1 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3 * tree-vector-builder.h
4 (tree_vector_builder::binary_encoded_nelts): Declare.
5 * tree-vector-builder.c
6 (tree_vector_builder::binary_encoded_nelts): New function.
7 * fold-const.c (negate_expr_p): Likewise.
8 (operand_equal_p, fold_checksum_tree): Likewise.
9 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
10 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
11 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
12 (uniform_vector_p): Likewise.
13 * varasm.c (const_hash_1, compare_constant): Likewise.
14 * tree-ssa-ccp.c: Include tree-vector-builder.h.
15 (valid_lattice_transition): Operate directly on the VECTOR_CST
17 * ipa-icf.c: Include tree-vector-builder.h.
18 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
19 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
21 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
23 * tree.c (build_vector): Delete.
24 * tree.h (build_vector): Make static and move into the self-testing
27 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
29 * vector-builder.h (vector_builder::derived): New const overload.
30 (vector_builder::elt): New function.
31 * tree-vector-builder.h (tree_vector_builder::type): New function.
32 (tree_vector_builder::apply_step): Declare.
33 * tree-vector-builder.c (tree_vector_builder::apply_step): New
35 * gimple-fold.h (tree_vector_builder): Declare.
36 (gimple_build_vector): Take a tree_vector_builder instead of a
37 type and vector of elements.
38 * gimple-fold.c (gimple_build_vector): Likewise.
39 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
41 (get_initial_defs_for_reduction): Likewise.
42 (vectorizable_induction): Likewise.
44 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
46 * tree-vector-builder.h
47 (tree_vector_builder::new_binary_operation): Declare.
48 * tree-vector-builder.c
49 (tree_vector_builder::new_binary_operation): New function.
50 * fold-const.c (fold_relational_const): Use it.
51 (const_binop): Likewise. Check that both input vectors have
52 the same number of elements, thus excluding things like WIDEN_SUM.
53 Check whether it is possible to operate directly on the encodings
56 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
58 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
59 new_unary_operation, operating only on the encoded elements.
60 (const_unop): Likewise.
61 (exact_inverse): Likewise.
62 (distributes_over_addition_p): New function.
63 (const_binop): Use tree_vector_builder and new_unary_operation
64 for combinations of VECTOR_CST and INTEGER_CST. Operate only
65 on the encoded elements unless the encoding is strided and the
66 operation does not distribute over addition.
67 (fold_convert_const): Use tree_vector_builder and
68 new_unary_operation. Operate only on the encoded elements
69 for truncating integer conversions, or for non-stepped encodings.
71 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
73 * config/sparc/sparc.c: Include tree-vector-builder.h.
74 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
75 * expmed.c: Include tree-vector-builder.h.
76 (make_tree): Use tree_vector_builder instead of build_vector.
77 * fold-const.c: Include tree-vector-builder.h.
78 (const_binop): Use tree_vector_builder instead of build_vector.
79 (const_unop): Likewise.
80 (native_interpret_vector): Likewise.
81 (fold_vec_perm): Likewise.
82 (fold_ternary_loc): Likewise.
83 * gimple-fold.c: Include tree-vector-builder.h.
84 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
86 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
87 (simplify_vector_constructor): Use tree_vector_builder instead
89 * tree-vect-generic.c: Include tree-vector-builder.h.
90 (add_rshift): Use tree_vector_builder instead of build_vector.
91 (expand_vector_divmod): Likewise.
92 (optimize_vector_constructor): Likewise.
93 * tree-vect-loop.c: Include tree-vector-builder.h.
94 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
95 of build_vector. Explicitly use a stepped encoding for
97 * tree-vect-slp.c: Include tree-vector-builder.h.
98 (vect_get_constant_vectors): Use tree_vector_builder instead
100 (vect_transform_slp_perm_load): Likewise.
101 (vect_schedule_slp_instance): Likewise.
102 * tree-vect-stmts.c: Include tree-vector-builder.h.
103 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
104 (vect_gen_perm_mask_any): Likewise.
105 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
106 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
108 (build_vector_from_val): Likewise. Explicitly use a duplicate
111 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
113 * doc/generic.texi (VECTOR_CST): Describe new representation of
115 * vector-builder.h: New file.
116 * tree-vector-builder.h: Likewise.
117 * tree-vector-builder.c: Likewise.
118 * Makefile.in (OBJS): Add tree-vector-builder.o.
119 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
120 * tree-core.h (tree_base): Add a vector_cst field to the u union.
121 (tree_vector): Change the number of elements to
122 vector_cst_encoded_nelts.
123 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
124 (VECTOR_CST_ELTS): Delete.
125 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
126 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
127 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
128 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
129 (VECTOR_CST_ENCODED_ELT): Likewise.
130 (vector_cst_encoded_nelts): New function.
131 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
132 VECTOR_CST_NELTS_PER_PATTERN as arguments.
133 (vector_cst_int_elt, vector_cst_elt): Declare.
134 * tree.c: Include tree-vector-builder.h.
135 (tree_code_size): Abort if passed VECTOR_CST.
136 (tree_size): Update for new VECTOR_CST layout.
137 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
138 VECTOR_CST_NELTS_PER_PATTERN as arguments.
139 (build_vector): Use tree_vector_builder.
140 (vector_cst_int_elt, vector_cst_elt): New functions.
141 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
142 encoded elements and then create the vector in the canonical form.
143 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
144 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
145 (tree_c_tests): Call test_vector_cst_patterns.
146 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
148 (hash_tree): Likewise.
149 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
150 (streamer_write_tree_header): Likewise.
151 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
152 (streamer_alloc_tree): Likewise. Update call to make_vector.
153 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
155 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
157 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
158 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
159 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
161 2017-12-07 Bin Cheng <bin.cheng@arm.com>
162 Richard Biener <rguenther@suse.de>
164 PR tree-optimization/81303
165 * Makefile.in (gimple-loop-interchange.o): New object file.
166 * common.opt (floop-interchange): Reuse the option from graphite.
167 * doc/invoke.texi (-floop-interchange): Ditto. New document for
168 -floop-interchange and mention it for -O3.
169 * opts.c (default_options_table): Enable -floop-interchange at -O3.
170 * gimple-loop-interchange.cc: New file.
171 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
172 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
173 * passes.def (pass_linterchange): New pass.
174 * timevar.def (TV_LINTERCHANGE): New time var.
175 * tree-pass.h (make_pass_linterchange): New declaration.
176 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
177 interchange. Record IV before/after increment in new parameters.
178 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
179 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
181 (check_reduction_path): ...New function here.
182 * tree-vectorizer.h (check_reduction_path): New declaration.
184 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
187 PR rtl-optimization/80818
188 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
189 always early clobbered.
190 * lra-lives.c (process_bb_lives): Check input hard regs for early
191 clobbered non-operand hard reg.
193 2017-12-07 Jakub Jelinek <jakub@redhat.com>
196 * tree-cfg.c (verify_gimple_assign_binary): Don't require
197 types_compatible_p, just that TYPE_MODE is the same.
199 2017-12-07 Martin Sebor <msebor@redhat.com>
202 * attribs.c (empty_attribute_table): Initialize new member of
203 struct attribute_spec.
204 (decl_attributes): Add argument. Handle mutually exclusive
205 combinations of attributes.
206 (selftests::test_attribute_exclusions): New function.
207 (selftests::attribute_c_tests): Ditto.
208 * attribs.h (decl_attributes): Add default argument.
209 * selftest.h (attribute_c_tests): Declare.
210 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
211 * tree-core.h (attribute_spec::exclusions, exclude): New type and
213 * doc/extend.texi (Common Function Attributes): Update const and pure.
214 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
215 of struct attribute_spec.
216 * config/arc/arc.c (arc_attribute_table): Same.
217 * config/arm/arm.c (arm_attribute_table): Same.
218 * config/avr/avr.c ( avr_attribute_table): Same.
219 * config/bfin/bfin.c (bfin_attribute_table): Same.
220 * config/cr16/cr16.c (cr16_attribute_table): Same.
221 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
222 * config/h8300/h8300.c (h8300_attribute_table): Same.
223 * config/i386/i386.c (ix86_attribute_table): Same.
224 * config/ia64/ia64.c (ia64_attribute_table): Same.
225 * config/m32c/m32c.c (m32c_attribute_table): Same.
226 * config/m32r/m32r.c (m32r_attribute_table): Same.
227 * config/m68k/m68k.c (m68k_attribute_table): Same.
228 * config/mcore/mcore.c (mcore_attribute_table): Same.
229 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
230 * config/mips/mips.c (mips_attribute_table): Same.
231 * config/msp430/msp430.c (msp430_attribute_table): Same.
232 * config/nds32/nds32.c (nds32_attribute_table): Same.
233 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
234 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
235 * config/rl78/rl78.c (rl78__attribute_table): Same.
236 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
237 * onfig/rx/rx.c (rx_attribute_table): Same.
238 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
239 * config/sh/sh.c (sh_attribute_table): Same.
240 * config/sparc/sparc.c (sparc_attribute_table): Same.
241 * config/spu/spu.c (spu_attribute_table): Same.
242 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
243 * config/v850/v850.c (v850_attribute_table): Same.
244 * config/visium/visium.c (visium_attribute_table): Same.
246 2017-12-07 Tamar Christina <tamar.christina@arm.com>
249 * config/arm/arm.c (INCLUDE_STRING): Define.
250 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
251 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
254 2017-12-07 Michael Matz <matz@suse.de>
256 Add unroll and jam pass
258 * gimple-loop-jam.c: New file.
259 * Makefile.in (OBJS): Add gimple-loop-jam.o.
260 * common.opt (funroll-and-jam): New option.
261 * opts.c (default_options_table): Add unroll-and-jam at -O3.
262 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
263 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
264 * passes.def: Add pass_loop_jam.
265 * timevar.def (TV_LOOP_JAM): Add.
266 * tree-pass.h (make_pass_loop_jam): Declare.
267 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
268 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
269 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
270 to flow_loop_tree_node_add.
271 (duplicate_subloops, copy_loops_to): Append to sibling list.
272 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
273 * doc/invoke.texi (-funroll-and-jam): Document new option.
274 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
276 2017-12-07 Richard Biener <rguenther@suse.de>
278 PR tree-optimization/83296
279 PR tree-optimization/67769
280 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
281 flow sensitive info in an unrelated BB.
282 (value_replacement): Use reset_flow_sensitive_info.
283 (minmax_replacement): Reset flow sensitive info on the def
284 we move. Do not reset flow sensitive info in the whole BB
286 (abs_replacement): Likewise.
288 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
291 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
292 rs6000_cpu to the given -mcpu=, or to the default processor.
294 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
296 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
297 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
298 instead of rs6000_cpu_attr.
299 (rs6000_adjust_cost): Ditto.
300 (is_microcoded_insn): Ditto.
301 (rs6000_adjust_priority): Ditto.
302 (rs6000_issue_rate): Ditto.
303 (rs6000_use_sched_lookahead): Ditto.
304 (rs6000_use_sched_lookahead_guard): Ditto.
305 (rs6000_sched_reorder): Ditto.
306 (force_new_group): Ditto.
307 * config/rs6000/rs6000.md (cpu attribute): Ditto.
308 (group_ending_nop): Ditto.
310 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
312 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
313 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
314 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
315 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
317 (rs6000_reassociation_width): Ditto.
318 (rs6000_emit_epilogue): Ditto.
319 (rs6000_adjust_cost): Ditto.
320 (is_microcoded_insn): Ditto.
321 (is_cracked_insn): Ditto.
322 (rs6000_adjust_priority): Ditto.
323 (rs6000_sched_reorder): Ditto.
324 (rs6000_sched_reorder2): Ditto.
325 (insn_must_be_first_in_group): Ditto.
326 (insn_must_be_last_in_group): Ditto.
