Likewise.
35034 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35035 * tree-ssa-loop-split.c (split_loop): Likewise.
35036 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35037 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35039 2017-02-05 Martin Liska <mliska@suse.cz>
35042 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35044 (print_operand_address): Initialize a struct to zero.
35046 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35048 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35049 garbage collector only in textual form.
35051 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35053 * doc/extend.texi (x86 specific memory model extensions for
35054 transactional memory): Simplify a phrase.
35056 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35059 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35060 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35061 (atomic_storedi_1): Likewise.
35063 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35065 PR tree-optimization/79338
35066 * tree-parloops.c (gather_scalar_reductions): Don't call
35067 vect_analyze_loop_form for loop->inner before destroying loop's
35070 2017-02-03 Martin Sebor <msebor@redhat.com>
35072 PR tree-optimization/79327
35073 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35074 when precision has resulted in leading zeros.
35075 (format_integer): Adjust the likely counter to assume an unknown
35076 argument that may be zero is non-zero.
35078 2017-02-03 Jason Merrill <jason@redhat.com>
35081 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35082 avoid copying non-taken branch.
35084 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35086 PR tree-optimization/79340
35087 * tree-vect-loop.c (vectorizable_reduction): Release
35088 vec_defs elements after safe_splicing them into other vectors.
35091 PR tree-optimization/79327
35092 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35093 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35095 (format_integer): Use wide_int_to_tree instead of build_int_cst
35096 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35097 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35098 of shortest and longest sequence.
35100 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35102 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35103 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35105 2017-02-03 Walter Lee <walt@tilera.com>
35108 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35109 after initial stackframe link reg save.
35110 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35112 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35115 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35116 wu for stxssp alternative.
35118 2017-02-03 Martin Sebor <msebor@redhat.com>
35120 PR tree-optimization/79352
35121 * gimple-fold.c (get_range_strlen): Add argument.
35122 (get_range_strlen): Change return type to bool.
35123 (get_maxval_strlen): Pass in a dummy argument.
35124 * gimple-fold.h (get_range_strlen): Change return type to bool.
35125 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35126 * tree.h (array_at_struct_end_p): Add argument.
35127 * tree.c (array_at_struct_end_p): Handle it.
35129 2017-02-03 Martin Liska <mliska@suse.cz>
35132 * multiple_target.c (create_dispatcher_calls): Redirect edge
35133 from a caller of a dispatcher.
35134 (expand_target_clones): Make the clones local.
35135 (ipa_target_clone): Do both target clones and resolvers.
35136 (ipa_dispatcher_calls): Remove the pass.
35137 (pass_dispatcher_calls::gate): Likewise.
35138 (make_pass_dispatcher_calls): Likewise.
35139 * passes.def (pass_target_clone): Put as very first IPA early
35142 2017-02-03 Martin Liska <mliska@suse.cz>
35144 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35145 in case of a function with ifunc attribute.
35147 2017-02-03 Martin Liska <mliska@suse.cz>
35149 * cgraph.c (cgraph_node::dump): Dump function version info.
35150 * symtab.c (symtab_node::dump_base): Add missing new line.
35152 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35154 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35155 (ifcombine_ifandif): Use it.
35157 2017-02-03 Martin Liska <mliska@suse.cz>
35159 * doc/invoke.texi: Document default value for
35160 use-after-scope-direct-emission-threshold.
35162 2017-02-03 Martin Liska <mliska@suse.cz>
35164 PR tree-optimization/79339
35165 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35166 (format_floating): Likewise.
35168 2017-02-03 Martin Liska <mliska@suse.cz>
35171 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35173 (ipa_node_params_t::remove): Likewise.
35174 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35175 initialization from removed ipa_node_params_t::insert.
35176 (ipa_node_params::~ipa_node_params): Move from removed
35177 ipa_node_params_t::release.
35178 * symbol-summary.h (symbol_summary::m_released): New member.
35179 Do not release a summary twice. Do not allow to call finalizer
35180 for types of a summary that live in GGC memory.
35182 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35184 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35187 2017-02-02 Martin Sebor <msebor@redhat.com>
35189 PR middle-end/79275
35190 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35191 (format_string): Tighten up the range of output for non-constant
35192 strings and correct the expected range for wide non-constant strings.
35194 2017-02-02 Martin Sebor <msebor@redhat.com>
35196 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35198 PR middle-end/32003
35199 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35201 (-fdump-tree-@var): Add to index and document how to come up
35202 with pass-specific option and dump file names.
35203 (-fdump-passes): Clarify where to look for output.
35205 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35207 PR middle-end/77445
35208 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35209 statistics of the analyzed path; allow threading for speed when
35210 any of BBs along the path are optimized for speed.
35212 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35214 PR middle-end/78468
35215 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35216 settings of the virtual registers.
35219 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35221 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35222 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35225 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35227 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35228 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35230 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35232 * config/s390/s390.md: Add missing comments with the expanded
35234 * config/s390/vector.md: Likewise.
35235 * config/s390/vx-builtins.md: Likewise.
35237 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35240 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35241 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35242 conditions on a single line.
35244 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35246 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35247 __S390_VX__ to __VX__.
35249 2017-02-01 Andrew Pinski <apinski@cavium.com>
35251 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35252 stmt_info to record_stmt_cost.
35253 (vect_get_known_peeling_cost): Pass stmt_info if known to
35255 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35256 cpu_vector_cost field into
35257 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35258 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35259 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35260 splitting of scalar_stmt_cost and vec_stmt_cost.
35261 (thunderx_vector_cost): Likewise.
35262 (cortexa57_vector_cost): LIkewise.
35263 (exynosm1_vector_cost): Likewise.
35264 (xgene1_vector_cost): Likewise.
35265 (thunderx2t99_vector_cost): Improve after the splitting of the two
35267 (aarch64_builtin_vectorization_cost): Update for the splitting of
35268 scalar_stmt_cost and vec_stmt_cost.
35270 2017-02-01 Torvald Riegel <triegel@redhat.com>
35271 Richard Henderson <rth@redhat.com>
35273 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35274 conditional on existance of a fast atomic load.
35275 * optabs-query.c (can_atomic_load_p): New function.
35276 * optabs-query.h (can_atomic_load_p): Declare it.
35277 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35278 no fast atomic load is available for the particular size of access.
35279 (expand_atomic_compare_and_swap): Likewise.
35280 (expand_atomic_load): Likewise.
35281 (expand_atomic_store): Likewise.
35282 (expand_atomic_fetch_op): Likewise.
35283 * testsuite/lib/target-supports.exp
35284 (check_effective_target_sync_int_128): Remove x86 because it provides
35285 no fast atomic load.
35286 (check_effective_target_sync_int_128_runtime): Likewise.
35288 2017-02-01 Richard Biener <rguenther@suse.de>
35290 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35291 (graphite_transform_loops): Provide opt-info for optimized nests.
35292 * tree-parloop.c (parallelize_loops): Provide opt-info for
35293 parallelized loops.
35295 2017-02-01 Richard Biener <rguenther@suse.de>
35297 PR middle-end/79315
35298 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35299 was not set before.
35301 2017-02-01 Richard Biener <rguenther@suse.de>
35303 PR tree-optimization/71824
35304 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35305 Verify the loops are valid in the merged SESE region.
35306 (scop_detection::can_represent_loop_1): Check analyzing the
35307 evolution of the number of iterations in the region succeeds.
35309 2017-01-31 Ian Lance Taylor <iant@golang.org>
35311 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35312 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35314 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35316 PR preprocessor/79210
35317 * input.c (get_substring_ranges_for_loc): Replace line_width
35318 assertion with error-handling.
35320 2017-01-31 Richard Biener <rguenther@suse.de>
35322 PR tree-optimization/77318
35323 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35324 (create_pw_aff_from_tree): Take loop parameter.
35325 (add_condition_to_pbb): Pass loop of the condition to
35326 create_pw_aff_from_tree.
35328 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35330 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35331 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35333 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35337 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35339 (convert_int_to_float128): Likewise.
35340 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35341 (convert_int_to_float128): Likewise.
35342 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35343 (UNSPEC_IEEE128_CONVERT): Likewise.
35344 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35345 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35346 Use local variables for IBM extended format.
35347 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35348 (fix_trunc<mode>si2_fprs): Likewise.
35349 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35350 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35351 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35352 to know that we can now have integers of all sizes in vector
35354 (fix<uns>_<mode>di2_hw): Likewise.
35355 (float<uns>_<mode>si2_hw): Likewise.
35356 (fix_<mode>si2_hw): Likewise.
35357 (fixuns_<mode>si2_hw): Likewise.
35358 (float<uns>_<mode>di2_hw): Likewise.
35359 (float_<mode>di2_hw): Likewise.
35360 (float_<mode>si2_hw): Likewise.
35361 (floatuns_<mode>di2_hw): Likewise.
35362 (floatuns_<mode>si2_hw): Likewise.
35363 (xscvqp<su>wz_<mode>): Delete, no longer used.
35364 (xscvqp<su>dz_<mode>): Likewise.
35365 (xscv<su>dqp_<mode>): Likewise.
35366 (ieee128_mfvsrd_64bit): Likewise.
35367 (ieee128_mfvsrd_32bit): Likewise.
35368 (ieee128_mfvsrwz): Likewise.
35369 (ieee128_mtvsrw): Likewise.
35370 (ieee128_mtvsrd_64bit): Likewise.
35371 (ieee128_mtvsrd_32bit): Likewise.
35373 2017-01-31 Martin Liska <mliska@suse.cz>
35376 * ipa-prop.c (ipa_free_all_node_params): Call release method
35377 instead of ~sumbol_summary to not to trigger double times
35380 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35382 PR tree-optimization/71691
35383 * bitmap.h (class auto_bitmap): New.
35384 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35385 is_maybe_undefined instead of ssa_undefined_value_p.
35387 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35389 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35390 __S390_ARCH_LEVEL__ to __ARCH__.
35392 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35394 PR tree-optimization/79267
35395 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35396 if should_remove_lhs_p is true.
35398 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35401 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35402 (add_alignment_attribute): New.
35403 (base_type_die): Add alignment attribute.
35404 (subrange_type_die): Likewise.
35405 (modified_type_die): Likewise.
35406 (gen_array_type_die): Likewise.
35407 (gen_descr_array_type_die: Likewise.
35408 (gen_enumeration_type_die): Likewise.
35409 (gen_subprogram_die): Likewise.
35410 (gen_variable_die): Likewise.
35411 (gen_field_die): Likewise.
35412 (gen_ptr_to_mbr_type_die): Likewise.
35413 (gen_struct_or_union_type_die): Likewise.
35414 (gen_subroutine_type_die): Likewise.
35415 (gen_typedef_die): Likewise.
35416 (base_type_cmp): Compare alignment attribute.
35418 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35421 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35422 (setb_unsigned) New pattern for setb with CCUNS.
35423 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35424 subfc./subfe sequence to avoid overflow problems. Generate a
35425 shorter sequence with cmpld/setb for power9.
35426 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35427 for generating subfc. instruction.
35428 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35429 now uses this instruction.
35431 2017-01-30 Ian Lance Taylor <iant@google.com>
35434 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35435 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35437 2017-01-30 Martin Sebor <msebor@redhat.com>
35439 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35440 Move constant to the right of a relational operator.
35441 (get_mpfr_format_length, format_character, format_string): Ditto.
35442 (should_warn_p, maybe_warn): Same.
35444 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35446 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35449 * asan.c (get_translation_unit_decl): Remove function.
35450 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35452 2017-01-30 Martin Liska <mliska@suse.cz>
35454 PR gcov-profile/79259
35455 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35456 -fprofile-generate.
35458 2017-01-30 Martin Liska <mliska@suse.cz>
35461 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35462 Initialize variables with NULL value.
35464 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35467 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35469 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35471 2017-01-30 Richard Biener <rguenther@suse.de>
35473 PR tree-optimization/79276
35474 * tree-vrp.c (process_assert_insertions): Properly adjust common
35475 when removing a duplicate.
35477 2017-01-30 Richard Biener <rguenther@suse.de>
35479 PR tree-optimization/79256
35480 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35481 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35483 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35485 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35488 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35489 ("*r<noxa>sbg_<mode>_sll_bitmask")
35490 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35491 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35492 Use contiguous_bitmask_nowrap_operand.