327 (rs6000_register_move_cost): Ditto.
328 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
331 2017-12-07 Julia Koval <julia.koval@intel.com>
333 * config.gcc: Add vaesintrin.h.
334 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI): New type.
335 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
336 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi): New builtins.
337 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
338 * config/i386/immintrin.h: Include vaesintrin.h.
339 * config/i386/sse.md (vaesdec_<mode>): New pattern.
340 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
341 _mm_aesdec_epi128): New intrinsics.
343 2017-12-06 David Malcolm <dmalcolm@redhat.com>
345 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
346 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
347 c-family/c-spellcheck.cc.
348 (best_macro_match::best_macro_match): Likewise.
350 (struct edit_distance_traits<cpp_hashnode *>): Move to
351 c-family/c-spellcheck.h.
352 (class best_macro_match): Likewise.
354 2017-12-06 Jakub Jelinek <jakub@redhat.com>
356 PR tree-optimization/83293
357 * gimple-ssa-strength-reduction.c (insert_initializers): Use
358 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
359 might insert into empty bb.
362 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
363 simplify for plus with :c added, and pointer_plus without that.
364 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
365 with undefined overflow and the conversion is not widening,
366 perform negation in utype and only convert to type afterwards.
367 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
368 simplify for plus with :c added, and pointer_plus without that.
369 If type is integral with undefined overflow and the conversion is
370 not widening, perform minus in utype and only convert to type
371 afterwards. Move the last pointer_diff_expr simplify into the
374 2017-12-06 Martin Sebor <msebor@redhat.com>
376 PR tree-optimization/82646
377 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
380 2017-12-06 Martin Sebor <msebor@redhat.com>
382 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
385 PR tree-optimization/83075
386 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
387 strncat/strncpy don't change length of source string.
389 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
392 2017-11-29 Martin Aberg <maberg@gaisler.com>
394 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
395 to prevent b2bst errata sequence.
396 (sqrtdf2_fix): Likewise.
398 2017-12-06 Jakub Jelinek <jakub@redhat.com>
400 PR tree-optimization/81945
401 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
402 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
403 to dest_cfun has orig_loop_num set, either remap it to the new
404 loop number if the loop got moved too, or clear it.
406 2017-12-05 Steve Ellcey <sellcey@cavium.com>
408 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
410 (thunderx2t99_nothing): New insn reservation.
411 (thunderx2t99_mrs): New insn reservation.
412 (thunderx2t99_multiple): New insn reservation.
413 (thunderx2t99_alu_basi): Add bfx to reservation.
414 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
416 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
419 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
421 2017-12-05 Bin Cheng <bin.cheng@arm.com>
423 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
424 tree-ssa-pre.c::remove_dead_inserted_code.
425 * tree-ssa-dce.h: New file.
426 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
427 (remove_dead_inserted_code): Move and rename to function
428 tree-ssa-dce.c::simple_dce_from_worklist.
429 (pass_pre::execute): Update use.
431 2017-12-05 Richard Biener <rguenther@suse.de>
433 PR tree-optimization/83277
434 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
435 to code-gen liveout vars.
437 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
439 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
440 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
441 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
444 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
446 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
448 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
449 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
452 2017-12-05 Richard Biener <rguenther@suse.de>
454 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL):
456 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
457 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
458 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
460 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
462 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
464 2017-12-05 Martin Liska <mliska@suse.cz>
465 Jakub Jelinek <jakub@redhat.com>
467 * doc/invoke.texi: Document the options.
468 * flag-types.h (enum sanitize_code): Add
469 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
470 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
471 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
472 * opts.c: Define new sanitizer options.
473 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
474 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
476 2017-12-05 Julia Koval <julia.koval@intel.com>
478 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
479 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
480 (ix86_handle_option): Handle -mavx512vnni.
481 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
482 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
483 * config/i386/i386-c (__AVX512VNNI__): New.
484 * config/i386/i386.c (ix86_target_string): Handle new option.
485 (ix86_valid_target_attribute_inner_p): Handle new option.
486 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
487 * config/i386/i386.opt (mavx512vnni): New option.
489 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
492 * athlon.md: Disable for generic.
493 * haswell.md: Enable for generic.
494 * i386.c (ix86_sched_init_global): Add core hooks for generic.
495 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
497 (ix86_adjust_cost): Move generic to haswell path.
499 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
501 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
502 instead of MEM_P in a couple more places. Fix formatting issues.
504 2017-12-04 Jim Wilson <jimw@sifive.com>
506 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
507 instead of GP_REG_LAST-1.
508 (riscv_adjust_libcall_cfi_prologue): Likewise.
509 (riscv_adjust_libcall_cri_epilogue): Likewise.
510 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
513 2017-12-04 Luis Machado <luis.machado@linaro.org>
515 * ipa-pure-const.c (check_decl): Add missing newline.
516 (state_from_flags): Likewise.
518 2017-12-04 Jeff Law <law@redhat.com>
520 PR tree-optimizatin/78496
521 * gimple-ssa-evrp-analyze.h
522 (evrp_range_analyzer::get_vr_values): Simplify.
523 * gimple-ssa-evrp-analyze.c: Corresponding changes.
524 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
525 and gimple-ssa-evrp-analyze.h.
526 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
527 (simplify_stmt_for_jump_threading): Copy a blob of code from
528 tree-vrp.c to use ranges to simplify statements.
529 (dom_opt_dom_walker::before_dom_children): Call
530 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
531 (dom_opt_dom_walker::after_dom_children): Similarly for
532 evrp_range_analyzer::leave.
533 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
536 * gimple-ssa-evrp-analyze.c
537 (evrp_range_analyzer::extract_range_from_stmt): Always use
538 vr_values::update_value_range so preexisting range info is
539 medged with new range info, even if the new range is VR_VARYING.
541 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
543 * combine.c: Adjust comment.
544 (use_crosses_set_p): Delete.
545 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
546 (try_combine): Ditto.
548 2017-12-04 Richard Biener <rguenther@suse.de>
550 PR tree-optimization/83255
551 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
552 Re-add zero-iteration check.
554 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
556 PR rtl-optimization/83245
557 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
558 hard registers as earlyclobber, also if not in an asm.
560 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
564 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
567 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
568 rs6000_cpu based on cpu_index, not tune_index.
570 2017-12-04 Richard Biener <rguenther@suse.de>
572 PR tree-optimization/83238
573 * graphite-scop-detection.c (scop_detection::merge_sese): Make
574 code match comment, rejecting invalid SESE regions.
576 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
578 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
579 require base operand is a REG_POINTER prior to reload on targets
580 with non-equivalent space registers.
582 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
584 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
585 (update_specialized_profile): Fix updating of counts.
586 (perhaps_add_new_callers): Likewise.
588 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
591 * x86-tune.def: Remove obsolette FIXMEs.
592 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
593 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
594 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
596 (X86_TUNE_PAD_RETURNS): Disable for generic.
598 2017-12-02 Jakub Jelinek <jakub@redhat.com>
600 PR tree-optimization/83170
601 PR tree-optimization/83241
602 * gimple-ssa-store-merging.c
603 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
604 gimple_vuse (ins_stmt) in case it has changed.
605 (imm_store_chain_info::output_merged_store): Likewise.
607 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
611 * tree-cfg.c (pass_warn_function_return::execute): Handle
612 __builtin_ubsan_handle_missing_return like __builtin_unreachable
613 with BUILTINS_LOCATION.
617 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
618 is BLKmode for vector field with vector raw mode, use TYPE_MODE
619 instead of DECL_MODE.
621 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
622 last argument to rtx pointer.
623 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
624 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
625 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
626 depending on the chosen ISAs.
627 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
628 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
629 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
631 * config/i386/sse.md (mov<mode>_internal): Likewise.
632 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
634 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
636 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
638 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
640 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
641 parameters from prototype.
642 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
643 parameters. Don't print a comment.
644 (emit_fusion_gpr_load): Adjust.
645 (emit_fusion_load_store): Adjust.
646 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
647 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
648 comment on the second line.
650 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
653 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
654 rs6000_cpu based on cpu_index, not tune_index.
656 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
658 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
659 which_alternative instead of which_alternative + 1.
660 (output_asm_insn): Print an extra tab if the template is short.
662 2017-12-01 Jim Wilson <jimw@sifive.com>
664 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
666 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
667 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
669 * doc/tm.texi: Regenerate.
671 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
674 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
675 whether we can allocate pseudos before trying to fix an address.
676 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
677 memory address is indexed or indirect.
678 (floatuns_<mode>si2_hw2): Likewise.
680 2017-12-01 Jason Merrill <jason@redhat.com>
682 * Makefile.in (TAGS): Add c-family/*.cc.
684 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
686 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
687 (call_value_insn): Likewise.
688 (sibcall_insn): Likewise.
689 (sibcall_value_insn): Likewise.
690 (movsi_aarch64): Likewise.
691 (movdi_aarch64): Likewise.
692 (add_losym_): Likewise.
693 (ldr_got_small_): Likewise.
694 (ldr_got_small_sidi): Likewise.
695 (ldr_got_small_28k_): Likewise.
696 (ldr_got_small_28k_sidi): Likewise.
697 * config/aarch64/aarch64.c (aarch64_print_address_internal):
698 Move output_addr_const to symbolic case. Add error check.
700 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
702 * config/s390/predicates.md (plus16_Q_operand): New predicate.
703 * config/s390/s390.md: Disable MVC merging peephole if it would
704 disable operand forwarding.
705 (new peephole2): Split MVCs if it would turn them into up to 2
708 2017-12-01 Richard Biener <rguenther@suse.de>
710 PR tree-optimization/83232
711 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
712 detection of same access. Instead of breaking the group here
713 do not consider the duplicate. Add comment explaining real fix.
715 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
717 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
719 2017-12-01 Sudakshina Das <sudi.das@arm.com>
721 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
723 2017-12-01 Jakub Jelinek <jakub@redhat.com>
725 * function.h (struct function): Remove cilk_frame_decl,
726 is_cilk_function and calls_cilk_spawn fields.
727 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
729 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
730 cilk_elemental field.
731 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
732 * target.def: Adjust comment.
733 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
734 Don't test cilk_elemental.
736 PR tree-optimization/83233
737 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
738 bswap_stat name for the struct.
741 * tree.h (SWITCH_BREAK_LABEL_P): Define.
742 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
743 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
744 SWITCH_BREAK_LABEL_P set on the label.
745 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
746 added for default case if it was missing and not all cases covered.
747 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
748 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
750 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
751 to avoid -Wimplicit-fallthrough warning.
752 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
753 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
755 PR tree-optimization/83221
756 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
758 (init_reassoc): Formatting fix.
761 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
762 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
766 * tree-cfg.c: Include asan.h.
767 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
769 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
771 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
773 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
775 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
776 vpcmpeqd instruction.
778 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
780 * config/i386/i386.c (standard_sse_constant_opcode): Fix
781 registers type for 128bit mode.
783 2017-11-30 Jakub Jelinek <jakub@redhat.com>
785 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
787 * gengtype-state.c (read_state_pair): Likewise.
788 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
789 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
790 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
791 * ifcvt.c (noce_try_store_flag_constants): Likewise.
792 * tree-ssa-ccp.c (ccp_finalize): Likewise.
793 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
794 * builtins.c (fold_builtin_3): Likewise.
795 * graphite-scop-detection.c
796 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
797 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
800 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
803 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
804 parameter. Return true if ignore_decl_rtl_set_p is true and other
805 conditions are satisfied.
806 * asan.h (asan_protect_global): Add new parameter.
807 * varasm.c (categorize_decl_for_section): Pass true as second parameter
808 to asan_protect_global calls.
810 2017-11-30 Jim Wilson <jimw@sifive.com>
812 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
813 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
814 -mstrict-align, add info on default value. Delete redundant lines for
815 -mabi. Add missing -mexplicit-relocs docs.
817 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
819 * config/arc/arc.md (trap): New pattern.
821 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
823 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
824 instruction to end into a delay slot.
825 * config/arc/arc.md (cond_delay_insn): Check if the instruction
826 can be placed into a delay slot against reg_note.
828 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
830 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
831 labels number of usages.
833 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
835 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
837 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
839 2017-11-30 Jakub Jelinek <jakub@redhat.com>
842 * internal-fn.c (expand_mul_overflow): Optimize unsigned
843 multiplication by power of 2 constant into two shifts + comparison.
845 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
848 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
850 2017-11-30 Richard Biener <rguenther@suse.de>
852 PR tree-optimization/83202
853 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
854 allow_peel argument and guard peeling.
855 (canonicalize_loop_induction_variables): Likewise.
856 (canonicalize_induction_variables): Pass false.
857 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
858 peeling from cunrolli.
860 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
862 * combine.c (try_combine): Print a message to dump file whenever
863 I0, I1, or I2 cannot be combined into I3.
865 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
867 PR rtl-optimization/83156
868 PR rtl-optimization/82621
869 * combine.c (try_combine): Don't split an I2 if one of the dests is
870 set again before I3. Allow unused dests.
872 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
874 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
876 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
878 PR rtl-optimization/80818
879 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
880 recursively. Use insn code for clobber.
881 (lra_set_insn_recog_data): Pass the new arg to
882 collect_non_operand_hard_regs.
883 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
885 (lra_update_insn_regno_info): Pass insn to
886 add_regs_to_insn_regno_info.
888 2017-11-29 Jim Wilson <jimw@sifive.com>
889 Andrew Waterman <andrew@sifive.com>
891 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
892 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
893 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
894 use SINGLE_SHIFT_COST.
895 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
896 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
898 2017-11-29 Julia Koval <julia.koval@intel.com>
900 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
901 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
902 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
903 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
904 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
905 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
906 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
907 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
908 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
909 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
910 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
911 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
912 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
913 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
914 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
915 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
916 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
917 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
918 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
919 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
920 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
921 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
922 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
923 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
924 __builtin_ia32_vpshldv_v2di_mask,
925 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
926 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
927 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
929 2017-11-29 Julia Koval <julia.koval@intel.com>
931 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
932 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
933 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
934 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
935 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
936 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
937 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
938 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
939 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
940 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
941 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
942 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
943 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
944 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
945 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
946 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
947 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
948 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
949 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
950 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
951 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
953 2017-11-29 Daniel Cederman <cederman@gaisler.com>
955 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
956 movsi_pic_gotdata_op instruction as a load for the UT699 errata
959 2017-11-29 Martin Aberg <maberg@gaisler.com>
961 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
962 to prevent b2bst errata sequence.
963 (sqrtdf2_fix): Likewise.
965 2017-11-29 Daniel Cederman <cederman@gaisler.com>
967 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
968 (div_sqrt_insn_p): New function.
969 (sparc_do_work_around_errata): Insert NOP instructions to
970 prevent sequences that could trigger the TN-0013 errata for
971 certain LEON3 processors.
972 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
973 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
974 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
975 (in_branch_delay): Prevent div and sqrt in delay slot if
977 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
979 2017-11-29 Daniel Cederman <cederman@gaisler.com>
981 * config/sparc/sparc.c (atomic_insn_p): New function.
982 (sparc_do_work_around_errata): Insert NOP instructions to
983 prevent sequences that could trigger the TN-0010 errata for
985 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
986 instruction referable in atomic_insns_p.
988 2017-11-29 Daniel Cederman <cederman@gaisler.com>
990 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
991 (atomic_compare_and_swap_leon3_1): Likewise.
994 2017-11-29 Daniel Cederman <cederman@gaisler.com>
996 * config/sparc/sparc.c (fpop_insn_p): New function.
997 (sparc_do_work_around_errata): Insert NOP instructions to
998 prevent sequences that could trigger the TN-0012 errata for
1000 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1001 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1002 (in_branch_annul_delay): Prevent floating-point instructions
1003 in delay slot of annulled integer branch.
1005 2017-11-29 Richard Biener <rguenther@suse.de>
1007 PR tree-optimization/83202
1008 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1010 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1011 nodes vectorized to the same stmts multiple times.
1012 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1013 (vect_analyze_slp_instance): Adjust.
1014 (scalar_stmts_to_slp_tree_map_t): New typedef.
1015 (vect_schedule_slp_instance): Add a map recording the SLP node
1016 representing the vectorized stmts for a set of scalar stmts.
1017 Avoid code-generating redundancies.
1018 (vect_schedule_slp): Allocate map and pass it down.
1020 2017-11-29 Nathan Sidwell <nathan@acm.org>
1023 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1026 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1028 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1029 cases for printing LDP/STP memory addresses.
1030 (aarch64_print_address_internal): Renamed from
1031 aarch64_print_operand_address, added parameter, add Pmode check.
1032 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1033 (aarch64_print_operand_address): Indirect to
1034 aarch64_print_address_internal.
1035 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1036 'y' operand output specifier.
1038 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1041 * tree.c (build_simple_mem_ref_loc): Handle
1042 get_addr_base_and_unit_offset returning a MEM_REF.
1045 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1048 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1051 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1053 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1055 (layout::move_to_column): Likewise.
1057 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1060 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1061 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1063 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1064 Martin Jambor <mjambor@suse.cz>
1067 * tree.h (expr_type_first_operand_type_p): Declare
1068 * tree.c (expr_type_first_operand_type_p): New function.
1069 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1070 (ipa_value_from_jfunc): Adjust declaration.
1071 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1072 Use it as result type for arithmetics, unless it is NULL in which case
1073 be more conservative.
1074 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1075 ipa_get_jf_pass_through_result.
1076 (propagate_vals_across_pass_through): Likewise.
1077 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1078 is to propagate_vals_across_pass_through.
1079 (propagate_constants_across_call): Pass PARM_TYPE to
1080 propagate_scalar_across_jump_function.
1081 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1082 ipa_value_from_jfunc.
1083 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1084 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1085 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1086 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1087 target_type, pass it to ipa_value_from_jfunc.
1088 (update_indirect_edges_after_inlining): Pass parameter type to
1089 try_make_edge_direct_simple_call.
1091 2017-11-28 Jeff Law <law@redhat.com>
1093 * gimple-ssa-evrp-analyze.c
1094 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1095 refine ranges if scev_initialized_p returns true.
1096 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1098 2017-11-28 Julia Koval <julia.koval@intel.com>
1100 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1101 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1102 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1103 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1104 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1105 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1106 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1107 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1108 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1109 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1110 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1111 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1112 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1113 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1114 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1115 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1116 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1117 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1118 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1119 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1120 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1121 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1123 2017-11-28 Julia Koval <julia.koval@intel.com>
1125 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1126 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1127 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1128 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1129 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1130 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1131 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1132 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1133 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1134 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1135 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1136 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1137 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1138 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1139 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1140 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1141 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1142 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1143 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1144 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1145 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1146 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1147 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1148 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1149 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1150 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1151 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1152 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1153 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1154 __builtin_ia32_vpshld_v2di_mask): New builtins.
1155 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1156 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1158 2017-11-28 Richard Biener <rguenther@suse.de>
1160 PR tree-optimization/80776
1161 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1163 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1165 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1166 If the incoming edge is an effective fallthru because the other
1167 edge only reaches a __builtin_unreachable () then record ranges
1168 derived from the controlling condition in SSA info.
1169 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1170 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1172 2017-11-28 Olivier Hainque <hainque@adacore.com>
1174 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1175 on mingw build hosts.
1177 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1180 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1182 2017-11-28 Richard Biener <rguenther@suse.de>
1185 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1186 copies generated from memcpy use a character array as reference
1189 2017-11-28 Julia Koval <julia.koval@intel.com>
1190 Sebastian Peryt <sebastian.peryt@intel.com>
1192 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1193 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1194 cilk-common.o, cilk.h, cilk-common.c): Remove.
1196 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1197 * builtins.c (is_builtin_name): Remove cilkplus condition.
1198 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1199 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1200 cilk-builtins.def, cilkplus.def): Remove.
1201 * cif-code.def (CILK_SPAWN): Remove.
1202 * cilk-builtins.def: Delete.
1203 * cilk-common.c: Ditto.
1205 * cilkplus.def: Ditto.
1206 * config/darwin.h (fcilkplus): Delete.
1207 * cppbuiltin.c: Ditto.
1208 * doc/extend.texi: Remove cilkplus doc.
1209 * doc/generic.texi: Ditto.
1210 * doc/invoke.texi: Ditto.
1211 * doc/passes.texi: Ditto.
1212 * gcc.c (fcilkplus): Remove.
1213 * gengtype.c (cilk.h): Remove.
1214 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
1216 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
1218 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
1220 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
1221 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
1222 cilkplus conditions.
1223 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
1224 inline_read_section): Ditto.
1225 * ipa-inline-analysis.c (cilk.h): Remove.
1226 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
1227 * lto-wrapper.c (merge_and_complain, append_compiler_options,
1228 append_linker_options): Remove condition for fcilkplus.
1229 * lto/lto-lang.c (cilk.h): Remove.
1230 (lto_init): Remove condition for fcilkplus.
1231 * omp-expand.c (expand_cilk_for_call): Delete.
1232 (expand_omp_taskreg, expand_omp_for_static_chunk,
1233 expand_omp_for): Remove cilkplus
1235 (expand_cilk_for): Delete.
1236 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
1237 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
1238 execute_lower_omp, diagnose_sb_0): Ditto.
1239 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
1240 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1241 * tree-nested.c: Ditto.
1242 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
1243 (dump_generic_node): Ditto.
1244 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1245 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
1247 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
1249 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1251 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
1252 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
1253 for vector mode and !TARGET_SIMD.
1255 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1257 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
1259 * tree.h (SWITCH_LABELS): Remove.
1260 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
1261 assert SWITCH_BODY is non-NULL.
1262 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
1264 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
1266 PR tree-optimization/80788
1267 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
1268 has TREE_OVERFLOW set, call drop_tree_overflow.
1270 2017-11-28 Richard Biener <rguenther@suse.de>
1272 PR tree-optimization/83158
1273 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
1276 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
1279 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
1280 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
1282 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
1284 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
1285 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
1287 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1290 * varasm.c (compare_constant): Take the mode of the constants into
1291 account when comparing floating point constants.
1293 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
1295 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
1296 from explicit instantiation of debug_helper.
1297 * vec.h (DEFINE_DEBUG_VEC): Ditto.
1299 2017-11-27 Richard Biener <rguenther@suse.de>
1301 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
1304 2017-11-27 Richard Biener <rguenther@suse.de>
1306 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
1307 (type_hash_canon): Also clear int_cst_hash_table entry for
1309 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
1311 2017-11-27 Tamar Christina <tamar.christina@arm.com>
1313 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1314 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1315 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1317 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1319 * hash-map.h (gt_cleare_cache): Avoid UB.
1321 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
1323 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
1324 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
1325 (decide_unroll_runtime_iterations): Likewise.
1326 (decide_unroll_stupid): Likewise.
1328 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1331 * config/i386/i386.md: Add a loop with incssp.
1333 2017-11-27 Martin Jambor <mjambor@suse.cz>
1335 PR tree-optimization/81248
1336 * tree-sra.c (splice_param_accesses): Remove size check.
1337 (decide_one_param_reduction): Fix size check.
1338 * gimple-pretty-print.c (dump_profile): Silence warning.
1339 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
1341 2017-11-27 Jakub Jelinek <jakub@redhat.com>
1344 * dbxout.c (lastlineno): New variable.
1345 (dbx_debug_hooks): Use dbxout_switch_text_section as
1346 switch_text_section debug hook.