35494 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35497 * config/rs6000/altivec.h (vec_xl): Revise #define.
35498 (vec_xst): Likewise.
35500 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35502 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35504 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35506 PR rtl-optimization/79194
35507 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35508 traps before call to bypass_conditional_jumps.
35510 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35512 PR tree-optimization/71374
35513 * lra-constraints.c (check_conflict_input_operands): New.
35514 (match_reload): Use it.
35516 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35519 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35520 account to calculate conflict_set.
35522 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35524 PR rtl-optimization/78559
35525 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35526 other_insn in combine.
35528 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35530 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35531 uint16_type_node for BT_UINT16.
35533 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35535 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35536 "RTL Tests" to menu.
35537 (GIMPLE Tests): New node.
35538 (RTL Tests): New node.
35540 2017-01-27 Richard Biener <rguenther@suse.de>
35542 PR tree-optimization/79245
35543 * tree-loop-distribution.c (distribute_loop): Apply cost
35544 modeling also to detected patterns.
35546 2017-01-27 Richard Biener <rguenther@suse.de>
35548 PR tree-optimization/71433
35549 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35550 (compare_assert_loc): New function.
35551 (process_assert_insertions): Sort and optimize assert locations
35552 to remove duplicates and push down identical assertions on
35553 edges to their destination block.
35555 2017-01-27 Richard Biener <rguenther@suse.de>
35557 PR tree-optimization/79244
35558 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35559 out SSA names even if abnormal.
35561 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35563 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35564 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35565 instead of MPFR_RNDN.
35567 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35570 * arm.c (arm_option_override): Don't call build_target_option_node
35571 until after doing all option overrides.
35572 (arm_valid_target_attribute_tree): Likewise.
35574 2017-01-27 Martin Liska <mliska@suse.cz>
35576 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35577 for {cd}tors and C++ {cd}tors.
35579 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35581 * config/s390/s390.md ("*setmem_long_and")
35582 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35584 2017-01-26 Martin Sebor <msebor@redhat.com>
35586 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35589 2017-01-26 Martin Sebor <msebor@redhat.com>
35591 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35592 HAVE_DFmode before using XFmode or DFmode.
35593 (parse_directive): Avoid using the z length modifier to avoid
35594 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
35596 PR middle-end/78703
35597 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
35598 to accept adjustment as an array.
35599 (get_int_range): New function.
35600 (struct directive): Make width and prec arrays.
35601 (directive::set_width, directive::set_precision): Call get_int_range.
35602 (format_integer, format_floating): Handle width and precision ranges.
35603 (format_string, parse_directive): Same.
35605 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35608 * dwarf2out.c (generate_skeleton_bottom_up): For children with
35609 comdat_type_p set, just clone them, but keep the children in the
35613 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
35614 which have direct callers with -fvar-tracking-assignments enabled
35616 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
35617 inside of type units.
35619 2017-01-26 Martin Sebor <msebor@redhat.com>
35621 PR middle-end/78703
35622 * gimple-ssa-sprintf.c (struct result_range): Add likely and
35624 (struct format_result): Replace number_chars, number_chars_min,
35625 and number_chars_max with a single member of struct result_range.
35627 (format_result::operator+=): Adjust.
35628 (struct fmtresult): Remove bounded. Handle likely and unlikely
35630 (fmtresult::adjust_for_width_or_precision): New function.
35631 (fmtresult:type_max_digits): New function.
35632 (bytes_remaining): Handle likely and unlikely counters.
35633 (min_bytes_remaining): Remove.
35634 (format_percent): Simplify.
35635 (format_integer, format_floating): Set likely and unlikely counters.
35636 (get_string_length, format_character, format_string): Same.
35637 (format_plain, should_warn_p): New function.
35638 (maybe_warn): Call should_warn_p. Update diagnostic messages
35639 and handle those for all directives, including plain strings.
35640 (format_directive): Handle likely and unlikely counters.
35641 Remove unnecessary quoting from diagnostics. Add an informational
35643 (add_bytes): Remove.
35644 (pass_sprintf_length::compute_format_length): Simplify.
35645 (try_substitute_return_value): Handle likely and unlikely counters.
35647 2017-01-26 Carl Love <cel@us.ibm.com>
35649 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
35650 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
35652 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
35655 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
35656 endianess for subregs into account.
35657 * lra-constraints.c (lra_constraints): Do risky transformations
35658 always on the first iteration.
35659 * lra-lives.c (check_pseudos_live_through_calls): Add arg
35660 last_call_used_reg_set.
35661 (process_bb_lives): Define and use last_call_used_reg_set.
35662 * lra.c (lra): Always continue after lra_constraints on the first
35665 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
35667 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
35669 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
35671 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35673 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
35674 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
35675 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
35676 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
35677 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
35678 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
35679 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
35680 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
35681 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
35683 2017-01-26 Marek Polacek <polacek@redhat.com>
35686 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
35687 for the third operand.
35689 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35691 PR middle-end/79236
35692 * omp-low.c (struct omp_context): Add simt_stmt field.
35693 (scan_omp_for): Return omp_context *.
35694 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
35695 context to the _simt_ SIMD stmt.
35696 (lower_omp_for): For combined SIMD with sibling _simt_
35697 SIMD, make sure to use the same decls in _looptemp_
35698 clauses as in the sibling.
35700 2017-01-26 David Sherwood <david.sherwood@arm.com>
35702 PR middle-end/79212
35703 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
35706 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35709 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
35710 emit fld b; fld a; if possible.
35712 * brig-builtins.def: Update copyright years.
35713 * config/arm/arm_acle_builtins.def: Update copyright years.
35715 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
35718 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
35719 constraint instead of o for the stxsd instruction.
35721 2017-01-25 Carl Love <cel@us.ibm.com>
35723 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
35724 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
35726 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
35728 * doc/invoke.texi (C++ Dialect Options): Fix typo.
35730 2017-01-25 Richard Biener <rguenther@suse.de>
35732 PR tree-optimization/69264
35733 * target.def (vector_alignment_reachable): Improve documentation.
35734 * doc/tm.texi: Regenerate.
35735 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
35737 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
35738 earlier changes with respect to TYPE_USER_ALIGN.
35739 (vector_alignment_reachable_p): Likewise. Improve dumping.
35741 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35744 * config/arm/arm.md (xordi3): Force constant operand into a register
35747 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35749 * doc/invoke.texi (-fstore-merging): Correct default optimization
35750 levels at which it is enabled.
35751 (-O): Move -fstore-merging from list to...
35754 2017-01-25 Richard Biener <rguenther@suse.de>
35757 * omp-expand.c: Include debug.h.
35758 (expand_omp_taskreg): Make sure to generate early debug before
35759 outlining anything from a function.
35760 (expand_omp_target): Likewise.
35761 (grid_expand_target_grid_body): Likewise.
35763 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
35766 * asan.c (get_translation_unit_decl): New function.
35767 (asan_add_global): Extract modules file name from globals
35768 TRANSLATION_UNIT_DECL name.
35770 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
35773 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
35774 for long calls with APCS frame and VFP.
35776 2017-01-24 David Malcolm <dmalcolm@redhat.com>
35778 * cfg.c (original_copy_tables_initialized_p): New function.
35779 * cfg.h (original_copy_tables_initialized_p): New decl.
35780 * cfgrtl.c (relink_block_chain): Guard the call to
35781 free_original_copy_tables with a call to
35782 original_copy_tables_initialized_p.
35783 * cgraph.h (symtab_node::native_rtl_p): New decl.
35784 * cgraphunit.c (symtab_node::native_rtl_p): New function.
35785 (symtab_node::needed_p): Don't assert for early assembly output
35786 for __RTL functions.
35787 (cgraph_node::finalize_function): Set "force_output" for __RTL
35789 (cgraph_node::analyze): Bail out early for __RTL functions.
35790 (analyze_functions): Update assertion to support __RTL functions.
35791 (cgraph_node::expand): Bail out early for __RTL functions.
35792 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
35794 * function.h (struct function): Update comment for field
35796 * gimple-expr.c: Include "tree-pass.h".
35797 (gimple_has_body_p): Return false for __RTL functions.
35798 * Makefile.in (OBJS): Add run-rtl-passes.o.
35799 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
35801 (gcc::pass_manager::get_clean_slate): New accessor.
35802 * passes.c: Include "insn-addr.h".
35803 (should_skip_pass_p): Add logging. Update logic for running
35804 "expand" to be compatible with both __GIMPLE and __RTL. Guard
35805 property-provider override so it is only done for gimple passes.
35807 (skip_pass): New function.
35808 (execute_one_pass): Call skip_pass when skipping passes.
35809 * read-md.c (md_reader::read_char): Support filtering
35810 the input to a subset of line numbers.
35811 (md_reader::md_reader): Initialize fields
35812 m_first_line and m_last_line.
35813 (md_reader::read_file_fragment): New function.
35814 * read-md.h (md_reader::read_file_fragment): New decl.
35815 (md_reader::m_first_line): New field.
35816 (md_reader::m_last_line): New field.
35817 * read-rtl-function.c (function_reader::create_function): Only
35818 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
35819 curr_properties. Set DECL_INITIAL to a dummy block.
35820 (read_rtl_function_body_from_file_range): New function.
35821 * read-rtl-function.h (read_rtl_function_body_from_file_range):
35823 * run-rtl-passes.c: New file.
35824 * run-rtl-passes.h: New file.
35826 2017-01-24 Jeff Law <law@redhat.com>
35828 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
35831 2017-01-24 Bin Cheng <bin.cheng@arm.com>
35833 PR tree-optimization/79159
35834 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
35835 (record_nonwrapping_iv): Improve boundary using above function if no
35836 value range information.
35838 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
35839 Martin Jambor <mjambor@suse.cz>
35841 * brig-builtins.def: New file.
35842 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
35843 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
35844 (DEF_HSAIL_SAT_BUILTIN): Likewise.
35845 (DEF_HSAIL_INTR_BUILTIN): Likewise.
35846 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
35847 * builtin-types.def (BT_INT8): New.
35848 (BT_INT16): Likewise.
35849 (BT_UINT8): Likewise.
35850 (BT_UINT16): Likewise.
35851 (BT_FN_ULONG): Likewise.
35852 (BT_FN_UINT_INT): Likewise.
35853 (BT_FN_UINT_ULONG): Likewise.
35854 (BT_FN_UINT_LONG): Likewise.
35855 (BT_FN_UINT_PTR): Likewise.
35856 (BT_FN_ULONG_PTR): Likewise.
35857 (BT_FN_INT8_FLOAT): Likewise.
35858 (BT_FN_INT16_FLOAT): Likewise.
35859 (BT_FN_UINT32_FLOAT): Likewise.
35860 (BT_FN_UINT16_FLOAT): Likewise.
35861 (BT_FN_UINT8_FLOAT): Likewise.
35862 (BT_FN_UINT64_FLOAT): Likewise.
35863 (BT_FN_UINT16_UINT32): Likewise.
35864 (BT_FN_UINT32_UINT16): Likewise.
35865 (BT_FN_UINT16_UINT16_UINT16): Likewise.
35866 (BT_FN_INT_PTR_INT): Likewise.
35867 (BT_FN_UINT_PTR_UINT): Likewise.
35868 (BT_FN_LONG_PTR_LONG): Likewise.
35869 (BT_FN_ULONG_PTR_ULONG): Likewise.
35870 (BT_FN_VOID_UINT64_UINT64): Likewise.
35871 (BT_FN_UINT8_UINT8_UINT8): Likewise.
35872 (BT_FN_INT8_INT8_INT8): Likewise.
35873 (BT_FN_INT16_INT16_INT16): Likewise.
35874 (BT_FN_INT_INT_INT): Likewise.
35875 (BT_FN_UINT_FLOAT_UINT): Likewise.
35876 (BT_FN_FLOAT_UINT_UINT): Likewise.
35877 (BT_FN_ULONG_UINT_UINT): Likewise.
35878 (BT_FN_ULONG_UINT_PTR): Likewise.
35879 (BT_FN_ULONG_ULONG_ULONG): Likewise.