1347 (dbxout_function_end): Switch to current_function_section
1348 rather than function_section. If crtl->has_bb_partition,
1349 output just one N_FUN, depending on in_cold_section_p.
1350 (dbxout_source_line): Remember last lineno in lastlineno.
1351 (dbxout_switch_text_section): New function.
1352 (dbxout_function_decl): Adjust dbxout_block caller.
1353 (dbx_block_with_cold_children): New function.
1354 (dbxout_block): Return true if any LBRAC/RBRAC have been
1355 emitted. Use dbx_block_with_cold_children at depth == 0
1356 in second partition. Add PARENT_BLOCKNUM argument, pass
1357 it optionally adjusted to children. Output LBRAC/RBRAC
1358 around recursive call only if the block is in the current
1359 partition, if not and anything was output, emit empty
1361 * final.c (final_scan_insn): Compute cold_function_name
1362 before calling switch_text_section debug hook. Call
1363 that hook even if dwarf2out_do_frame if not emitting
1367 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
1368 TREE_READONLY decls.
1370 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1372 PR rtl-optimization/82488
1373 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
1375 2017-11-26 Julia Koval <julia.koval@intel.com>
1377 * config/i386/i386.c (processor_target_table): Add skylake_cost for
1379 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
1382 2017-11-26 Julia Koval <julia.koval@intel.com>
1384 * config/i386/driver-i386.c (host_detect_local_cpu):
1385 Detect skylake-avx512.
1387 2017-11-26 Julia Koval <julia.koval@intel.com>
1389 * config.gcc: Add -march=cannonlake.
1390 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
1391 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
1392 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
1393 (PTA_CANNONLAKE): New.
1394 (processor_target_table): Add cannonlake.
1395 (ix86_option_override_internal): Ditto.
1396 (fold_builtin_cpu): Ditto.
1397 (get_builtin_code_for_version): Handle cannonlake.
1398 (M_INTEL_COREI7_CANNONLAKE): New.
1399 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
1400 * doc/invoke.texi: Add -march=cannonlake.
1402 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
1404 * plugin.c (add_new_plugin): Use platform-specific library extensions.
1405 (try_init_one_plugin): Alternative implementation for MinGW.
1406 * Makefile.in (plugin_implib): New.
1407 (gengtype-lex.c): Fix broken AIX workaround.
1408 * configure: Regenerate.
1409 * doc/plugins.texi: Document support for MinGW.
1411 2017-11-25 Jakub Jelinek <jakub@redhat.com>
1413 PR rtl-optimization/81553
1414 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
1415 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
1416 is a shift where C1 has different mode than the whole shift, use C1's
1417 mode for MULT rather than the shift's mode.
1420 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
1421 builtins not enabled in the currently selected ISA.
1423 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
1425 PR tree-optimization/71026
1426 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
1427 (insert_reciprocals): Change to insert reciprocals before a division
1428 by a square and to insert the square of a reciprocal.
1429 (execute_cse_reciprocals_1): Change to consider division by a square.
1430 (register_division_in): Add importance parameter.
1432 2017-11-24 Richard Biener <rguenther@suse.de>
1434 PR tree-optimization/82402
1435 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
1436 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
1438 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
1440 * match.pd (0-ptr): New transformation.
1442 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
1445 * ipa-inline.c (inline_small_functions): Set current badnes correctly
1446 when skipping checking.
1448 2017-11-24 Richard Biener <rguenther@suse.de>
1450 PR tree-optimization/83128
1451 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
1452 (vn_reference_lookup_3): Likewise.
1454 2017-11-24 Jakub Jelinek <jakub@redhat.com>
1457 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
1458 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
1461 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
1462 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
1463 check for the latter with gimple_call_builtin_p. Do not handle
1464 BUILT_IN_STPNCPY_CHK which is not a pass through call.
1466 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
1468 * config/arm/arm_neon.h: Fix pragma GCC push_options before
1471 2017-11-23 Julia Koval <julia.koval@intel.com>
1473 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
1474 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
1475 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
1476 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
1477 _mm512_maskz_expandloadu_epi16): New intrinsics.
1478 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
1479 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
1480 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
1481 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
1482 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
1483 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
1484 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
1485 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
1486 _mm256_maskz_expandloadu_epi8): New intrinsics.
1487 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
1488 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
1489 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
1490 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
1491 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
1492 * config/i386/sse.md (VI248_VLBW): New iterator.
1493 (expand<mode>_mask, expand<mode>_maskz): New patterns.
1495 2017-11-23 Julia Koval <julia.koval@intel.com>
1497 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
1498 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
1499 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
1500 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
1501 _mm512_mask_compressstoreu_epi16): New.
1502 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
1503 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
1504 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
1505 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
1506 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
1507 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
1508 _mm256_mask_compressstoreu_epi8): New.
1509 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
1510 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
1511 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
1512 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
1513 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
1514 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
1515 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
1516 __builtin_ia32_compresshi128_mask,
1517 __builtin_ia32_compressstoreuqi512_mask,
1518 __builtin_ia32_compressstoreuhi512_mask,
1519 __builtin_ia32_compressstoreuqi256_mask,
1520 __builtin_ia32_compressstoreuqi128_mask,
1521 __builtin_ia32_compressstoreuhi256_mask,
1522 __builtin_ia32_compressstoreuhi128_mask): New builtins.
1523 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
1525 (ix86_expand_special_args_builtin): Handle new types.
1526 (s4fma_expand): Handle new builtin array.
1527 * config/i386/immintrin.h: Include new headers.
1528 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
1529 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
1531 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1533 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
1534 fixes. Declare temp and g variables at the top in order to avoid
1535 {} in most of the cases.
1537 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
1539 * match.pd (ptr-0): New transformation.
1541 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1543 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
1544 (struct addr_mode_cost_table): New.
1545 (struct tune_params): Add field addr_mode_costs.
1546 * config/arm/arm.c (generic_addr_mode_costs): New.
1547 (arm_slowmul_tune): Initialise addr_mode_costs field.
1548 (arm_fastmul_tune): Likewise.
1549 (arm_strongarm_tune): Likewise.
1550 (arm_xscale_tune): Likewise.
1551 (arm_9e_tune): Likewise.
1552 (arm_marvell_pj4_tune): Likewise.
1553 (arm_v6t2_tune): Likewise.
1554 (arm_cortex_tune): Likewise.
1555 (arm_cortex_a8_tune): Likewise.
1556 (arm_cortex_a7_tune): Likewise.
1557 (arm_cortex_a15_tune): Likewise.
1558 (arm_cortex_a35_tune): Likewise.
1559 (arm_cortex_a53_tune): Likewise.
1560 (arm_cortex_a57_tune): Likewise.
1561 (arm_exynosm1_tune): Likewise.
1562 (arm_xgene1_tune): Likewise.
1563 (arm_cortex_a5_tune): Likewise.
1564 (arm_cortex_a9_tune): Likewise.
1565 (arm_cortex_a12_tune): Likewise.
1566 (arm_cortex_a73_tune): Likewise.
1567 (arm_v7m_tune): Likewise.
1568 (arm_cortex_m7_tune): Likewise.
1569 (arm_v6m_tune): Likewise.
1570 (arm_fa726te_tune): Likewise.
1571 (arm_mem_costs): Use table lookup to calculate cost of addressing
1574 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1576 * config/arm/arm.c (arm_mem_costs): New function.
1577 (arm_rtx_costs_internal): Use arm_mem_costs.
1579 2017-11-23 Mark Wielaard <mark@klomp.org>
1581 * dwarf2out.c (init_sections_and_labels): Use generation to create
1582 unique ranges_section_label and ranges_base_label. Return generation.
1583 (output_rnglists): Add generation argument. Use generation to create
1584 unique ranges labels.
1585 (dwarf2out_finish): Get generation from init_sections_and_labels
1586 and pass generation to output_rnglists.
1588 2017-11-23 Mike Stump <mikestump@comcast.net>
1589 Eric Botcazou <ebotcazou@adacore.com>
1591 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
1592 * cfgloop.h (struct loop): Add unroll field.
1593 * function.h (struct function): Add has_unroll bitfield.
1594 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
1595 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
1596 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
1597 (pass_rtl_unroll_loops::gate): Likewise.
1598 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
1599 for which loop->unroll==1.
1600 (decide_unroll_constant_iterations): Use note for consistency and
1601 take loop->unroll into account. Return early if loop->unroll is set.
1602 Fix thinko in existing test.
1603 (decide_unroll_runtime_iterations): Use note for consistency and
1604 take loop->unroll into account.
1605 (decide_unroll_stupid): Likewise.
1606 * lto-streamer-in.c (input_cfg): Read loop->unroll.
1607 * lto-streamer-out.c (output_cfg): Write loop->unroll.
1608 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
1610 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
1611 (print_loop): Print loop->unroll if set.
1612 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
1613 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
1614 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
1616 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
1617 loop->unroll is set and smaller than the trip count. Otherwise bypass
1618 entirely the heuristics if loop->unroll is set. Remove dead note.
1619 Fix off-by-one bug in other note.
1620 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
1621 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
1623 (tree_unroll_loops_completely): Make static.
1624 (pass_complete_unroll::execute): Use correct type for variable.
1625 (pass_complete_unrolli::execute): Fix formatting.
1626 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
1628 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
1630 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
1631 enable when TARGET_PREFER_AVX128 is set.
1633 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1635 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
1637 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
1638 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
1640 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1642 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
1643 frequencies. Fix estimation of aggregate parameters.
1645 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1647 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
1650 2017-11-23 Tom de Vries <tom@codesourcery.com>
1652 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
1654 2017-11-23 Tom de Vries <tom@codesourcery.com>
1656 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
1659 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
1662 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
1663 sibcall_value_pcrel_fdpic): Use local variable instead of
1665 (calli_tbr_rel): Add missing operand 2.
1666 (call_valuei_tbr_rel): Add missing operand 3.
1668 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1671 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
1672 bitpos/bitsize covering the whole destination, use store_expr only if
1673 the complex mode is the same. Otherwise, use expand_normal and if
1674 it returns CONCAT, subreg each part separately instead of trying to
1675 subreg the whole result.
1677 2017-11-23 Richard Biener <rguenther@suse.de>
1679 PR tree-optimization/23094
1680 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
1681 come here from walking over backedges in the first iteration.
1682 (vn_reference_lookup_3): Skip clobbers that store the same value.
1684 2017-11-23 Richard Biener <rguenther@suse.de>
1686 PR tree-optimization/81403
1687 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
1688 a block we need a leader relative to.
1689 (phi_translate_1): For nary processing require a leader from
1690 get_representative_for given we run expression simplification
1691 using match-and-simplify. Remove previous fix.
1693 2017-11-22 Jeff Law <law@redhat.com>
1695 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
1696 Use new method allocate_value_range rather than accessing the
1697 vrp_value_range_pool data member directly.
1698 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
1699 to use extract_range_from_stmt method to avoid need for
1700 extract_range_from_assignment method.
1701 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
1702 method rather than setting values_propgated data member directly.
1703 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
1704 and values propagated data members and extract_range_from_assignment
1705 method. Reorder private data members to conform to standards.
1706 Add new methods set_lattice_propagation_complete and
1707 allocate_value_range.
1709 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
1711 PR rtl-optimization/83030
1712 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
1713 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
1714 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
1715 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
1716 CROSSING_JUMP_P flag is already set before setting it.
1717 * cfgrtl.c (fixup_partition_crossing): Likewise.
1718 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
1721 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1723 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
1724 Handle the case where both arguments are using gen_const_vec_series.
1726 2017-11-22 David Malcolm <dmalcolm@redhat.com>
1729 * pretty-print.c (pp_format): Move quoting implementation to
1730 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
1731 to pass address of "quote" local.
1732 (pp_begin_quote): New function.
1733 (pp_end_quote): New function.
1734 * pretty-print.h (printer_fn): Convert penultimate param from bool
1736 (pp_begin_quote): New decl.
1737 (pp_end_quote): New decl.
1738 * tree-diagnostic.c (default_tree_printer): Convert penultimate
1739 param from bool to bool *.
1740 * tree-diagnostic.h (default_tree_printer): Likewise.
1742 2017-11-22 Jeff Law <law@redhat.com>
1744 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
1745 Initialize vr_values.
1746 (evrp_range_analyzer::try_find_new_range): Call methods attached to
1747 vr_values via vr_values class instance rather than delegators.
1748 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
1749 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1750 (evrp_range_analyzer::push_value_range): Likewise.
1751 (evrp_range_analyzer::pop_value_range): Likewise.
1752 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
1753 most delegators. Those remaining are exposed as public interfaces.
1754 Make vr_values a pointer and private.
1755 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
1757 (evrp_range_analyzer::get_vr_value): New method.
1758 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
1759 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
1760 (class evrp_dom_walker): Attach evrp_folder class, initialize
1761 it in the ctor. Remove temporary delegators.
1762 (evrp_dom_walker::before_dom_children): Call methods in attached
1763 evrp_range_analyzer class via class instance pointer. Use
1764 free value_range_constant_singleton to remove need for
1765 op_with_constant_singleton_value delegator method. Do not
1766 create a vrp_prop class instance for every call! Narrow
1767 scope of a couple variables.
1768 (evrp_dom_walker::cleanup): Call methods in attached
1769 evrp_range_analyzer class via class instance pointer.
1770 * vr-values.h (class vr_values): Privatize many methods and
1773 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1775 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
1776 * cfgexpand.c (expand_debug_expr): Remove handling for them.
1777 * expr.c (expand_expr_real_2): Likewise.
1778 * fold-const.c (const_unop): Likewise.
1779 * optabs-tree.c (optab_for_tree_code): Likewise.
1780 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1781 * tree-inline.c (estimate_operator_cost): Likewise.
1782 * tree-pretty-print.c (dump_generic_node): Likewise.
1783 (op_code_prio): Likewise.
1784 (op_symbol_code): Likewise.
1785 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
1786 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
1787 * internal-fn.c (direct_internal_fn_optab): New function.
1788 (direct_internal_fn_array, direct_internal_fn_supported_p
1789 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
1790 * fold-const-call.c (fold_const_reduction): New function.
1791 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
1793 * tree-vect-loop.c: Include internal-fn.h.
1794 (reduction_code_for_scalar_code): Rename to...
1795 (reduction_fn_for_scalar_code): ...this and return an internal
1797 (vect_model_reduction_cost): Take an internal_fn rather than
1799 (vect_create_epilog_for_reduction): Likewise. Build calls rather
1801 (vectorizable_reduction): Use internal functions rather than tree
1802 codes for the reduction operation. Update calls to the functions
1804 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
1805 Use calls to internal functions rather than REDUC tree codes.
1806 * config/aarch64/aarch64-simd.md: Update comment accordingly.
1808 2017-11-22 Olivier Hainque <hainque@adacore.com>
1810 * config/vxworks.c (vxworks_override_options): Pick default
1811 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
1812 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
1813 DWARF_GNAT_ENCODINGS_DEFAULT.
1814 * config/vxworksae.h: Likewise.
1816 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
1818 PR tree-optimization/83104
1819 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
1822 2017-11-22 Marek Polacek <polacek@redhat.com>
1823 H.J. Lu <hongjiu.lu@intel.com>
1824 Jason Merrill <jason@redhat.com>
1829 * calls.c (initialize_argument_information): Call
1830 warn_parameter_passing_abi target hook.
1831 (store_one_arg): Use 0 for empty record size. Don't push 0 size
1832 argument onto stack.
1833 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
1834 * common.opt: Update -fabi-version description.
1835 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
1836 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
1838 (ix86_is_empty_record): New function.
1839 (ix86_warn_parameter_passing_abi): New function.
1840 (TARGET_EMPTY_RECORD_P): Redefine.
1841 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
1842 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
1843 * doc/tm.texi: Regenerated.
1844 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
1845 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
1846 * dwarf2out.c (get_ultimate_context): Move to tree.c.
1847 * explow.c (hard_function_value): Call arg_int_size_in_bytes
1848 instead of int_size_in_bytes.
1849 * expr.c (copy_blkmode_to_reg): Likewise.
1850 * function.c (aggregate_value_p): Return 0 for empty types.
1851 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
1852 (locate_and_pad_parm): Call arg size_in_bytes instead
1854 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
1855 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
1856 * target.def (empty_record_p, warn_parameter_passing_abi): New target
1858 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
1859 (std_gimplify_va_arg_expr): Skip empty records. Call
1860 arg_size_in_bytes instead size_in_bytes.
1861 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
1862 * tree-core.h (tree_type_common): Add empty_flag.
1863 (tree_decl_common): Update comments.
1864 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
1866 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1867 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
1869 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1870 * tree.c (default_is_empty_type): New function.
1871 (default_is_empty_record): New function.
1872 (arg_int_size_in_bytes): New function.
1873 (arg_size_in_bytes): New function.
1874 (get_ultimate_context): New function.
1875 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
1876 TRANSLATION_UNIT_WARN_EMPTY_P.
1877 (default_is_empty_record, arg_int_size_in_bytes,
1878 arg_size_in_bytes, get_ultimate_context): Declare.
1880 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1882 * config/arm/arm.c (cmse_clear_registers): New function.
1883 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
1884 code by call to cmse_clear_registers.
1885 (cmse_nonsecure_entry_clear_before_return): Likewise.
1887 2017-11-22 Tamar Christina <tamar.christina@arm.com>
1889 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
1890 (vdot_s32, vdotq_s32): New.
1891 (vdot_lane_u32, vdotq_lane_u32): New.
1892 (vdot_lane_s32, vdotq_lane_s32): New.
1895 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1898 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
1899 for unsigned values with fewer HWIs than the precision.
1900 (test_overflow): New function.
1901 (wide_int_cc_tests): Call it.
1903 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1905 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
1906 has the same mode class as Pmode.
1908 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1910 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
1911 padding_bits_to_clear_ptr.
1912 (cmse_nonsecure_entry_clear_before_return): Likewise.
1914 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1916 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
1917 auto_sbitap instead of integer bitfield to control register needing
1920 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1922 PR tree-optimization/83044
1923 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
1924 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
1925 that need the upper bound. Subtract offset from
1926 get_addr_base_and_unit_offset only if positive and subtract it
1927 before division by eltsize rather than after it.
1930 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
1931 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
1932 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
1936 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
1937 expand_binop with *mul_widen_optab, make sure at least one of the
1938 operands doesn't have VOIDmode.
1941 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
1943 PR rtl-optimization/82044
1944 PR tree-optimization/82042
1945 * dse.c (record_store): Check for overflow.
1946 (check_mem_read_rtx): Properly check for overflow if width == -1, call
1947 add_wild_read instead of clear_rhs_from_active_local_stores on
1948 overflow and log it into dump_file.
1950 2017-11-22 Richard Biener <rguenther@suse.de>
1952 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
1953 fake edges to exit when looking for a place to insert.
1954 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
1956 (insert_into_preds_of_block): Commit edge insertion immediately,
1957 assert that doesn't require new BBs.
1958 (fini_pre): Release expressions.
1959 (pass_pre::execute): Shuffle things around a bit, if the fn
1960 is too large do not compute AVAIL either as this is really the
1963 2017-11-22 Richard Biener <rguenther@suse.de>
1965 PR tree-optimization/83089
1966 * tree-if-conv.c (pass_if_conversion::execute): If anything
1967 changed reset SCEV and free the number of iteration estimates.
1969 2017-11-21 Martin Sebor <msebor@redhat.com>
1971 PR tree-optimization/82945
1972 * calls.h (warn_nonstring_bound): Remove unused function.
1974 2017-11-21 Martin Sebor <msebor@redhat.com>
1976 PR tree-optimization/82945
1977 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
1978 * calls.h (maybe_warn_nonstring_arg): Declare new function.
1979 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
1981 (initialize_argument_information): Call maybe_warn_nonstring_arg.
1982 * calls.h (get_attr_nonstring_decl): Declare new function.
1983 * doc/extend.texi (attribute nonstring): Update.
1984 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
1985 get_attr_nonstring_decl and handle it.
1986 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
1987 detection of nul-termination.
1988 (strlen_to_stridx): Change to a pointer.
1989 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
1990 (pass_strlen::execute): Same.
1992 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
1994 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
1995 for the new option -mprefer-vector-width=[none|128|256|512].
1996 * config/i386/i386.c (ix86_target_string): remove old style options
1997 -mprefer-avx256 and make -mprefer-avx128 as alias.
1998 (ix86_option_override_internal): Apply defaults for the
1999 -mprefer-vector-width=[128|256] option.
2000 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2001 Implement macros to work with -mprefer-vector-width=.
2002 * config/i386/i386.opt: Implemented option
2003 -mprefer-vector-width=[none|128|256|512].
2004 * doc/invoke.texi: Documentation for
2005 -mprefer-vector-width=[none|128|256|512].
2007 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2009 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2011 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2013 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2015 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2017 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2019 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2020 * expr.c (expand_expr_real_2): Likewise.
2021 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2022 fold_binary_loc): Likewise.
2023 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2024 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2025 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2026 MINUS_EXPR transformations.
2027 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2028 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2029 * tree-inline.c (estimate_operator_cost): Likewise.
2030 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2031 op_symbol_code): Likewise.
2032 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2033 * vr-values.c (extract_range_from_binary_expr): Likewise.
2034 * varasm.c (initializer_constant_valid_p_1): Likewise.
2035 * tree.def: New tree code POINTER_DIFF_EXPR.
2037 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2039 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2040 integer suffix to movbe mnemonic.
2041 (*bswaphi2_movbe): Ditto.
2042 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2044 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2047 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2048 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2049 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2051 2017-11-21 Martin Liska <mliska@suse.cz>
2053 * tree-inline.c (expand_call_inline): Remove not needed
2056 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2057 Jakub Jelinek <jakub@redhat.com>
2060 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2061 Remove static keyword from f variable.
2063 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2065 PR tree-optimization/83086
2066 * gimple-ssa-store-merging.c
2067 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2068 rather than n.base_addr.
2070 2017-11-21 Martin Liska <mliska@suse.cz>
2072 PR rtl-optimization/82044
2073 PR tree-optimization/82042
2074 * dse.c (check_mem_read_rtx): Check for overflow.
2076 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2078 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2081 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2083 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2084 and its corresponding call arg location note.
2086 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2087 Andrew Burgess <andrew.burgess@embecosm.com>
2089 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2091 (arc_return_slot_offset): Likewise.
2092 (arc_eh_return_address_location): New declaration.
2093 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2094 (MUST_SAVE_REGISTER): Add exception handler case.
2095 (MUST_SAVE_RETURN_ADDR): Likewise.
2096 (arc_frame_pointer_required): Likewise.
2097 (arc_frame_pointer_needed): New function.
2098 (arc_compute_frame_size): Changed.
2099 (arc_expand_prologue): Likewise.
2100 (arc_expand_epilogue): Likewise.
2101 (arc_initial_elimination_offset): Likewise.
2102 (arc_return_slot_offset): Delete.
2103 (arc_eh_return_address_location): New function.
2104 (arc_builtin_setjmp_frame_value): Likewise.
2105 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2106 (EH_RETURN_STACKADJ_RTX): Define.
2107 (EH_RETURN_HANDLER_RTX): Likewise.
2108 * config/arc/arc.md (eh_return): Delete.
2110 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2112 * print-tree.h (debug_vec_tree): Remove prototype.