35880 (BT_FN_UINT_UINT_UINT): Likewise.
35881 (BT_FN_VOID_UINT_PTR): Likewise.
35882 (BT_FN_UINT_UINT_PTR: Likewise.
35883 (BT_FN_UINT32_UINT64_PTR): Likewise.
35884 (BT_FN_INT_INT_UINT_UINT): Likewise.
35885 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
35886 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
35887 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
35888 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
35889 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
35890 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
35891 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
35892 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
35893 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
35894 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
35895 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
35896 * doc/frontends.texi: List BRIG FE.
35897 * doc/install.texi (Testing): Add BRIG tesring requirements.
35898 * doc/invoke.texi (Overall Options): Mention BRIG.
35899 * doc/standards.texi (Standards): Doucment BRIG HSA version.
35901 2017-01-24 Richard Biener <rguenther@suse.de>
35903 PR translation/79208
35904 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
35906 2017-01-24 Martin Jambor <mjambor@suse.cz>
35909 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
35910 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
35911 and known_contexts.
35913 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
35915 PR middle-end/79123
35916 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
35917 casts from signed to unsigned really don't have a range.
35919 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
35921 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
35922 GMP_RNDx for compatiblity.
35924 2017-01-24 Martin Liska <mliska@suse.cz>
35927 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
35928 that would prevent us to call alloca with -1 as argument.
35930 2017-01-24 Jakub Jelinek <jakub@redhat.com>
35932 * dwarf2out.c (output_compilation_unit_header, output_file_names):
35933 Avoid -Wformat-security warning.
35935 2017-01-23 Andrew Pinski <apinski@cavium.com>
35937 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
35940 2017-01-23 Martin Sebor <msebor@redhat.com>
35942 PR middle-end/78703
35943 * gimple-ssa-sprintf.c (warn_level): New global.
35944 (format_integer): Use it here and throughout the rest of the file.
35945 Use the same switch to compute sign as base.
35946 (maybe_warn): New function.
35947 (format_directive): Factor out warnings into maybe_warn.
35948 Add debugging output. Use warn_level.
35949 (add_bytes): Use warn_level.
35950 (pass_sprintf_length::compute_format_length): Add debugging output.
35951 (try_substitute_return_value): Same.
35952 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
35954 PR middle-end/78703
35955 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
35956 (struct fmtresult, format_integer, format_floating): Adjust.
35957 (fmtresult::fmtresult): Set max correctly in two argument ctor.
35958 (get_string_length, format_string,format_directive): Same.
35959 (pass_sprintf_length::compute_format_length): Same.
35960 (try_substitute_return_value): Simplify slightly.
35962 PR middle-end/78703
35963 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
35964 (fmtresult::operator+=): Outlined.
35965 (struct fmtresult): Add ctors.
35966 (struct conversion_spec): Rename...
35967 (struct directive): ...to this. Add and remove data members.
35968 (directive::set_width, directive::set_precision): New functions.
35969 (format_percent): Use fmtresult ctor.
35970 (get_width_and_precision): Remove.
35971 (format_integer): Make naming changes. Avoid computing width and
35973 (format_floating): Same. Adjust indentation.
35974 (format_character, format_none): New functions.
35975 (format_string): Moved character handling to format_character.
35976 (format_directive): Remove arguments, change return type.
35977 (parse_directive): New function.
35978 (pass_sprintf_length::compute_format_length): Move directive
35979 parsing to parse_directive.
35981 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35983 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
35984 (assign_assembler_name_if_needed): ... this.
35985 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
35986 (assign_assembler_name_if_needed): ... this.
35987 (free_lang_data_in_cgraph): Adjust callers.
35988 * cgraphunit.c (cgraph_node::analyze): Likewise.
35989 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
35992 2017-01-23 Richard Biener <rguenther@suse.de>
35994 PR tree-optimization/79088
35995 PR tree-optimization/79188
35996 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
35997 resetting loop bounds after last path deletion. Reset loop
35998 bounds of the target loop, make code match the comments.
35999 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36000 Make sure loops need no fixups.
36002 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36004 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36005 exponent support with double type for first argument.
36006 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36007 type returned by __builtin_vec_extract_sig,
36008 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36009 functions from "vector int" to "vector unsigned int" or from
36010 "vector long long int" to "vector unsigned long long int".
36011 Changed type returned by __builtin_vec_extract_exp,
36012 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36013 functions from "vector int" to "vector unsigned int" or from
36014 "vector long long int" to "vector unsigned long long int".
36015 Changed return type of __builtin_vec_test_data_class,
36016 __builtin_vec_test_data_class_sp, and
36017 __builtin_vec_test_data_class_dp from "vector int" to
36018 "vector bool int" or from "vector long long int" to "vector bool
36019 long long int" and changed second argument type from "unsigned
36020 int" to "int". Added new overloaded function forms "vector float
36021 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36022 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36023 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36024 double, vector unsigned long long int)" and "vector double
36025 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36026 long int)". Changed return type of
36027 __builtin_scalar_test_data_class and
36028 __builtin_scalar_test_data_class_sp and
36029 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36030 int" and changed second argument from "unsigned int" to "int".
36031 Changed type returned by __builtin_scalar_test_neg,
36032 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36033 from "int" to "bool int". Added new overloaded function form
36034 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36035 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36036 exponent double-precision with floating point first argument.
36037 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36038 documentation of scalar_test_data_class, scalar_test_neg,
36039 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36040 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36041 vec_test_data_class built-in functions to reflect refinements in
36042 their type signatures.
36044 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36046 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36048 (aarch64_elf_asm_destructor): Likewise.
36050 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36052 PR rtl-optimization/78634
36053 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36054 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36055 * ifcvt.c (noce_try_cmove): Add missing cost check.
36057 PR rtl-optimization/71724
36058 * combine.c (if_then_else_cond): Look for situations where it is
36059 beneficial to undo the work of one of the recursive calls.
36061 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36063 PR tree-optimization/70754
36064 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36065 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36066 combined stmt before it if not NULL.
36067 (combine_chains): Process refs reversely and compute dominance point
36070 2017-01-23 Martin Liska <mliska@suse.cz>
36072 PR tree-optimization/79196
36073 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36074 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36076 (strlen_optimize_stmt): Call the renamed function.
36078 2017-01-23 Michael Matz <matz@suse.de>
36080 PR tree-optimization/78384
36081 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36083 2017-01-23 Richard Biener <rguenther@suse.de>
36085 PR tree-optimization/79186
36086 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36087 both incoming edges before moving an assert.
36089 2017-01-23 Martin Jambor <mjambor@suse.cz>
36091 * ipa-prop.c (load_from_param_1): Removed.
36092 (load_from_unmodified_param): Bits from load_from_param_1 put back
36094 (load_from_param): Removed.
36095 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36096 with stmt. Reverted back to use of load_from_unmodified_param.
36098 2017-01-23 Martin Jambor <mjambor@suse.cz>
36101 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36102 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36103 field a pointer to garbage collected vector, mark lattices and
36104 ipcp_orig_node with GTY((skip)).
36105 (ipa_get_param_count): Adjust to descriptors being a pointer.
36106 (ipa_get_param): Likewise.
36107 (ipa_get_type): Likewise.
36108 (ipa_get_param_move_cost): Likewise.
36109 (ipa_set_param_used): Likewise.
36110 (ipa_get_controlled_uses): Likewise.
36111 (ipa_set_controlled_uses): Likewise.
36112 (ipa_is_param_used): Likewise.
36113 (ipa_node_params_t): Move into garbage collector. New methods insert
36115 (ipa_node_params_sum): Annotate wth GTY(()).
36116 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36118 (ipa_load_from_parm_agg): Adjust declaration.
36119 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36120 * ipa-profile.c (ipa_profile): Likewise.
36121 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36122 (ipa_populate_param_decls): Make descriptors parameter garbage
36124 (ipa_dump_param): Adjust to descriptors being a pointer.
36125 (ipa_alloc_node_params): Likewise.
36126 (ipa_initialize_node_params): Likewise.
36127 (load_from_param_1): Make descriptors parameter garbage collected.
36128 (load_from_unmodified_param): Likewise.
36129 (load_from_param): Likewise.
36130 (ipa_load_from_parm_agg): Likewise.
36131 (ipa_node_params::~ipa_node_params): Removed.
36132 (ipa_free_all_node_params): Remove call to delete operator.
36133 (ipa_node_params_t::insert): New.
36134 (ipa_node_params_t::remove): Likewise.
36135 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36136 copy known_csts and known_contexts vectors.
36137 (ipa_read_node_info): Adjust to descriptors being a pointer.
36138 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36140 (ipcp_transform_function): Make descriptors variable garbage
36143 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36145 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36146 * config/i386/avx512dqintrin.h: Ditto.
36147 * config/i386/avx512fintrin.h: Ditto.
36148 * config/i386/i386.c: Handle new builtins.
36149 * config/i386/i386-builtin.def: Add new builtins.
36150 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36151 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36153 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36154 Martin Liska <mliska@suse.cz>
36156 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36157 * asan.c (asan_expand_poison_ifn): Support stores and use
36158 appropriate ASAN report function.
36159 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36160 * internal-fn.def (ASAN_POISON_USE): Declare.
36161 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36162 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36163 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36164 ASAN_POISON calls w/o LHS.
36165 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36166 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36167 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36168 * gimplify.c (asan_poison_variables): Add attribute
36169 use_after_scope_memory to variables that really needs to live
36171 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36172 having the attribute.
36174 2017-01-23 Martin Liska <mliska@suse.cz>
36176 * asan.c (create_asan_shadow_var): New function.
36177 (asan_expand_poison_ifn): Likewise.
36178 * asan.h (asan_expand_poison_ifn): New declaration.
36179 * internal-fn.c (expand_ASAN_POISON): Likewise.
36180 * internal-fn.def (ASAN_POISON): New builtin.
36181 * sanopt.c (pass_sanopt::execute): Expand
36182 asan_expand_poison_ifn.
36183 * tree-inline.c (copy_decl_for_dup_finish): Make function
36185 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36186 * tree-ssa.c (is_asan_mark_p): New function.
36187 (execute_update_addresses_taken): Rewrite local variables
36188 (identified just by use-after-scope as addressable) into SSA.
36190 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36192 * doc/install.texi (Specific): opensource.apple.com uses https
36193 now. Remove trailing slash.
36195 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36197 * README.Portability: Remove note on an Irix compatibility issue.
36199 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36201 * gcov.c (INCLUDE_ALGORITHM): Define.
36202 (INCLUDE_VECTOR): Define.
36203 No longer include <vector> and <algorithm> directly.
36205 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36207 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36209 * doc/invoke.texi (Code Gen Options): Ditto.
36211 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36214 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36216 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36218 rtl-optimization/79125
36219 * cprop.c (local_cprop_pass): Handle cases where we make an
36220 unconditional trap.
36222 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36226 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36227 read from, for big endian.
36229 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36231 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36232 register pauth builtins for LP64 only.
36234 2017-01-20 Marek Polacek <polacek@redhat.com>
36237 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36240 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36242 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36244 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36245 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36246 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36248 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36251 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36252 in big-endian mode when they are not a single duplicated value.
36254 2017-01-20 Richard Biener <rguenther@suse.de>
36256 * BASE-VER: Bump to 7.0.1.
36258 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36260 * omp-low.c (omplow_simd_context): New struct. Use it...
36261 (lower_rec_simd_input_clauses): ...here and...
36262 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36263 references to idx, lane, max_vf, is_simt.
36265 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36267 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36270 2017-01-20 Martin Jambor <mjambor@suse.cz>
36272 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36273 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36275 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36276 (GTFILES): Rename hsa.c to hsa-common.c.
36277 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36278 * hsa-dump.c: Likewise.
36279 * hsa-gen.c: Likewise.
36280 * hsa-regalloc.c: Likewise.
36281 * ipa-hsa.c: Likewise.
36282 * omp-expand.c: Likewise.
36283 * omp-low.c: Likewise.
36284 * toplev.c: Likewise.
36286 2017-01-20 Marek Polacek <polacek@redhat.com>
36289 * doc/invoke.texi: Document -Wduplicated-branches.