2113 * gdbinit.in (pvt): Remove macro.
2115 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2117 PR tree-optimization/83047
2118 * gimple-ssa-store-merging.c
2119 (imm_store_chain_info::output_merged_store): If the loads with the
2120 same vuse are in different basic blocks, for load_gsi pick a load
2121 location that is dominated by the other loads.
2124 * config/i386/i386.c (ix86_memmodel_check): Start
2125 -Winvalid-memory-model diagnostics with lowercase letter.
2128 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2129 set in_cold_section_p to the partition containing loc_list->first.
2130 When seeing loc_list->last_before_switch node, update secname and
2131 perform range_across_switch second partition handling only after that.
2134 * run-rtl-passes.c: Include debug.h.
2135 (run_rtl_passes): Call debug_hooks->assembly_start.
2136 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2140 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2141 OPTAB_DIRECT in calls to expand_simple_binop.
2143 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2146 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2148 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2151 * spellcheck.h (best_match::blithely_get_best_candidate): New
2154 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2156 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2157 with lower case letter.
2159 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2161 * config/i386/i386.md (bswaphi2): New expander.
2162 (*bswaphi2_movbe): New insn pattern.
2163 (bswaphi -> rorhi pepehole2): New peephole pattern.
2165 2017-11-20 Jeff Law <law@redhat.com>
2167 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2168 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2169 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2170 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2171 gimple-ssa-evrp-analyze.h.
2173 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2174 set BB_VISITED here.
2175 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2177 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2179 * tree-predcom.c: Add general comment on Store-Store chains.
2180 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2182 (get_chain_last_ref_at): Rename into...
2183 (get_chain_last_write_at): ...this.
2184 (get_chain_last_write_before_load): New function.
2185 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2186 CT_STORE_STORE when write reference is added.
2187 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2188 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2189 (initialize_root_vars_store_elim_1): Ditto.
2190 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2191 definition is created.
2192 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2193 chain by replacing it with dominant stored value.
2195 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2197 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2199 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2201 * vr-values.c (extract_range_from_binary_expr): Use a full range
2204 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2206 * config/arm/arm.md (R4_REGNUM): Define constant.
2207 (nonsecure_call_internal): Remove r4 clobber.
2208 (nonsecure_call_value_internal): Likewise.
2209 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2210 clobber and resequence match_operands.
2211 (nonsecure_call_value_reg_thumb1_v5): Likewise.
2212 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
2213 (nonsecure_call_value_reg_thumb2): Likewise.
2215 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2217 PR tree-optimization/78821
2218 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
2219 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
2220 to the address of the base rather than the base itself.
2221 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
2222 (find_bswap_or_nop_finalize): New function.
2223 (find_bswap_or_nop): Use it.
2224 (bswap_replace): Return a tree rather than bool, change first
2225 argument from gimple * to gimple_stmt_iterator, allow inserting
2226 into an empty sequence, allow ins_stmt to be NULL - then emit
2227 all stmts into gsi. Fix up MEM_REF address gimplification.
2228 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
2229 (struct store_immediate_info): Add N and INS_STMT non-static
2231 (store_immediate_info::store_immediate_info): Initialize them
2232 from newly added ctor args.
2233 (merged_store_group::apply_stores): Formatting fixes. Sort by
2235 (stmts_may_clobber_ref_p): For stores call also
2236 refs_anti_dependent_p.
2237 (gather_bswap_load_refs): New function.
2238 (imm_store_chain_info::try_coalesce_bswap): New method.
2239 (imm_store_chain_info::coalesce_immediate_stores): Use it.
2240 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
2241 (imm_store_chain_info::output_merged_store): Fail if number of
2242 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
2243 and NOP_EXPR rhs_code.
2244 (pass_store_merging::process_store): Compute n and ins_stmt, if
2245 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
2246 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
2248 (pass_store_merging::execute): Calculate dominators.
2250 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
2251 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
2252 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
2253 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2254 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
2255 class pass_optimize_bswap, bswap_replace,
2256 pass_optimize_bswap::execute): Moved to ...
2257 * gimple-ssa-store-merging.c: ... this file.
2258 Include optabs-tree.h.
2259 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
2260 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2261 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
2262 anonymous namespace, remove static keywords.
2263 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
2264 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
2266 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
2269 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
2271 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
2273 * vec.h (debug_helper): New function.
2274 (DEFINE_DEBUG_VEC): New macro.
2275 * hash-set.h (debug_helper): New function.
2276 (DEFINE_DEBUG_HASH_SET): New macro.
2277 * cfg.c (debug_slim (edge)): New function.
2278 Call DEFINE_DEBUG_VEC for edges.
2279 Call DEFINE_DEBUG_HASH_SET for edges.
2280 * cfghooks.c (debug_slim (basic_block)): New function.
2281 Call DEFINE_DEBUG_VEC for basic blocks.
2282 Call DEFINE_DEBUG_HASH_SET for basic blocks.
2283 * print-tree.c (debug_slim): New function to handle trees.
2284 Call DEFINE_DEBUG_VEC for trees.
2285 Call DEFINE_DEBUG_HASH_SET for trees.
2286 (debug (vec<tree, va_gc>) &): Remove.
2287 (debug (<vec<tree, va_gc>) *): Remove.
2288 * print-rtl.c (debug_slim): New function to handle const_rtx.
2289 Call DEFINE_DEBUG_VEC for rtx_def.
2290 Call DEFINE_DEBUG_VEC for rtx_insn.
2291 Call DEFINE_DEBUG_HASH_SET for rtx_def.
2292 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
2293 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
2294 (debug (vec<rtx_insn *> *ptr): Remove.
2295 (debug_insn_vector): Remove.
2296 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
2298 2017-11-20 Tom de Vries <tom@codesourcery.com>
2300 PR rtl-optimization/82020
2301 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
2302 IF_THEN_ELSE condition.
2304 2017-11-19 Jeff Law <law@redhat.com>
2306 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2307 of degenerates resulting from ignoring an edge.
2309 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2312 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
2314 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2317 * profile-count.c (profile_count::to_sreal_scale): Fix return value
2318 for uninitialied counts.
2320 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2323 * tree-inline.c (estimate_num_insns): Set to 1 at least.
2325 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2328 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
2330 2017-11-19 Tom de Vries <tom@codesourcery.com>
2332 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
2335 2017-11-19 Tom de Vries <tom@codesourcery.com>
2337 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
2340 2017-11-19 Tom de Vries <tom@codesourcery.com>
2342 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2345 2017-11-19 Tom de Vries <tom@codesourcery.com>
2347 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
2349 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
2351 2017-11-19 Tom de Vries <tom@codesourcery.com>
2353 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
2355 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
2356 semicolon after MCORE_EXPORT_NAME call.
2358 2017-11-19 Tom de Vries <tom@codesourcery.com>
2361 * vmsdbgout.c (vmsdbgout_early_finish): New function.
2362 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
2364 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
2366 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
2367 (cgraph_edge::clone): Cleanup updating of profile.
2368 * ipa-cp.c (update_profiling_info): Likewise.
2369 * ipa-inline-transform.c (inline_transform): Likewise.
2370 * ipa-inline.c (inline_small_functions): Add missing space to dump.
2371 * ipa-split.c (execute_split_functions): Do not split when function
2373 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
2374 * profile-count.c (profile_count::dump): Add global0.
2375 (profile_count::to_cgraph_frequency): Do not ICE when entry is
2377 (profile_count::to_sreal_scale): Likewise.
2378 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
2379 (profile_count::combine_with_ipa_count): New function.
2380 * profile-count.h (profile_guessed_global0adjusted): New.
2381 (profile_count::adjusted_zero): New.
2382 (profile_count::global0adjusted): New.
2383 (profile_count::combine_with_ipa_count): New.
2384 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
2385 correct profile of return block of split functions.
2386 (copy_cfg_body): Remove unused profile_count.
2387 (copy_body): Likewise.
2388 (expand_call_inline): Update.
2389 (tree_function_versioning): Update.
2391 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
2393 * hash-set.h (hash_set::empty): New.
2394 * tree-ssa-threadbackward.h: Delete.
2395 * tree-ssa-threadbackward.c (class thread_jumps): New.
2396 Move max_threaded_paths into class.
2397 (fsm_find_thread_path): Remove arguments that are now in class.
2398 (profitable_jump_thread_path): Rename to...
2399 (thread_jumps::profitable_jump_thread_path): ...this.
2400 (convert_and_register_jump_thread_path): Rename to...
2401 (thread_jumps::convert_and_register_current_path): ...this.
2402 (check_subpath_and_update_thread_path): Rename to...
2403 (thread_jumps::check_subpath_and_update_thread_path): ...this.
2404 (register_jump_thread_path_if_profitable): Rename to...
2405 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
2406 (handle_phi): Rename to...
2407 (thread_jumps::handle_phi): ...this.
2408 (handle_assignment): Rename to...
2409 (thread_jumps::handle_assignment): ...this.
2410 (fsm_find_control_statement_thread_paths): Rename to...
2411 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
2412 (find_jump_threads_backwards): Rename to...
2413 (thread_jumps::find_jump_threads_backwards): ...this.
2414 Initialize path local data.
2415 (pass_thread_jumps::execute): Call find_jump_threads_backwards
2416 from within thread_jumps class.
2417 (pass_early_thread_jumps::execute): Same.
2419 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2421 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
2423 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2425 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
2426 to canonical location.
2428 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2431 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
2433 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
2435 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
2437 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
2438 rcrt1.o%s/grcrt1.o%s for -static-pie.
2440 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2442 * i386.c (ix86_multiplication_cost, ix86_division_cost,
2443 ix86_shift_rotate_cost): Break out from ...
2444 (ix86_rtx_costs): ... here.
2445 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
2448 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2450 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
2451 when entry block was promoted unlikely.
2452 (estimate_bb_frequencies): Increase frequency scale.
2453 * profile-count.h (profile_count): Export precision info.
2455 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2457 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
2458 disturbin profile of entry block.
2460 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2462 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
2465 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2467 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
2468 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
2470 2017-11-17 Jeff Law <law@redhat.com>
2472 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
2473 from evrp_dom_walker class. Various methods moved into new class.
2474 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
2475 (evrp_range_analyzer::enter): New method.
2476 (evrp_range_analyzer::leave): New method.
2477 (evrp_dom_walker): Remove delegators no longer needed by this class.
2478 Replace vr_values data member with evrp_range_analyzer
2480 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
2481 method extracted from evrp_dom_walker::before_dom_children.
2482 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
2483 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
2485 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
2486 Add private copy constructor and move assignment operators.
2487 Privatize methods and class data where trivially possible.
2488 (evrp_dom_walker::cleanup): New function, extracted from
2489 execute_early_vrp. Simplify access to class data.
2491 * vr-values.h (get_output_for_vrp): Prototype.
2492 * vr-values.c (get_output_for_vrp): New function extracted from
2493 vrp_visit_assignment_or_call and extract_range_from_stmt.
2494 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
2496 2017-11-17 Luis Machado <luis.machado@linaro.org>
2498 * config/aarch64/aarch64.c
2499 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
2500 (qdf24xx_tunings) <autoprefetcher_model>: Set to
2501 tune_params::AUTOPREFETCHER_WEAK.
2503 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2506 * config/arm/arm.c (arm_valid_target_attribute_rec):
2507 Parse "arch=" and "+<ext>".
2508 (arm_valid_target_attribute_tree): Re-init global options.
2509 (arm_option_override): Make non-static.
2510 (arm_options_perform_arch_sanity_checks): Make errors fatal.
2511 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
2512 (__ARM_FEATURE_CRC32): Support undef.
2513 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
2514 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
2516 2017-11-17 David Malcolm <dmalcolm@redhat.com>
2518 * gdbinit.in (break-on-diagnostic): New command.
2520 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2522 * config/i386/i386.c (ix86_expand_epilogue): Change simple
2523 return to indirect jump for EH return if control-flow
2524 protection is enabled. Change explicit 'false' argument in
2525 pro_epilogue_adjust_stack with a value of
2527 * config/i386/i386.md (simple_return_indirect_internal):
2528 Remove SImode restriction to support 64-bit.
2530 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2532 * combine.c (added_notes_insn): New.
2533 (try_combine): Handle added_notes_insn like added_links_insn.
2534 Rewrite return value code.
2535 (distribute_notes): Set added_notes_insn to the earliest insn we added
2538 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2540 PR rtl-optimization/82621
2541 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
2542 dest of one of those SETs is unused.
2544 2017-11-17 Richard Biener <rguenther@suse.de>
2547 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
2548 * tree-pretty-print.c (dump_generic_node): Handle
2549 annot_expr_parallel_kind.
2550 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
2551 * gimplify.c (gimple_boolify): Likewise.
2553 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2555 * config.gcc (extra_headers): Add cet.h for x86 targets.
2556 * config/i386/cet.h: New file.
2557 * doc/install.texi: Add --enable-cet/--disable-cet.
2559 2017-11-17 Richard Biener <rguenther@suse.de>
2561 PR tree-optimization/83017
2562 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
2563 (gen_parallel_loop): Properly count iterations.
2564 (parallelize_loops): Handle loop->can_be_parallel independent
2565 of flag_loop_parallelize_all. Make static profitability test match
2567 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
2568 * invoke.texi (parloops-min-per-thread): Document.
2570 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
2572 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
2573 upstreaming review comments.
2575 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2577 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
2578 with fast unaligned access.
2579 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
2581 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
2583 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
2584 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
2586 2017-11-17 Richard Biener <rguenther@suse.de>
2588 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
2589 folding of references.
2591 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
2594 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
2595 of replacing call to strncmp with corresponding call to strcmp when
2598 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
2600 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
2601 option prefer-avx256 for skylake-avx512 configuration.
2602 * config/i386/i386.c (ix86_option_override_internal): Ditto.
2603 (get_builtin_code_for_version): Ditto.
2605 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2606 Monk Chiang <sh.chiang04@gmail.com>
2608 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
2609 (FIXED_REGISTERS): Reserve more register numbers.
2610 (CALL_USED_REGISTERS): Likewise.
2611 (REG_ALLOC_ORDER): Likewise.
2612 (REG_CLASS_CONTENTS): Likewise.
2613 (REGISTER_NAMES): Likewise.
2615 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2616 Kito Cheng <kito.cheng@gmail.com>
2618 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
2620 * config/nds32/iterators.md: Add vector mode iterators and attributes.
2622 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
2624 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
2625 parameter list for vec_splats.
2627 2017-11-16 Joseph Myers <joseph@codesourcery.com>
2629 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
2631 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
2633 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
2636 * doc/install.texi: Document incompatibility of Perl >=5.6.26
2637 with the required version of automake 1.11.6.
2639 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
2641 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
2643 (DU_C2_3_power9): Correct reservation combinations.
2644 (FP_DIV_power9, VEC_DIV_power9): New.
2645 (power9-alu): Split out rotate/shift...
2646 (power9-rot): ...to here, correct dispatch resource.
2647 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
2649 (power9-fp): Correct latency.
2650 (power9-sdiv): Add div/sqrt resource.
2651 (power9-ddiv): Correct latency, add div/sqrt resource.
2652 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
2653 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
2655 (power9-qpdiv, power9-qpmul): Adjust resource usage.
2657 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2659 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
2660 switch statement mapping KF built-ins to TF built-ins if we don't
2661 have the proper ISA 3.0 assembler support.
2663 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2665 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
2666 (lower_emutls_function_body): Do not compute it.
2668 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2670 * ipa-split.c (split_bb_info): Turn time to sreal.
2671 (split_point): Likewise.
2672 (dump_split_point): Likewise.
2673 (fine_split_points): Likewise.
2674 (execute_split_functions): Only zero split_bbs; turn time to sreals.
2676 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2678 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
2681 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2683 * predict.c (combine_predictions_for_bb): Preserve zero predicted
2685 (expensive_function_p): Remove useless assert.
2686 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
2688 2017-11-16 Martin Sebor <msebor@redhat.com>
2690 PR tree-optimization/82588
2691 PR tree-optimization/82583
2692 * tree-vrp.c (check_array_ref): Handle flexible array members,
2693 string literals, and inner indices.
2694 (search_for_addr_array): Add detail to diagnostics.
2696 2017-11-16 Nathan Sidwell <nathan@acm.org>
2700 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
2701 (SET_DECL_ASSEMBLER_NAME): Forward to
2702 overwrite_decl_assembler_name.
2703 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
2704 (overwrite_decl_assembler_name): Declare.
2705 * tree.c (overwrite_decl_assembler_name): New.
2706 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
2707 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
2708 (LANG_HOOKS_INITIALIZER): Add it.
2709 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
2710 * langhooks.c (lhd_set_decl_assembler_name): Use
2711 SET_DECL_ASSEMBLER_NAME.
2712 (lhd_overwrite_decl_assembler_name): Default implementation.
2714 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2715 Jackson Woodruff <jackson.woodruff@arm.com>
2717 PR tree-optimization/71026
2718 * match.pd: Canonicalize constant multiplies in division.
2720 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2722 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
2723 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
2724 Remove OPT_fomit_frame_pointer entry.
2725 * common/config/arc/arc-common.c: Likewise.
2726 * common/config/arm/arm-common.c: Likewise.
2727 * common/config/avr/avr-common.c: Likewise.
2728 * common/config/c6x/c6x-common.c: Likewise.
2729 * common/config/cr16/cr16-common.c: Likewise.
2730 * common/config/cris/cris-common.c: Likewise.
2731 * common/config/epiphany/epiphany-common.c: Likewise.
2732 * common/config/fr30/fr30-common.c: Likewise.
2733 * common/config/frv/frv-common.c: Likewise.
2734 * common/config/ia64/ia64-common.c: Likewise.
2735 * common/config/iq2000/iq2000-common.c: Likewise.
2736 * common/config/lm32/lm32-common.c: Likewise.
2737 * common/config/m32r/m32r-common.c: Likewise.
2738 * common/config/mcore/mcore-common.c: Likewise.
2739 * common/config/microblaze/microblaze-common.c: Likewise.
2740 * common/config/mips/mips-common.c: Likewise.
2741 * common/config/mmix/mmix-common.c: Likewise.
2742 * common/config/mn10300/mn10300-common.c: Likewise.
2743 * common/config/nios2/nios2-common.c: Likewise.
2744 * common/config/pa/pa-common.c: Likewise.
2745 * common/config/pdp11/pdp11-common.c: Likewise.
2746 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
2747 * common/config/riscv/riscv-common.c: Likewise.
2748 * common/config/rs6000/rs6000-common.c: Likewise.
2749 * common/config/rx/rx-common.c: Likewise.
2750 * common/config/s390/s390-common.c: Likewise.
2751 * common/config/sh/sh-common.c: Likewise.
2752 * common/config/sparc/sparc-common.c: Likewise.
2753 * common/config/tilegx/tilegx-common.c: Likewise.
2754 * common/config/tilepro/tilepro-common.c: Likewise.
2755 * common/config/v850/v850-common.c: Likewise.
2756 * common/config/visium/visium-common.c: Likewise.
2757 * common/config/xstormy16/xstormy16-common.c: Likewise.
2758 * common/config/xtensa/xtensa-common.c: Likewise.
2759 * invoke.texi (-fomit-frame-pointer): Update documentation.
2761 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2763 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
2765 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2767 * cfg.c (scale_bbs_frequencies_int,
2768 cale_bbs_frequencies_gcov_type): Remove.
2769 * cfg.h (scale_bbs_frequencies_int,
2770 cale_bbs_frequencies_gcov_type): Remove.
2772 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2774 * tree-ssa-loop-manip.c
2775 (scale_dominated_blocks_in_loop): Update to profile counts.
2776 (tree_transform_and_unroll_loop): Likewise.
2778 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2780 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
2781 scale_bbs_frequencies_int.
2783 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2785 * final.c (compute_alignments): Use counts rather than frequencies.
2787 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2789 * cfgloopanal.c: Include sreal.h
2790 (average_num_loop_insns): Use counts and sreal for accounting.
2792 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2794 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
2797 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2798 Kito Cheng <kito.cheng@gmail.com>
2800 * config/nds32/constraints.md: Provide more constraints.
2801 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
2802 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
2803 support constraints usage.
2805 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2807 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
2809 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2810 Kito Cheng <kito.cheng@gmail.com>
2812 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
2813 * config/nds32/nds32.opt: Refine the layout.
2814 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
2815 TARGET_EXT_STRING): Support new options.
2816 * config/nds32/nds32.h: Likewise.
2817 * config/nds32/nds32.md: Likewise.
2818 * config/nds32/nds32-predicates.c: Likewise.
2819 * config/nds32/constraints.md: Likewise.
2820 * common/config/nds32/nds32-common.c: Likewise.
2822 2017-11-16 Julia Koval <julia.koval@intel.com>
2825 * config/i386/gfniintrin.h: Add sse check.
2826 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
2828 2017-11-16 Julia Koval <julia.koval@intel.com>
2830 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
2831 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
2832 (ix86_handle_option): Handle -mavx512vbmi2.
2833 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
2834 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2835 * config/i386/i386-c.c (__AVX512VBMI2__): New.
2836 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
2837 (ix86_valid_target_attribute_inner_p): Ditto.
2838 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
2839 * config/i386/i386.opt (mavx512vbmi2): New option.
2840 * doc/invoke.texi: Add new option.
2842 2017-11-16 Julia Koval <julia.koval@intel.com>
2844 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
2845 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
2846 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
2847 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
2848 _mm512_gf2p8mul_epi8): New intrinsics.
2849 * config/i386/i386-builtin-types.def
2850 (V64QI_FTYPE_V64QI_V64QI): New type.
2851 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
2852 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
2853 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
2854 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
2855 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
2856 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2858 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
2860 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
2861 explicitly as a stream of bytes.
2863 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2865 * config/rs6000/altivec.h (vec_xst_be): New #define.
2866 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
2867 and externalize from *altivec_vperm_<mode>_internal.
2868 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
2870 (XL_BE_V8HI): Likewise.
2871 (XL_BE_V4SI): Likewise.
2872 (XL_BE_V4SI): Likewise.
2873 (XL_BE_V2DI): Likewise.
2874 (XL_BE_V4SF): Likewise.
2875 (XL_BE_V2DF): Likewise.
2876 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
2877 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
2878 all array entries with these keys: VSX_BUILTIN_VEC_XL,
2879 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
2880 VSX_BUILTIN_VEC_XST_BE.
2881 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
2882 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
2884 (altivec_init_builtins): Replace conditional calls to def_builtin
2885 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
2886 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
2887 with unconditional calls. Remove calls to def_builtin for
2888 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
2889 __builtin_vec_xst_be.
2890 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
2891 to define_expand, and add alternate RTL generation for P8.
2892 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
2893 vsx_ld_elemrev_v8hi.
2894 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
2895 add alternate RTL generation for P8.
2896 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
2897 vsx_ld_elemrev_v16qi.
2898 (vsx_st_elemrev_v8hi): Convert define_insn
2899 to define_expand, and add alternate RTL generation for P8.
2900 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
2901 vsx_st_elemrev_v8hi.
2902 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
2903 add alternate RTL generation for P8.
2904 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
2905 vsx_st_elemrev_v16qi.
2907 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
2910 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
2911 TARGET_AVX512ER check.
2912 (ix86_option_override_internal): Set MASK_VZEROUPPER if
2913 neither -mzeroupper nor -mno-zeroupper is used and
2914 TARGET_EMIT_VZEROUPPER is set.
2915 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
2916 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
2918 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2920 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2921 folding of vector compares.