36290 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36291 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36292 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36293 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36294 return 0 only when not OEP_LEXICOGRAPHIC.
36295 (fold_build_cleanup_point_expr): Use the expression
36296 location when building CLEANUP_POINT_EXPR.
36297 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36298 * tree.c (add_expr): Handle error_mark_node.
36300 2017-01-20 Martin Liska <mliska@suse.cz>
36303 * tree-profile.c (init_ic_make_global_vars): Do not call
36305 (gimple_init_gcov_profiler): Likewise.
36307 2017-01-20 Martin Liska <mliska@suse.cz>
36310 * cgraph.h (maybe_create_reference): Remove argument and
36312 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36314 * ipa-cp.c (create_specialized_node): Likewise.
36315 * symtab.c (symtab_node::maybe_create_reference): Handle
36316 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36318 2017-01-20 Martin Liska <mliska@suse.cz>
36320 * read-rtl-function.c (function_reader::create_function): Use
36321 build_decl instread of build_decl_stat.
36323 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36325 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36326 * config/i386/avx512dqintrin.h: Ditto.
36327 * config/i386/avx512fintrin.h: Ditto.
36328 * config/i386/i386-builtin-types.def: Add new types.
36329 * config/i386/i386.c: Handle new types.
36330 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36331 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36332 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36333 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36334 (__builtin_ia32_kshiftridi): New.
36335 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36337 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36341 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36342 define to rs6000_init_stack_protect_guard.
36343 (rs6000_init_stack_protect_guard): New function.
36345 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36346 Yunqiang Su <yunqiang.su@imgtec.com>
36348 * config.gcc (supported_defaults): Add madd4.
36349 (with_madd4): Add validation.
36350 (all_defaults): Add madd4.
36351 * config/mips/mips.opt (mmadd4): New option.
36352 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36354 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36356 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36357 (ISA_HAS_FUSED_MADD4): Likewise.
36358 * doc/invoke.texi (-mmadd4): Document the new option.
36359 * doc/install.texi (--with-madd4): Document the new option.
36361 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36363 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36364 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36365 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36366 (aarch64_init_pauth_hint_builtins): New.
36367 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36368 (aarch64_expand_builtin): Expand new builtins.
36370 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36372 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36373 * combine-stack-adj.c (no_unhandled_cfa): Handle
36374 REG_CFA_TOGGLE_RA_MANGLE.
36375 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36376 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36377 info for return address signing.
36378 (aarch64_expand_epilogue): Likewise.
36380 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36382 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36383 * config/aarch64/aarch64-protos.h
36384 (aarch64_return_address_signing_enabled): New declaration.
36385 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36387 (aarch64_expand_prologue): Sign return address before it's pushed onto
36389 (aarch64_expand_epilogue): Authenticate return address fetched from
36391 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36392 (aarch64_attributes): New function attributes for "sign-return-address".
36393 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36394 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36395 ("*do_return"): Generate combined instructions according to key index.
36396 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36397 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36399 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36400 * config/aarch64/aarch64.opt (msign-return-address=): New.
36401 * doc/extend.texi (AArch64 Function Attributes): Documents
36402 "sign-return-address=".
36403 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36405 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36407 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36408 overall option summary.
36410 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36412 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36413 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36414 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36415 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36417 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36419 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36420 -mpower9-minmax by default for -mcpu=power9.
36421 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36422 128-bit floating point.
36424 2017-01-20 Alan Modra <amodra@gmail.com>
36426 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36427 optimizing for size.
36429 2017-01-20 Alan Modra <amodra@gmail.com>
36432 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36433 for strcmp and strncmp from corresponding builtin decl.
36435 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36437 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36438 instead of i386/rtems-64.h.
36439 * config/i386/rtems-64.h: Remove.
36441 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36445 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36447 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36449 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36451 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36452 Change int to HOST_WIDE_INT.
36453 * config/aarch64/aarch64-protos.h
36454 (aarch64_simd_gen_const_vector_dup): Likewise.
36455 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36457 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36459 * langhooks-def.h (lhd_type_for_size): New decl.
36460 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36461 * langhooks.c (lhd_type_for_size): New function, taken from
36464 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36466 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36467 define_bypass for CR latency.
36468 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36469 (power9-alu2): Add define_bypass for CR latency.
36471 (power9-mul): Update insn latency.
36472 (power9-mul-compare): Update insn latency, bypass latency and remove
36475 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36477 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36479 * config/aarch64/aarch64.md
36480 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36481 aarch64_nopcrelative_literal_loads.
36482 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36484 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36486 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36487 TARGET_LOONGSON_3A.
36488 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36490 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36493 * config.gcc (supported_defaults): Add lxc1-sxc1.
36494 (with_lxc1_sxc1): Add validation.
36495 (all_defaults): Add lxc1-sxc1.
36496 * config/mips/mips.opt (mlxc1-sxc1): New option.
36497 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36499 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36500 __mips_no_lxc1_sxc1.
36501 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36502 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36503 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36505 2017-01-19 Richard Biener <rguenther@suse.de>
36507 PR tree-optimization/72488
36508 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36509 sure to restore SSA info.
36510 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36512 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36514 PR rtl-optimization/79121
36515 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36516 of the inner type when shifting an extended value.
36518 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36521 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36522 interposable aliases.
36524 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36527 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36528 Use the evmergelohi instruction.
36529 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36530 (mov_sitf_e500_subreg8_2_be): Likewise.
36531 (mov_sitf_e500_subreg12_2_le): Likewise.
36532 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36533 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36534 (mov_sitf_e500_subreg8_2_le): Likewise.
36535 (mov_sitf_e500_subreg12_2_be): Likewise.
36537 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36539 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36540 attribute from vecsimple to vecperm.
36541 (altivec_vbpermq2): Likewise.
36543 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36546 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36548 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36549 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36550 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36551 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36552 case where N arg is SIZE_MAX.
36553 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36554 (cmpstrsi): Add pattern.
36556 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36558 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36559 __builtin_vec_revb builtins.
36560 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36561 built-in functions to support generation of the ISA 3.0 XXBR<x>
36562 vector byte reverse instructions.
36563 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36564 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36565 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36566 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36567 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36568 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36569 (P9V_BUILTIN_VEC_REVB): Likewise.
36570 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36571 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36572 (p9_xxbrq_v16qi): Likewise.
36573 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36574 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36575 (p9_xxbrh_v8hi): Likewise.
36576 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36577 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36578 vec_revb built-in functions.
36580 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36582 PR rtl-optimization/78952
36583 * config/i386/i386.md (any_extract): New code iterator.
36584 (*insvqi_2): Use any_extract for source operand.
36585 (*insvqi_3): Use any_shiftrt for source operand.
36587 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36589 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36591 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36593 2017-01-18 Matthias Klose <doko@ubuntu.com>
36595 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
36597 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36599 * config/rs6000/altivec.h (vec_bperm): Change #define.
36600 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
36601 (altivec_vbpermq2): New define_insn.
36602 (altivec_vbpermd): Likewise.
36603 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
36604 function interface.
36605 (VBPERMD): Likewise.
36606 (VBPERM): New polymorphic function interface.
36607 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
36608 Add entries for P9V_BUILTIN_VEC_VBPERM.
36609 * doc/extend.texi: Add interfaces for vec_bperm.
36611 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36613 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
36614 first letter of error messages.
36615 (s390_resolve_overloaded_builtin): Likewise.
36616 * config/s390/s390.c (s390_expand_builtin): Likewise.
36617 (s390_invalid_arg_for_unprototyped_fn): Likewise.
36618 (s390_valid_target_attribute_inner_p): Likewise.
36619 * config/s390/s390.md ("tabort"): Likewise.
36621 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
36623 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
36624 (ISA_AVOID_DIV_HILO): New macro.
36625 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
36626 (ISA_HAS_DDIV): Likewise.
36628 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36630 * doc/invoke.texi (fabi-version): Correct number of occurrences.
36632 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36634 * doc/invoke.texi (fabi-version): Spelling fix.
36636 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36639 * doc/invoke.texi (fabi-version): Mention mangling fix for
36642 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36645 * doc/invoke.texi (fabi-version): Document discriminator mangling.
36647 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
36650 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
36651 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
36653 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
36655 (stack_protect_test): Ditto.
36656 * config/rs6000/rs6000.opt (mstack-protector-guard=,
36657 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
36659 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
36660 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
36661 -mstack-protector-guard-offset=.
36662 (RS/6000 and PowerPC Options): Ditto.
36664 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36666 * config/i386/i386.h (MASK_CLASS_P): New define.
36667 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
36668 there are no registers from different register sets also when
36669 mask registers are used. Update function comment.
36670 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
36671 to (*k/*r) and (*k/*km) alternatives.
36673 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
36675 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
36676 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
36677 (EH_RETURN_HANDLER_RTX): New define.
36678 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
36679 Force frame pointer in EH return functions.
36680 (aarch64_expand_epilogue): Add barrier for eh_return.
36681 (aarch64_final_eh_return_addr): Remove.
36682 (aarch64_eh_return_handler_rtx): New function.
36683 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
36685 (aarch64_eh_return_handler_rtx): New prototype.
36687 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36689 * config/rs6000/altivec.h (vec_rlmi): New #define.
36690 (vec_vrlnm): Likewise.
36691 (vec_rlnm): Likewise.
36692 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
36693 (UNSPEC_VRLNM): Likewise.
36694 (VIlong): New mode iterator.
36695 (altivec_vrl<VI_char>mi): New define_insn.
36696 (altivec_vrl<VI_char>nm): Likewise.
36697 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
36699 (VRLDNM): Likewise.
36700 (RLNM): New polymorphic function entry.
36701 (VRLWMI): New monomorphic function entry.
36702 (VRLDMI): Likewise.
36703 (RLMI): New polymorphic function entry.
36704 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
36705 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
36706 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
36709 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36712 * dwarf2out.c (field_byte_offset): Restore the
36713 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
36714 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
36715 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
36718 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36721 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
36723 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36726 * dwarf2out.c (add_data_member_location_attribute): For constant
36727 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
36728 instead of DW_AT_data_member_location, DW_AT_bit_offset and
36729 DW_AT_byte_size attributes.
36731 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36733 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
36734 after forcing to constant memory when the code model is medium.
36736 2017-01-17 Julia Koval <julia.koval@intel.com>
36739 * config/i386/avx512fintrin.h
36740 (_mm512_i32gather_ps): Change __addr type to void const*.
36741 (_mm512_mask_i32gather_ps): Ditto.
36742 (_mm512_i32gather_pd): Ditto.
36743 (_mm512_mask_i32gather_pd): Ditto.
36744 (_mm512_i64gather_ps): Ditto.
36745 (_mm512_mask_i64gather_ps): Ditto.
36746 (_mm512_i64gather_pd): Ditto.
36747 (_mm512_mask_i64gather_pd): Ditto.
36748 (_mm512_i32gather_epi32): Ditto.
36749 (_mm512_mask_i32gather_epi32): Ditto.
36750 (_mm512_i32gather_epi64): Ditto.
36751 (_mm512_mask_i32gather_epi64): Ditto.
36752 (_mm512_i64gather_epi32): Ditto.
36753 (_mm512_mask_i64gather_epi32): Ditto.
36754 (_mm512_i64gather_epi64): Ditto.
36755 (_mm512_mask_i64gather_epi64): Ditto.
36756 (_mm512_i32scatter_ps): Change __addr type to void*.
36757 (_mm512_mask_i32scatter_ps): Ditto.
36758 (_mm512_i32scatter_pd): Ditto.
36759 (_mm512_mask_i32scatter_pd): Ditto.
36760 (_mm512_i64scatter_ps): Ditto.
36761 (_mm512_mask_i64scatter_ps): Ditto.
36762 (_mm512_i64scatter_pd): Ditto.
36763 (_mm512_mask_i64scatter_pd): Ditto.
36764 (_mm512_i32scatter_epi32): Ditto.
36765 (_mm512_mask_i32scatter_epi32): Ditto.
36766 (_mm512_i32scatter_epi64): Ditto.
36767 (_mm512_mask_i32scatter_epi64): Ditto.
36768 (_mm512_i64scatter_epi32): Ditto.
36769 (_mm512_mask_i64scatter_epi32): Ditto.
36770 (_mm512_i64scatter_epi64): Ditto.
36771 (_mm512_mask_i64scatter_epi64): Ditto.
36772 * config/i386/avx512pfintrin.h
36773 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
36774 (_mm512_mask_prefetch_i32gather_ps): Ditto.
36775 (_mm512_mask_prefetch_i64gather_pd): Ditto.
36776 (_mm512_mask_prefetch_i64gather_ps): Ditto.
36777 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
36778 (_mm512_prefetch_i32scatter_ps): Ditto.
36779 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
36780 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
36781 (_mm512_prefetch_i64scatter_pd): Ditto.
36782 (_mm512_prefetch_i64scatter_ps): Ditto.
36783 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
36784 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
36785 * config/i386/avx512vlintrin.h
36786 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
36787 (_mm_mmask_i32gather_ps): Ditto.
36788 (_mm256_mmask_i32gather_pd): Ditto.
36789 (_mm_mmask_i32gather_pd): Ditto.
36790 (_mm256_mmask_i64gather_ps): Ditto.
36791 (_mm_mmask_i64gather_ps): Ditto.
36792 (_mm256_mmask_i64gather_pd): Ditto.
36793 (_mm_mmask_i64gather_pd): Ditto.
36794 (_mm256_mmask_i32gather_epi32): Ditto.
36795 (_mm_mmask_i32gather_epi32): Ditto.
36796 (_mm256_mmask_i32gather_epi64): Ditto.
36797 (_mm_mmask_i32gather_epi64): Ditto.
36798 (_mm256_mmask_i64gather_epi32): Ditto.
36799 (_mm_mmask_i64gather_epi32): Ditto.
36800 (_mm256_mmask_i64gather_epi64): Ditto.
36801 (_mm_mmask_i64gather_epi64): Ditto.
36802 (_mm256_i32scatter_ps): Change __addr type to void*.
36803 (_mm256_mask_i32scatter_ps): Ditto.
36804 (_mm_i32scatter_ps): Ditto.
36805 (_mm_mask_i32scatter_ps): Ditto.
36806 (_mm256_i32scatter_pd): Ditto.
36807 (_mm256_mask_i32scatter_pd): Ditto.
36808 (_mm_i32scatter_pd): Ditto.
36809 (_mm_mask_i32scatter_pd): Ditto.
36810 (_mm256_i64scatter_ps): Ditto.
36811 (_mm256_mask_i64scatter_ps): Ditto.
36812 (_mm_i64scatter_ps): Ditto.
36813 (_mm_mask_i64scatter_ps): Ditto.
36814 (_mm256_i64scatter_pd): Ditto.
36815 (_mm256_mask_i64scatter_pd): Ditto.
36816 (_mm_i64scatter_pd): Ditto.
36817 (_mm_mask_i64scatter_pd): Ditto.
36818 (_mm256_i32scatter_epi32): Ditto.
36819 (_mm256_mask_i32scatter_epi32): Ditto.
36820 (_mm_i32scatter_epi32): Ditto.
36821 (_mm_mask_i32scatter_epi32): Ditto.
36822 (_mm256_i32scatter_epi64): Ditto.
36823 (_mm256_mask_i32scatter_epi64): Ditto.
36824 (_mm_i32scatter_epi64): Ditto.
36825 (_mm_mask_i32scatter_epi64): Ditto.
36826 (_mm256_i64scatter_epi32): Ditto.
36827 (_mm256_mask_i64scatter_epi32): Ditto.
36828 (_mm_i64scatter_epi32): Ditto.
36829 (_mm_mask_i64scatter_epi32): Ditto.
36830 (_mm256_i64scatter_epi64): Ditto.
36831 (_mm256_mask_i64scatter_epi64): Ditto.
36832 (_mm_i64scatter_epi64): Ditto.
36833 (_mm_mask_i64scatter_epi64): Ditto.
36834 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
36835 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
36836 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
36837 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
36838 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
36839 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
36840 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
36841 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
36842 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
36843 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
36844 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
36845 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
36846 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
36847 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
36848 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
36849 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
36850 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
36851 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
36852 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
36853 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
36854 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
36855 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
36856 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
36857 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
36858 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
36859 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
36860 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
36861 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
36862 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
36863 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
36864 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
36865 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
36866 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
36867 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
36868 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
36869 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
36870 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
36871 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
36872 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
36873 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
36874 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
36875 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
36876 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
36877 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
36878 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
36879 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
36880 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
36881 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
36882 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
36883 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
36884 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
36885 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
36886 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
36887 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
36888 definitions accordingly.
36890 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
36891 Kuan-Lin Chen <kuanlinchentw@gmail.com>
36894 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
36897 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
36900 * ira-conflicts.c (ira_build_conflicts): Update total conflict
36901 hard regs for inner regno.
36903 2017-01-17 Martin Liska <mliska@suse.cz>
36906 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
36907 assumption and add comment.
36909 2017-01-17 Nathan Sidwell <nathan@acm.org>
36911 * ipa-visibility.c (localize_node): New function, broken out of ...
36912 (function_and_variable_visibility): ... here. Call it.
36914 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36916 PR middle-end/77445
36917 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
36918 correctly set frequency of oudgoing edge.
36919 (duplicate_thread_path): Fix profile updating.
36921 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36924 * configure.ac: Add GCC_BASE_VER.
36925 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
36926 version from BASE-VER file.
36927 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
36928 (gcc.o): Depend on $(BASEVER).
36929 * common.opt (dumpfullversion): New option.
36930 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
36931 * doc/invoke.texi: Document -dumpfullversion.
36932 * doc/install.texi: Document --with-gcc-major-version-only.
36933 * configure: Regenerated.
36935 2017-01-17 Richard Biener <rguenther@suse.de>
36937 PR tree-optimization/71433
36938 * tree-vrp.c (register_new_assert_for): Merge same asserts
36939 on all incoming edges.
36940 (process_assert_insertions_for): Handle insertions at the
36943 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
36945 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
36946 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
36948 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
36951 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
36954 2017-01-17 Alan Modra <amodra@gmail.com>
36957 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
36958 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
36959 symbolic stack limit when pic.
36961 2017-01-16 Martin Sebor <msebor@redhat.com>
36963 PR tree-optimization/78608
36964 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
36966 2017-01-16 Jeff Law <law@redhat.com>
36969 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
36970 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
36971 for several include directories that may be relative to sysroot.
36972 * config/i386/x-mingw32 (gplus_includedir): Define.
36973 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
36974 (native_system_includedir): Likewise.
36975 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
36976 override if TARGET_SYSTEM_ROOT is defined.
36977 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
36979 PR tree-optimization/79090
36980 PR tree-optimization/33562
36981 PR tree-optimization/61912
36982 PR tree-optimization/77485
36983 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
36984 and computed trims into the dump file.
36986 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36988 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
36990 2017-01-16 Jakub Jelinek <jakub@redhat.com>
36993 * gimplify.c (gimplify_init_constructor): If want_value and
36994 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
36998 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
36999 sequence. Formatting fixes.
37000 (doloop_optimize): Formatting fixes.
37003 * gcc.c (debug_level_greater_than_spec_func): New function.
37004 (static_spec_functions): Add debug-level-gt spec function.
37005 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37007 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37008 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37009 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37010 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37011 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37012 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37014 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37016 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37017 QImode fixups to general and mask registers only.
37019 2017-01-16 Carl Love <cel@us.ibm.com>
37021 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37022 for built-in functions
37023 vector signed char vec_nabs (vector signed char)
37024 vector signed short vec_nabs (vector signed short)
37025 vector signed int vec_nabs (vector signed int)
37026 vector signed long long vec_nabs (vector signed long long)
37027 vector float vec_nabs (vector float)
37028 vector double vec_nabs (vector double)
37029 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37031 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37032 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37033 * doc/extend.texi: Update the documentation file for the new built-in
37036 2017-01-16 Martin Sebor <msebor@redhat.com>
37038 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37041 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37043 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37044 UNSPEC_VSX__XXSPLTD to require special splat handling.
37046 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37049 * system.h: Poison strndup.
37051 2017-01-16 Alan Modra <amodra@gmail.com>
37054 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37057 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37059 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37061 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37063 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37064 call recog here. Assert that INSN_CODE (insn) is non-negative.
37066 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37069 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37071 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37072 in the currently scheduled RTL fragment.
37074 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37076 PR rtl-optimization/78751
37077 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37080 2017-01-14 Jeff Law <law@redhat.com>
37082 PR tree-optimization/79090
37083 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37084 variable length stores.
37085 (compute_trims): Delete dead assignment to *trim_tail.
37086 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37089 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37091 PR rtl-optimization/78626
37092 PR rtl-optimization/78727
37093 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37094 of a block, and split such blocks after everything else is finished.
37096 2017-01-14 Alan Modra <amodra@gmail.com>
37099 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37100 target legitimate_combined_insn.
37101 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37102 (rs6000_legitimate_combined_insn): New function.
37103 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37105 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37106 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37107 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37109 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37111 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37113 2017-01-13 Jeff Law <law@redhat.com>
37115 PR tree-optimization/33562
37116 PR tree-optimization/61912
37117 PR tree-optimization/77485
37118 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37120 (delete_dead_assignment): Likewise.
37121 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37122 statement to delete_dead_call and delete_dead_assignment.
37124 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37127 * substring-locations.c (format_warning_va): Strengthen case 1 so
37128 that both endpoints of the substring must be within the format
37129 range for just the substring to be printed.
37131 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37133 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37134 * config/i386/i386.c (ix86_target_string): Add missing options
37135 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37136 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37137 flags_other and ix86_target_other to flags2_other. Display unknown
37139 (ix86_valid_target_attribute_inner_p): Add missing options and
37140 reorder options by implied ISAs, as in ix86_target_string.
37142 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37144 * hash-table.h (hash_table::too_empty_p): New function.
37145 (hash_table::expand): Use it.
37146 (hash_table::traverse): Likewise.
37147 (hash_table::empty_slot): Use sizeof (value_type) instead of
37148 sizeof (PTR) to convert bytes to elements. Shrink the table
37149 if the current size is excessive for the current number of
37152 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37154 * ira-costs.c (record_reg_classes): Break from the inner loop
37155 early once alt_fail is known to be true. Update outer loop
37156 handling accordingly.
37158 2017-01-13 Jeff Law <law@redhat.com>
37160 * tree-ssa-dse.c (decrement_count): New function.
37161 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37162 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37163 when we know the partially dead statement is a mem* function.
37165 PR tree-optimization/61912
37166 PR tree-optimization/77485
37167 * tree-ssa-dse.c: Include expr.h.
37168 (maybe_trim_constructor_store): New function.
37169 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37171 PR tree-optimization/33562
37172 PR tree-optimization/61912
37173 PR tree-optimization/77485
37174 * doc/invoke.texi: Document new dse-max-object-size param.
37175 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37176 * tree-ssa-dse.c: Include params.h.
37177 (dse_store_status): New enum.
37178 (initialize_ao_ref_for_dse): New, partially extracted from
37180 (valid_ao_ref_for_dse, normalize_ref): New.
37181 (setup_live_bytes_from_ref, compute_trims): Likewise.
37182 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37183 (maybe_trim_partially_dead_store): Likewise.
37184 (maybe_trim_complex_store): Likewise.
37185 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37186 Track what bytes live from the original store. Return tri-state
37187 for dead, partially dead or live.
37188 (dse_dom_walker): Add constructor, destructor and new private members.
37189 (delete_dead_call, delete_dead_assignment): New extracted from
37191 (dse_optimize_stmt): Make a member of dse_dom_walker.
37192 Use initialize_ao_ref_for_dse.
37194 PR tree-optimization/33562
37195 PR tree-optimization/61912
37196 PR tree-optimization/77485
37197 * sbitmap.h (bitmap_count_bits): Prototype.
37198 (bitmap_clear_range, bitmap_set_range): Likewise.
37199 * sbitmap.c (bitmap_clear_range): New function.
37200 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37202 2017-01-13 Martin Liska <mliska@suse.cz>
37205 * function.c (set_cfun): Add new argument force.
37206 * function.h (set_cfun): Likewise.
37207 * ipa-inline-transform.c (inline_call): Use the function when
37208 strict alising from is dropped for function we inline to.
37210 2017-01-13 Richard Biener <rguenther@suse.de>
37212 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37213 for dumping GIMPLE INTEGER_CSTs.
37215 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37217 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37218 to 201112L since C++17.
37220 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37223 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37224 if -fsanitize=kernel-address is present.
37226 2017-01-13 Richard Biener <rguenther@suse.de>
37228 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37229 as _Literal ( type ) number in case usual suffixes do not
37230 preserve all information.
37232 2017-01-13 Richard Biener <rguenther@suse.de>
37234 PR tree-optimization/77283
37235 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37236 and ssa-iterators.h.
37237 (is_feasible_trace): Implement a cost model based on joiner
37240 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37243 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37244 char or short to __float128/_Float128 directly.
37246 2017-01-12 Martin Sebor <msebor@redhat.com>
37248 to -Wformat-overflow.
37249 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37250 (min_bytes_remaining): Same.
37251 (get_string_length): Same.
37252 (format_string): Same.
37253 (format_directive): Same.
37255 (pass_sprintf_length::handle_gimple_call): Same.
37257 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37259 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37260 info.nowrite calls with no lhs that can't throw. Return bool
37261 whether gsi_remove has been called or not.
37262 (pass_sprintf_length::handle_gimple_call): Return bool whether
37263 try_substitute_return_value called gsi_remove. Formatting fix.
37264 (pass_sprintf_length::execute): Don't use gsi_remove if
37265 handle_gimple_call returned true.
37268 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37269 be removed due to side-effects, don't remove following barrier nor
37270 turn the successor edge into fallthru edge.
37272 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37275 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37276 element-reversing loads and stores as not swappable.
37278 2017-01-12 Nathan Sidwell <nathan@acm.org>
37279 Nicolai Stange <nicstange@gmail.com>
37281 * combine.c (try_combine): Don't ignore result of overlap checking
37282 loop. Combine overlap & asm check into single loop.
37284 2017-01-12 Richard Biener <rguenther@suse.de>
37286 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37287 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37289 2017-01-12 Richard Biener <rguenther@suse.de>
37291 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37292 and TS_TARGET_OPTION directly derive from TS_BASE.
37293 * tree-core.h (tree_optimization_option): Derive from tree_base.
37294 (tree_target_option): Likewise.
37296 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37298 * config/i386/i386.c (memory_address_length): Increase len
37299 only when rip_relative_addr_p returns false.
37301 2017-01-11 Julia Koval <julia.koval@intel.com>
37303 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37304 (OPTION_MASK_ISA_SGX_SET): New.
37305 (ix86_handle_option): Handle OPT_msgx.
37306 * config.gcc: Added sgxintrin.h.
37307 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37308 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37309 * config/i386/i386.c (ix86_target_string): Add -msgx.
37311 (ix86_option_override_internal): Handle new options.
37312 (ix86_valid_target_attribute_inner_p): Add sgx.
37313 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37314 * config/i386/i386.opt: Add msgx.
37315 * config/i386/sgxintrin.h: New file.
37316 * config/i386/x86intrin.h: Add sgxintrin.h.
37318 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37321 * fold-const.c (maybe_nonzero_address): Return 1 for function
37323 (tree_single_nonzero_warnv_p): Don't handle function local objects
37327 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37330 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37333 * opts.c: Include "spellcheck.h"
37334 (struct string_fragment): New struct.
37335 (struct edit_distance_traits<const string_fragment &>): New
37337 (get_closest_sanitizer_option): New function.
37338 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37340 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37342 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37344 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37345 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37346 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37347 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37348 for initial die_offset if dwarf_split_debug_info.
37349 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37350 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37352 (output_skeleton_debug_sections): Formatting fix. Use
37353 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37354 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37356 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37358 * config/arm/cortex-a53.md: Add bypasses for
37359 cortex_a53_r2f_cvt.
37360 (cortex_a53_r2f): Only use for transfers.
37361 (cortex_a53_f2r): Likewise.
37362 (cortex_a53_r2f_cvt): Add reservation for conversions.
37363 (cortex_a53_f2r_cvt): Likewise.
37365 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37367 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37368 to all inlined functions, change static to extern.
37370 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37373 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37375 (arm_assemble_integer): Likewise.
37377 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37379 * config.gcc: Use new awk script to check CPU, FPU and architecture
37380 parameters for --with-... options.
37381 * config/arm/parsecpu.awk: New file
37382 * config/arm/arm-cpus.in: New file.
37383 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37385 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37387 * config/arm/t-arm: Update dependency rules.
37388 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37389 of processing .def files.
37390 * config/arm/genopt.sh: Deleted.
37391 * config/arm/gentune.sh: Deleted.
37392 * config/arm/arm-cores.def: Deleted.
37393 * config/arm/arm-arches.def: Deleted.
37394 * config/arm/arm-fpus.def: Deleted.
37395 * config/arm/arm-tune.md: Regenerated.
37396 * config/arm/arm-tables.opt: Regenerated.
37397 * config/arm/arm-cpu.h: New generated file.
37398 * config/arm/arm-cpu-data.h: New generated file.
37399 * config/arm/arm-cpu-cdata.h: New generated file.
37401 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37404 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37406 (input_varpool_node): Unpack dynamically_initialized bit.
37408 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37410 PR rtl-optimization/79032
37411 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37412 the alignment of the adjusted memory reference against that of MODE,
37413 instead of the alignment of the original memory reference.
37415 2017-01-11 Martin Jambor <mjambor@suse.cz>
37417 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37419 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37420 decorated functions.
37422 2017-01-11 Richard Biener <rguenther@suse.de>
37424 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37425 set range/nonnull info for PHI results. Do not set it on
37426 stmts marked for removal.
37428 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37430 * expr.c (store_field): In the bitfield case, fetch the return value
37431 from the registers before applying a single big-endian adjustment.
37432 Always do a final load for a BLKmode value not larger than a word.
37434 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37437 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37438 that we correctly handle column numbers greater than
37439 LINE_MAP_MAX_COLUMN_NUMBER.
37441 2017-01-10 Martin Sebor <msebor@redhat.com>
37443 PR middle-end/78245
37444 * gimple-ssa-sprintf.c (get_destination_size): Call
37445 {init,fini}object_sizes.
37446 * tree-object-size.c (addr_object_size): Adjust.
37447 (pass_through_call): Adjust.
37448 (pass_object_sizes::execute): Adjust.
37449 * tree-object-size.h (fini_object_sizes): Declare.
37451 2017-01-10 Martin Sebor <msebor@redhat.com>
37453 PR tree-optimization/78775
37454 * builtins.c (get_size_range): Move...
37455 * calls.c: ...to here.
37456 (alloc_max_size): Accept zero argument.
37457 (operand_signed_p): Remove.
37458 (maybe_warn_alloc_args_overflow): Call get_size_range.
37459 * calls.h (get_size_range): Declare.
37461 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37463 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37464 from TI's devices.csv file as of September 2016.
37465 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37467 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37469 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37470 * doc/invoke.texi: Likewise.
37471 * doc/md.texi: Likewise.
37472 * doc/objc.texi: Likewise.
37474 2017-01-10 Joshua Conner <joshconner@google.com>
37476 * config/arm/fuchsia-elf.h: New file.
37477 * config/fuchsia.h: New file.
37478 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37479 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37481 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37483 2016-01-10 Richard Biener <rguenther@suse.de>
37485 PR tree-optimization/79034
37486 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37487 Propagate out degenerate PHIs in the joiner.
37489 2017-01-10 Martin Liska <mliska@suse.cz>
37491 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37492 (sort_congruence_classes_by_decl_uid): Likewise.
37493 (sort_congruence_class_groups_by_decl_uid): Likewise.
37494 (sem_item_optimizer::merge_classes): Sort class, groups in these
37495 classes and members in the groups by DECL_UID of declarations.
37496 This would make merge operations stable.
37498 2017-01-10 Martin Liska <mliska@suse.cz>
37500 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37501 usage of m_classes_vec.
37502 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37503 (sem_item_optimizer::get_group_by_hash): Likewise.
37504 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37505 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37506 (sem_item_optimizer::verify_classes): Likewise.
37507 (sem_item_optimizer::process_cong_reduction): Likewise.
37508 (sem_item_optimizer::dump_cong_classes): Likewise.
37509 (sem_item_optimizer::merge_classes): Likewise.
37510 * ipa-icf.h (congruence_class_hash): Rename from
37511 congruence_class_group_hash. Remove declaration of m_classes_vec.
37513 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37515 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37516 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37517 * config.gcc: Add avx512vpopcntdqintrin.h.
37518 * config/i386/avx512vpopcntdqintrin.h: New.
37519 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37520 * config/i386/i386-builtin-types.def: Add new types.
37521 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37522 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37523 __builtin_ia32_vpopcountq_v8di_mask): New.
37524 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37525 __AVX512VPOPCNTDQ__.
37526 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37527 (PTA_AVX512VPOPCNTDQ): Define.
37528 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37529 TARGET_AVX512VPOPCNTDQ_P): Define.
37530 * config/i386/i386.opt: Add mavx512vpopcntdq.
37531 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37532 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37534 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37536 PR middle-end/77484
37537 * predict.def (PRED_CALL): Set to 67.
37539 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37541 * expr.c (store_field): In the bitfield case, if the value comes from
37542 a function call and is of an aggregate type returned in registers, do
37543 not modify the field mode; extract the value in all cases if the mode
37544 is BLKmode and the size is not larger than a word.
37546 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37549 * config/i386/cpuid.h: Fix undefined behavior.
37551 2017-01-04 Jeff Law <law@redhat.com>
37553 PR tree-optimization/79007
37554 PR tree-optimization/67955
37555 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37556 conservative for pt.null when flag_non_call_exceptions is on.
37558 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37560 PR translation/79019
37561 PR translation/79020
37562 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37563 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37564 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37566 * config/avr/avr.opt (maccumulate-args): Likewise.
37567 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37568 * common.opt (freport-bug): Likewise.
37569 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37570 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37571 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37572 translatable string.
37573 * config/i386/i386.c (function_value_32): Likewise.
37574 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37575 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37577 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37578 * common/config/msp430/msp430-common.c (msp430_handle_option):
37580 * symtab.c (symtab_node::verify_base): Likewise.
37581 * opts.c (set_debug_level): Likewise.
37582 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37583 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37584 missing whitespace to translatable strings.
37585 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37586 * config/sh/superh.h: Likewise.
37587 * config/i386/xopintrin.h: Likewise.
37588 * config/i386/znver1.md: Likewise.
37589 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37590 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37591 * double-int.h (struct double_int): Likewise.
37592 * double-int.c (div_and_round_double): Likewise.
37593 * wide-int.cc: Likewise.
37594 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
37595 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
37596 * cfgcleanup.c (crossjumps_occured): Renamed to ...
37597 (crossjumps_occurred): ... this.
37598 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
37601 PR tree-optimization/78899
37602 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
37603 returning bool return struct loop *, NULL for failure and the new
37605 (versionable_outer_loop_p): Don't version outer loop if it has
37606 dont_vectorized bit set.
37607 (tree_if_conversion): When versioning outer loop, ensure
37608 tree_if_conversion is performed also on the inner loop of the
37609 non-vectorizable outer loop copy.
37610 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
37611 LOOP_VECTORIZED in inner loop of the scalar outer loop and
37612 prevent vectorization of it.
37613 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
37614 the outer loop vectorization of the non-scalar version is attempted
37615 before vectorization of the inner loop in scalar version. If
37616 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
37617 vectorization of its inner loop.
37618 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
37619 has 2 inner loops, rename also on edges from bb whose single pred
37620 is outer_loop->header. Fix typo in function comment.
37622 2017-01-09 Martin Sebor <msebor@redhat.com>
37625 * asan.c (asan_emit_stack_protection): Increase local buffer size
37626 to avoid snprintf truncation warning.
37628 2017-01-09 Andrew Pinski <apinski@cavium.com>
37630 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
37631 to reference thunderx2t99 for the tuning structure
37632 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
37634 (thunderx2t99_extra_costs): This.
37635 * config/aarch64/aarch64-tune.md: Regenerate.
37636 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
37637 (vulcan_addrcost_table): This.
37638 (vulcan_regmove_cost): Rename to ...
37639 (thunderx2t99_regmove_cost): This.
37640 (vulcan_vector_cost): Rename to ...
37641 (thunderx2t99_vector_cost): this.
37642 (vulcan_branch_cost): Rename to ...
37643 (thunderx2t99_branch_cost): This.
37644 (vulcan_tunings): Rename to ...
37645 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
37646 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
37648 2017-01-09 Martin Jambor <mjambor@suse.cz>
37652 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
37653 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
37654 (propagate_vr_accross_jump_function): Use the above function for all
37655 value range computations for pass-through jump functions and type
37656 converasion from explicit value range values.
37657 (ipcp_propagate_stage): Do not attempt to deduce types of formal
37658 parameters from TYPE_ARG_TYPES.
37659 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
37660 (ipa_write_node_info): Stream type of the actual argument.
37661 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
37663 2017-01-09 Martin Liska <mliska@suse.cz>
37666 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
37667 (lookup_compiler): Do not show error message with have_E.
37669 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37671 PR tree-optimization/78938
37672 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
37673 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
37674 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
37675 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
37678 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37680 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
37683 2017-01-09 Richard Biener <rguenther@suse.de>
37685 PR tree-optimization/78997
37686 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
37687 name condition properly.
37689 2017-01-09 Richard Biener <rguenther@suse.de>
37692 * dwarf2out.c (is_cxx): New overload with context.
37693 (is_naming_typedef_decl): Use it.
37695 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37697 * invoke.texi (Option Summary): Correct spacing in option lists
37698 and add line breaks to fix over-long lines.
37700 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37702 PR middle-end/17660
37704 * extend.texi (Common Variable Attributes): Add xref to GCC
37705 Internals manual to explain mode attribute keywords.
37707 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37710 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
37711 and Preprocessor Options.
37712 (Options for Linking): Document -pthread here....
37713 (RS/6000 and PowerPC Options): ...not here.
37714 (Solaris 2 Options): ...or here.
37715 * doc/cppopts.texi: Document -pthread.
37717 2017-01-08 Martin Sebor <msebor@redhat.com>
37719 PR middle-end/77708
37720 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
37721 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
37722 New member functions.
37723 (format_directive): Used them.
37725 (pass_sprintf_length::handle_gimple_call): Same.
37726 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
37727 to avoid truncation for any argument.
37728 (extract_affine_mul): Same.
37729 * tree.c (get_file_function_name): Same.
37731 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37733 PR middle-end/77484
37734 * predict.def (PRED_INDIR_CALL): Set to 86.
37736 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37738 PR preprocessor/54124
37739 * doc/cppopts.texi: Reformat -d subtable to list the full name
37740 of the options. Add cross-reference to the docs for the general
37741 compiler -d options.
37742 * doc/invoke.texi (Developer Options): Add cross-reference to the
37743 preprocessor-specific -d option documentation.
37745 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37747 PR preprocessor/13498
37748 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
37749 redudant material, and reflect new command-line options.
37750 (System Headers): Likewise.
37752 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37754 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
37755 -isystem, and -idirafter. Copy-edit.
37756 * doc/cppopts.texi: Copy-edit. Remove contradiction about
37757 default for -ftrack-macro-expansion. Delete obsolete and
37758 badly-formatted implementation details about -fdebug-cpp output.
37759 * doc/cppwarnopts.texi: Copy-edit.
37761 2017-01-07 David Malcolm <dmalcolm@redhat.com>
37764 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37765 that the transition from a max line width >= 1<<10 to narrower
37766 lines works correctly.
37768 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
37770 * doc/options.texi (PerFunction): New.
37771 * opt-functions.awk (switch_flags): Map both Optimization and
37772 PerFunction to CL_OPTIMIZATION.
37773 * opth-gen.awk: Test for PerFunction flag along with
37775 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
37776 it only when the latter is present. Skip those that don't in
37777 the hash function generator.
37778 * common.opt (fvar-tracking): Mark as PerFunction instead of
37780 (fvar-tracking-assignments): Likewise.
37781 (fvar-tracking-assignments-toggle): Likewise.
37782 (fvar-tracking-uninit): Likewise.
37784 2017-01-07 Jakub Jelinek <jakub@redhat.com>
37786 PR translation/79018
37787 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
37790 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
37793 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
37794 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
37795 TARGET_LONG_JUMP_TABLE_OFFSETS.
37796 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
37797 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
37798 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
37799 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
37800 * config/m68k/m68k.md (tablejump expander): Likewise.
37801 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
37802 TARGET_LONG_JUMP_TABLE_OFFSETS.
37803 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
37804 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
37806 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
37807 David Holsgrove <david.holsgrove@xilinx.com>
37809 * common/config/microblaze/microblaze-common.c
37810 (TARGET_EXCEPT_UNWIND_INFO): Remove.
37811 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
37813 * config/microblaze/microblaze.c (microblaze_must_save_register)
37814 (microblaze_expand_epilogue, microblaze_return_addr): Handle
37816 (microblaze_eh_return): New function.
37817 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
37818 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
37819 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
37820 * config/microblaze/microblaze.md (eh_return): New pattern.
37822 2017-01-06 Jakub Jelinek <jakub@redhat.com>
37824 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
37825 GCC_DIAGNOSTIC_STRINGIFY): Define.
37827 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
37829 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37831 * config/arm/arm.md (<mcrr>): New.
37833 * config/arm/arm.c (arm_arch5te): New.
37834 (arm_option_override): Set arm_arch5te.
37835 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
37837 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
37838 (arm_mcrr_qualifiers): ... this. New.
37839 (MRRC_QUALIFIERS): Define to...
37840 (arm_mrrc_qualifiers): ... this. New.
37841 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
37843 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
37844 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
37845 (MRRCI, mrrc, MRRC): New.
37846 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
37847 VUNSPEC_MRRC2): New.
37849 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37851 * config/arm/arm.md (<mcr>): New.
37853 * config/arm/arm.c (arm_coproc_builtin_available): Add
37854 support for mcr, mrc, mcr2 and mrc2.
37855 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
37856 (arm_mcr_qualifiers): ... this. New.
37857 (MRC_QUALIFIERS): Define to ...
37858 (arm_mrc_qualifiers): ... this. New.
37859 (MCR_QUALIFIERS): Define to ...
37860 (arm_mcr_qualifiers): ... this. New.
37861 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
37863 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
37864 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
37865 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
37866 VUNSPEC_MRC2): New.
37868 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37870 * config/arm/arm.md (*ldc): New.
37874 * config/arm/arm.c (arm_coproc_builtin_available): Add
37875 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
37876 (arm_coproc_ldc_stc_legitimate_address): New.
37877 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
37878 'qualifier_const_pointer'.
37879 (LDC_QUALIFIERS): Define to...
37880 (arm_ldc_qualifiers): ... this. New.
37881 (STC_QUALIFIERS): Define to...
37882 (arm_stc_qualifiers): ... this. New.
37883 * config/arm/arm-protos.h
37884 (arm_coproc_ldc_stc_legitimate_address): New.
37885 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
37886 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
37887 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
37888 stc2, stcl, stc2l): New.
37889 * config/arm/constraints.md (Uz): New.
37890 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
37891 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
37892 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
37893 VUNSPEC_STC2L): New.
37895 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37897 * config/arm/arm.md (<cdp>): New.
37898 * config/arm/arm.c (neon_const_bounds): Rename this ...
37899 (arm_const_bounds): ... this.
37900 (arm_coproc_builtin_available): New.
37901 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
37902 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
37903 (CDP_QUALIFIERS): Define to...
37904 (arm_cdp_qualifiers): ... this. New.
37906 (arm_expand_builtin_args): Add case for 6 arguments.
37907 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
37908 (arm_const_bounds): ... this.
37909 (arm_coproc_builtin_available): New.
37910 * config/arm/arm_acle.h (__arm_cdp): New.
37912 * config/arm/arm_acle_builtins.def (cdp): New.
37914 * config/arm/iterators.md (CDPI,CDP,cdp): New.
37915 * config/arm/neon.md: Rename all 'neon_const_bounds' to
37916 'arm_const_bounds'.
37917 * config/arm/types.md (coproc): New.
37918 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
37919 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
37920 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
37921 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
37923 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37925 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
37926 (UBINOP_QUALIFIERS): New.
37928 (acle_builtin_data): New. Change comment.
37929 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
37930 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
37931 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
37932 arm_acle_builtins.def.
37933 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
37934 (arm_init_acle_builtins): New.
37935 (CRC32_BUILTIN): Remove.
37936 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
37937 crc32cb, crc32ch and crc32cw.
37938 (arm_init_crc32_builtins): Remove.
37939 (arm_init_builtins): Use arm_init_acle_builtins rather
37940 than arm_init_crc32_builtins.
37941 (arm_expand_acle_builtin): New.
37942 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
37943 * config/arm/arm_acle_builtins.def: New.
37945 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37947 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
37948 (arm_builtin_datum): ... this.
37949 (arm_init_neon_builtin): Rename to ...
37950 (arm_init_builtin): ... this. Add a new parameters PREFIX
37951 and USE_SIG_IN_NAME.
37952 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
37953 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
37954 'arm_builtin_datum'.
37955 (arm_init_vfp_builtins): Likewise.
37956 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
37957 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
37958 (arm_expand_neon_args): Rename to ...
37959 (arm_expand_builtin_args): ... this. Rename builtin_arg
37960 enum values and differentiate between ARG_BUILTIN_MEMORY
37961 and ARG_BUILTIN_NEON_MEMORY.
37962 (arm_expand_neon_builtin_1): Rename to ...
37963 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
37964 values, arm_expand_builtin_args and add bool parameter NEON.
37965 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
37966 (arm_expand_vfp_builtin): Likewise.
37967 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
37969 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37971 PR middle-end/77484
37972 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
37973 * predict.c (tree_estimate_probability_bb): Reverse direction of
37974 polymorphic call predictor.
37976 2017-01-06 David Malcolm <dmalcolm@redhat.com>
37978 * passes.c (execute_one_pass): Split out pass-skipping logic into...
37979 (determine_pass_name_match): ...this new function and...
37980 (should_skip_pass_p): ...this new function.
37982 2017-01-06 Nathan Sidwell <nathan@acm.org>
37984 * ipa-visibility.c (function_and_variable_visibility): Reformat
37985 comments and long lines. Remove extrneous if.
37986 * symtab.c (symtab_node::make_decl_local): Fix code format.
37987 (symtab_node::set_section_for_node): Fix comment typo.
37989 2017-01-06 Martin Liska <mliska@suse.cz>
37992 * lra-constraints.c: Rename invariant to lra_invariant.
37993 * predict.c (set_even_probabilities): Initialize e to NULL.
37995 2017-01-05 Martin Sebor <msebor@redhat.com>
37997 PR tree-optimization/78910
37998 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
37999 (format_integer): Correct off-by-one error in the handling
38000 of precision with negative numbers in signed conversions..
38002 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38004 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38006 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38008 PR tree-optimization/71016
38009 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38010 factor_out_conditional_conversion. Formatting fix.
38011 (factor_out_conditional_conversion): Add cond_stmt argument.
38012 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38013 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38016 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38018 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38019 read-rtl-function.o, and selftest-rtl.o.
38020 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38021 (selftest::aarch64_test_loading_full_dump): New function.
38022 (selftest::aarch64_run_selftests): New function.
38023 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38024 selftest::aarch64_run_selftests.
38025 * config/i386/i386.c
38026 (selftest::ix86_test_loading_dump_fragment_1): New function.
38027 (selftest::ix86_test_loading_call_insn): New function.
38028 (selftest::ix86_test_loading_full_dump): New function.
38029 (selftest::ix86_test_loading_unspec): New function.
38030 (selftest::ix86_run_selftests): Call the new functions.
38031 * emit-rtl.c (maybe_set_max_label_num): New function.
38032 * emit-rtl.h (maybe_set_max_label_num): New decl.
38033 * function.c (instantiate_decls): Guard call to
38034 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38035 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38037 * gensupport.c (gen_reader::gen_reader): Pass "false"
38038 for new "compact" param of rtx_reader.
38039 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38040 rather than an empty string for NULL strings.
38041 * read-md.c: Potentially include config.h rather than bconfig.h.
38042 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38043 (have_error): New global, copied from errors.c.
38044 (md_reader::read_name): Rename to...
38045 (md_reader::read_name_1): ...this, adding "out_loc" param,
38046 and converting "missing name or number" to returning false, rather
38048 (md_reader::read_name): Reimplement in terms of read_name_1.
38049 (md_reader::read_name_or_nil): New function.
38050 (md_reader::read_string): Handle "(nil)" by returning NULL.
38051 (md_reader::md_reader): Add new param "compact".
38052 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38053 (md_reader::read_file): New method.
38054 * read-md.h (md_reader::md_reader): Add new param "compact".
38055 (md_reader::read_file): New method.
38056 (md_reader::is_compact): New accessor.
38057 (md_reader::read_name): Convert return type from void to file_location.
38058 (md_reader::read_name_or_nil): New decl.
38059 (md_reader::read_name_1): New decl.
38060 (md_reader::m_compact): New field.
38061 (noop_reader::noop_reader): Pass "false" for new "compact" param
38063 (rtx_reader::rtx_reader): Add new "compact" param.
38064 (rtx_reader::read_rtx_operand): Make virtual and convert return
38065 type from void to rtx.
38066 (rtx_reader::read_until): New decl.
38067 (rtx_reader::handle_any_trailing_information): New virtual function.
38068 (rtx_reader::postprocess): New virtual function.
38069 (rtx_reader::finalize_string): New virtual function.
38070 (rtx_reader::m_in_call_function_usage): New field.
38071 (rtx_reader::m_reuse_rtx_by_id): New field.
38072 * read-rtl-function.c: New file.
38073 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38074 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38075 (selftest::verify_three_block_rtl_cfg): New decl.
38076 * read-rtl-function.h: New file.
38077 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38078 For host, include function.h, memmodel.h, and emit-rtl.h.
38079 (one_time_initialization): New function.
38080 (struct compact_insn_name): New struct.
38081 (compact_insn_names): New array.
38082 (find_code): Handle insn codes in compact dumps.
38083 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38084 (bind_subst_iter_and_attr): Likewise.
38085 (add_condition_to_string): Likewise.
38086 (add_condition_to_rtx): Likewise.
38087 (apply_attribute_uses): Likewise.
38088 (add_current_iterators): Likewise.
38089 (apply_iterators): Likewise.
38090 (initialize_iterators): Guard usage of apply_subst_iterator with
38091 #ifdef GENERATOR_FILE.
38092 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38093 (md_reader::read_mapping): Likewise.
38094 (add_define_attr_for_define_subst): Likewise.
38095 (add_define_subst_attr): Likewise.
38096 (read_subst_mapping): Likewise.
38097 (check_code_iterator): Likewise.
38098 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38100 (one_time_initialization): New function.
38101 (rtx_reader::read_until): New method.
38102 (read_flags): New function.
38103 (parse_reg_note_name): New function.
38104 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38105 Handle reuse_rtx ids.
38106 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38107 Add parsing support for RTL dumps, mirroring the special-cases in
38108 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38109 values, and calling handle_any_trailing_information.
38110 (rtx_reader::read_rtx_operand): Convert return type from void
38111 to rtx, returning return_rtx. Handle case 'e'. Call
38112 finalize_string on XSTR and XTMPL fields.
38113 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38114 "(nil)" values were omitted. Call the postprocess vfunc on the
38116 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38117 class ctor. Initialize m_in_call_function_usage. Call
38118 one_time_initialization.
38119 * rtl-tests.c (selftest::test_uncond_jump): Call
38120 set_new_first_and_last_insn.
38121 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38122 * selftest-rtl.c: New file.
38123 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38124 (selftest::get_insn_by_uid): New decl.
38125 * selftest-run-tests.c (selftest::run_tests): Call
38126 read_rtl_function_c_tests.
38127 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38128 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38131 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38133 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38134 operands in a special way. Assert that pos+len <= mode precision.
38136 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38138 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38139 3 argument Alias with unlimited for the negative form.
38140 (fno-vect-cost-model): Removed.
38142 2017-01-05 Martin Liska <mliska@suse.cz>
38144 * hsa-gen.c (gen_hsa_divmod): New function.
38145 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38147 2017-01-05 Martin Liska <mliska@suse.cz>
38150 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38153 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38155 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38156 small constant length operands.
38158 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38160 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38161 between loop iterations.
38163 2017-01-05 Martin Liska <mliska@suse.cz>
38166 * gimplify.c (gimplify_decl_expr): Compare to
38167 asan_poisoned_variables instread of checking flags.
38168 (gimplify_target_expr): Likewise.
38169 (gimplify_expr): Likewise.
38170 (gimplify_function_tree): Conditionally initialize
38171 asan_poisoned_variables.
38173 2017-01-04 Jeff Law <law@redhat.com>
38175 PR tree-optimizatin/78812
38176 * rtl.h (contains_mem_rtx_p): Prototype.
38177 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38178 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38179 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38180 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38181 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38183 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38185 * input.c (assert_char_at_range): Default-initialize actual_range.
38187 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38189 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38190 to match the caller.
38192 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38194 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38195 insns after final jump in test to emit dummy move.
38197 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38199 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38200 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38202 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38204 * multiple_target.c (create_dispatcher_calls): Init e_next.
38205 * tree-ssa-loop-split.c (split_loop): Init border.
38206 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38209 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38214 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38215 (altivec_register_operand): Do not return true if the operand
38216 contains a SUBREG mixing SImode and SFmode.
38217 (vsx_register_operand): Likewise.
38218 (vsx_reg_sfsubreg_ok): New predicate.
38219 (vfloat_operand): Do not return true if the operand contains a
38220 SUBREG mixing SImode and SFmode.
38221 (vint_operand): Likewise.
38222 (vlogical_operand): Likewise.
38223 (gpc_reg_operand): Likewise.
38224 (int_reg_operand): Likewise.
38225 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38226 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38227 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38229 (rs6000_emit_move_si_sf_subreg): New helper function.
38230 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38231 fixup SUBREGs involving SImode and SFmode.
38232 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38233 numbers for the new peephole2 optimization.
38234 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38235 the GLIBC math library that do AND/IOR/XOR operations on single
38236 precision floating point.
38237 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38238 target macros to say whether we need to avoid SUBREGs mixing
38240 (TARGET_ALLOW_SF_SUBREG): Likewise.
38241 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38242 (UNSPEC_SI_FROM_SF): Likewise.
38243 (iorxor): Change spacing.
38244 (and_ior_xor): New iterator for AND, IOR, and XOR.
38245 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38246 (movdi_from_sf_zero_ext): Likewise.
38247 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38248 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38249 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38250 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38251 (fms<mode>4): Likewise.
38252 (fnma<mode>4): Likewise.
38253 (fnms<mode>4): Likewise.
38254 (nfma<mode>4): Likewise.
38255 (nfms<mode>4): Likewise.
38257 2017-01-04 Marek Polacek <polacek@redhat.com>
38260 * doc/invoke.texi: Document -Wpointer-compare.
38262 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38264 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38267 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38268 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38271 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38274 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38275 documentation of the powerpc_popcntb_ok attribute.
38276 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38277 code to issue warning messages if a requested CPU configuration is
38278 not supported by the binary (assembler and loader) toolchain.
38279 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38280 made to define a built-in function that has been disabled.
38281 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38282 made to define a built-in function that has been disabled.
38283 (altivec_init_builtins): Add comment explaining why definition
38284 of the DST built-in functions is not preceded by an assertion
38285 check. Add assertions to prevent ICE if attempts are made to
38286 define an altivec predicate or an abs* built-in function that has
38288 (htm_init_builtins): Add comment explaining why definition of the
38289 htm built-in functions is not preceded by an assertion check.
38291 2017-01-04 Jeff Law <law@redhat.com>
38293 PR tree-optimizatin/67955
38294 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38295 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38296 the points-to solution does not include pt_null. Use DECL_PT_UID
38299 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38301 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38302 Use gen_int_mode instead of gen_lopwart for const_int operands.
38304 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38306 PR tree-optimization/71563
38307 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38308 out of range value - has low bits known to be zero.
38310 2017-01-04 Alan Modra <amodra@gmail.com>
38312 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38313 * configure: Regenerate.
38314 * config.in: Regenerate.
38316 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38319 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38320 a substring of the message, but strcmp with the whole message. Ifdef
38321 ENABLE_NLS, translate the message first using dgettext.
38323 2017-01-03 Jeff Law <law@redhat.com>
38325 PR tree-optimizatin/78856
38326 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38327 (mark_threaded_blocks): Remove code to truncate thread paths that
38328 cross multiple loop headers. Instead invalidate the cached loop
38329 iteration information and handle case of a thread path walking
38330 into an irreducible region.
38332 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38335 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38336 assertions. Add support for doing the signbit if the IEEE 128-bit
38337 floating point value is in a GPR.
38338 * config/rs6000/rs6000.md (Fsignbit): Delete.
38339 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38340 Update the length attribute if the value is in a GPR.
38341 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38342 the sign or zero extension instruction, since the value is always 0/1.
38343 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38346 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38347 extracting SImode to a GPR register so that we can generate a
38348 store, limit the vector to be in a traditional Altivec register
38349 for the vextuwrx instruction.
38351 2017-01-03 Ian Lance Taylor <iant@google.com>
38353 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38355 2017-01-03 Martin Sebor <msebor@redhat.com>
38357 PR tree-optimization/78696
38358 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38359 precision. Use MPFR for %f for greater fidelity. Correct handling
38361 (pass_sprintf_length::compute_format_length): Set width and precision
38362 specified by asrerisk to void_node for vararg functions.
38363 (try_substitute_return_value): Adjust dump output.
38365 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38367 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38369 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38371 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38372 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38373 -mlra/-mno-lra was passed to the compiler.
38375 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38377 PR rtl-optimization/65618
38378 * emit-rtl.c (try_split): Move initialization of "before" and
38379 "after" to just before the call to emit_insn_after_setloc.
38381 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38383 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38385 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38387 * dwarf2out.c (gen_enumeration_type_die): When
38388 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38390 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38392 PR tree-optimization/78965
38393 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38394 Change first argument from const call_info & to call_info &. For %n
38395 set info.nowrite to false.
38397 PR middle-end/78901
38398 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38399 possibly throwing calls.
38401 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38402 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38403 and fns handling, rather than in a separate case SSA_NAME.
38405 2017-01-02 Jeff Law <law@redhat.com>
38407 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38408 fixes for first_period and second_period variables.
38410 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38413 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38414 (*insvqi_1): New insn pattern.
38415 (*insvqi_1_mem_rex64): Ditto.
38416 (*insvqi_2): Ditto.
38417 (*insvqi_3): Rename from *insvqi.
38419 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38421 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38423 * doc/cfg.texi (Edges): Remove reference to Java.
38424 (Maintaining the CFG): Ditto.
38426 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38428 PR middle-end/77674
38429 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38430 transparent aliases.
38432 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38434 PR middle-end/77484
38435 * predict.def (PRED_CALL): Update hitrate.
38436 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38437 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38438 into direct/indirect/polymorphic variants.
38440 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38442 Update copyright years.
38444 * gcc.c (process_command): Update copyright notice dates.
38445 * gcov-dump.c (print_version): Ditto.
38446 * gcov.c (print_version): Ditto.
38447 * gcov-tool.c (print_version): Ditto.
38448 * gengtype.c (create_file): Ditto.
38449 * doc/cpp.texi: Bump @copying's copyright year.
38450 * doc/cppinternals.texi: Ditto.
38451 * doc/gcc.texi: Ditto.
38452 * doc/gccint.texi: Ditto.
38453 * doc/gcov.texi: Ditto.
38454 * doc/install.texi: Ditto.
38455 * doc/invoke.texi: Ditto.
38457 Copyright (C) 2017 Free Software Foundation, Inc.
38459 Copying and distribution of this file, with or without modification,
38460 are permitted in any medium without royalty provided the copyright
38461 notice and this notice are preserved.