2922 (fold_build_vec_cmp): New helper function.
2923 (fold_compare_helper): New helper function.
2924 (builtin_function_type): Add compare builtins to the list of functions
2925 having unsigned arguments. Cosmetic updates to comment indentation.
2926 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
2927 the not+eq combination.
2929 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2931 PR tree-optimization/82726
2932 PR tree-optimization/70754
2933 * tree-predcom.c (order_drefs_by_pos): New function.
2934 (combine_chains): Move code setting has_max_use_after to...
2935 (try_combine_chains): ...here. New parameter. Sort combined chains
2936 according to position information.
2937 (tree_predictive_commoning_loop): Update call to above function.
2938 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
2940 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2942 PR tree-optimization/82726
2944 2017-01-23 Bin Cheng <bin.cheng@arm.com>
2946 PR tree-optimization/70754
2947 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
2948 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
2949 combined stmt before it if not NULL.
2950 (combine_chains): Process refs reversely and compute dominance point
2954 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2956 PR tree-optimization/79663
2957 * tree-predcom.c (combine_chains): Process refs in reverse order
2958 only for ZERO length chains, and add explaining comment.
2960 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2962 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
2963 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
2964 * doc/invoke.texi (ARM Options): Add armv8.3-a.
2966 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2968 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
2970 2017-11-15 Martin Liska <mliska@suse.cz>
2972 * tree-cfg.c (pass_warn_function_return::execute):
2973 Compare warn_return_type for greater than zero.
2975 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
2979 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
2980 to return true on Xeon and not on Xeon Phi.
2981 (ix86_check_avx256_register): Changed to ...
2982 (ix86_check_avx_upper_register): ... this. Add extra check for
2983 VALID_AVX512F_REG_OR_XI_MODE.
2984 (ix86_avx_u128_mode_needed): Changed
2985 ix86_check_avx256_register to ix86_check_avx_upper_register.
2986 (ix86_check_avx256_stores): Changed to ...
2987 (ix86_check_avx_upper_stores): ... this. Changed
2988 ix86_check_avx256_register to ix86_check_avx_upper_register.
2989 (ix86_avx_u128_mode_after): Changed
2990 avx_reg256_found to avx_upper_reg_found. Changed
2991 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
2992 (ix86_avx_u128_mode_entry): Changed
2993 ix86_check_avx256_register to ix86_check_avx_upper_register.
2994 (ix86_avx_u128_mode_exit): Ditto.
2995 * config/i386/i386.h: (host_detect_local_cpu): New define.
2997 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2999 * config/arm/xgene1.md (xgene1): Split into automatons
3000 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3001 (xgene1_f_load): Adjust reservations and/or types.
3002 (xgene1_f_store): Likewise.
3003 (xgene1_load_pair): Likewise.
3004 (xgene1_store_pair): Likewise.
3005 (xgene1_fp_load1): Likewise.
3006 (xgene1_load1): Likewise.
3007 (xgene1_store1): Likewise.
3008 (xgene1_move): Likewise.
3009 (xgene1_alu): Likewise.
3010 (xgene1_simd): Likewise.
3011 (xgene1_bfm): Likewise.
3012 (xgene1_neon_load1): Likewise.
3013 (xgene1_neon_store1): Likewise.
3014 (xgene1_neon_logic): Likewise.
3015 (xgene1_neon_st1): Likewise.
3016 (xgene1_neon_ld1r): Likewise.
3017 (xgene1_alu_cond): Added.
3018 (xgene1_shift_reg): Likwise.
3019 (xgene1_bfx): Likewise.
3020 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3022 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3025 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3027 (can_widen_mult_without_libcall): New function.
3028 (expand_mul_overflow): If only checking unsigned mul overflow,
3029 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3030 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3031 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3032 (expand_DIVMOD): Formatting fix.
3033 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3034 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3035 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3037 PR tree-optimization/82977
3038 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3039 constructed temporary to strlen_to_stridx.put.
3041 2017-11-15 Martin Liska <mliska@suse.cz>
3043 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3044 * configure: Regenerate.
3046 2017-11-15 Martin Liska <mliska@suse.cz>
3049 * config/sh/sh-mem.cc: Use proper probability for
3052 2017-11-14 Jeff Law <law@redhat.com>
3054 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3055 the red zone for stack_clash_protection_final_dynamic_probe targets
3056 when the total dynamic stack size is zero bytes.
3058 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3059 blocks is post order.
3061 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3063 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3064 * final.c (rest_of_clean_state): Set it for the
3065 -fcompare-debug dump.
3066 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3067 class when TDF_COMPARE_DEBUG is set.
3069 * dwarf2out.c (gen_producer_string): Discard
3072 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3075 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3076 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3077 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3078 (__TGMATH_REAL_2_3): Remove macros.
3079 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3080 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3082 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3084 (remquo): Define using __TGMATH_REAL_3.
3086 2017-11-14 Jeff Law <law@redhat.com>
3088 * vr-values.c: New file with contents extracted from tree-vrp.c.
3089 * Makefile.in (OBJS): Add vr-values.o
3090 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3091 (set_value_range, set_and_canonicalize_value_range): Likewise.
3092 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3093 (value_range_constant_singleton, symbolic_range_p): Likewise.
3094 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3095 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3096 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3097 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3098 (find_case_label_range, find_case_label_index): Likewise.
3099 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3100 (range_int_cst_singleton_p, value_inside_range): Likewise.
3101 (get_single_symbol): Likewise.
3102 (switch_update): Move structure definition here.
3103 (to_remove_edges, to_update_switch_stmts): Provide externs.
3104 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3105 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3106 (vrp_val_is_min, set_value_range): Likewise.
3107 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3108 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3109 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3110 (range_is_nonnull, range_int_cst_p): Likewwise.
3111 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3112 (get_single_symbol, operand_less_p): Likewise
3113 (compare_values_warnv, compare_values): Likewise.
3114 (value_inside_range, value_range_constant_singleton): Likewise.
3115 (zero_nonzero_bitgs_from_vr): Likewise.
3116 (extract_range_from_binary_expr_1): Likewise.
3117 (overflow_comparison_p): Likewise.
3118 (to_remove_edges, to_update_switch_stmts): Likewise.
3119 (find_case_label-index, find_case_label_range): Likewise.
3120 (switch_update, set_value_range_to_nonnegative): Remove.
3121 (set_value_range_to_truthvalue): Likewise.
3122 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3123 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3124 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3125 (find_case_label_ranges, test_for_singularity): Likewise.
3126 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3127 (x_vr_values): Move to its remaining use site.
3129 2017-11-10 Jeff Law <law@redhat.com>
3131 * vr-values.h (VR_INITIALIZER): Move #define here.
3132 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3133 * Makefile.in (OBJS): Add tree-evrp.o
3134 * tree-vrp.h (assert_info): Move structure definition here.
3135 (set_value_range_to_varying): Prototype.
3136 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3137 (infer_value_range, register_edge_assert_for): Likewise.
3138 (stmt_interesting_for_vrp): Likewise.
3139 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3140 (set_value_range_to_varying): No longer static.
3141 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3142 (infer_value_range, register_edge_assert_for): Likewise.
3144 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3146 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3147 generate the XXBRD instruction.
3149 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3150 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3151 as being compatible if -mabi=ieeelongdouble.
3152 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3153 to setup float128 built-ins with hardware support.
3154 (BU_FLOAT128_HW_2): Likewise.
3155 (BU_FLOAT128_HW_3): Likewise.
3156 (BU_FLOAT128_HW_VSX_1): Likewise.
3157 (BU_FLOAT128_HW_VSX_2): Likewise.
3158 (scalar_extract_expq): Change float128 built-in functions to
3159 accommodate having both KFmode and TFmode functions. Use the
3160 KFmode variant as the default.
3161 (scalar_extract_sigq): Likewise.
3162 (scalar_test_neg_qp): Likewise.
3163 (scalar_insert_exp_q): Likewise.
3164 (scalar_insert_exp_qp): Likewise.
3165 (scalar_test_data_class_qp): Likewise.
3166 (sqrtf128_round_to_odd): Delete processing the round to odd
3167 built-in functions as special built-in functions, and define them
3168 as float128 built-ins. Use the KFmode variant as the default.
3169 (truncf128_round_to_odd): Likewise.
3170 (addf128_round_to_odd): Likewise.
3171 (subf128_round_to_odd): Likewise.
3172 (mulf128_round_to_odd): Likewise.
3173 (divf128_round_to_odd): Likewise.
3174 (fmaf128_round_to_odd): Likewise.
3175 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3176 support for KFmode and TFmode xststdcqp calls.
3177 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3178 point, switch the built-in handlers for the get/set float128
3179 exponent, get float128 mantissa, float128 test built-ins, and the
3180 float128 round to odd built-in functions. Eliminate creating the
3181 float128 round to odd built-in functions as special built-ins.
3182 (rs6000_init_builtins): Eliminate special creation of the float128
3183 round to odd built-in functions.
3184 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3185 function insns to support both TFmode and KFmode variants.
3186 (xsxsigqp_<mode>): Likewise.
3187 (xsiexpqpf_<mode>): Likewise.
3188 (xsiexpqp_<mode>): Likewise.
3189 (xststdcqp_<mode>): Likewise.
3190 (xststdcnegqp_<mode>): Likewise.
3191 (xststdcqp_<mode>): Likewise.
3193 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3195 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3196 unused path_in_freq_ptr parameter.
3197 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3199 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3201 * ipa-inline.c (edge_badness): Dump sreal frequency.
3202 (compute_inlined_call_time): Match natural implementaiton ...
3203 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3204 forgotten division by CGRAPH_FREQ_BASE.
3206 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3208 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3209 Solaris 11. Update comment.
3210 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
3212 * config/sol2.h (STARTFILE_SPEC): Likewise.
3213 * configure: Regenerate.
3215 2017-11-14 Carl Love <cel@us.ibm.com>
3217 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
3218 le_ and be_ prefixes to swap* variables. Remove
3219 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
3222 2017-11-14 Jason Merrill <jason@redhat.com>
3224 Support GTY((cache)) on hash_map.
3225 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
3226 (ggc_cache_remove): Override it instead of ggc_mx.
3227 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
3228 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
3229 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
3230 (simple_cache_map_traits): Override maybe_mx.
3231 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
3232 (hash_map): Friend gt_cleare_cache.
3233 (gt_cleare_cache): New.
3234 * tree.h (tree_cache_traits): New hash_map traits class.
3235 (tree_cache_map): New typedef.
3237 2017-11-14 Richard Biener <rguenther@suse.de>
3239 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
3240 paramter and handling.
3241 (cleanup_control_flow_bb): Likewise.
3242 (cleanup_control_flow_pre): New helper performing a DFS walk
3243 to call cleanup_control_flow_bb in PRE order.
3244 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
3245 via cleanup_control_flow_pre.
3247 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
3249 * config/aarch64/aarch64-simd.md
3250 (aarch64_simd_bsl<mode>_internal): Remove DImode.
3251 (*aarch64_simd_bsl<mode>_alt): Likewise.
3252 (aarch64_simd_bsldi_internal): New.
3253 (aarch64_simd_bsldi_alt): Likewise.
3255 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3257 * tracer.c (better_p): Do not compare frequencies.
3258 * reg-stack.c (better_edge): Likewise.
3259 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
3262 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3264 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
3265 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
3266 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
3267 * lto-streamer-in.c (input_function): Use update_max_bb_count.
3268 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
3269 * predict.c (maybe_hot_frequency_p): Inline to ...
3270 (maybe_hot_count_p): ... here; rewrite to counts.
3271 (counts_to_freqs): Rename to ...
3272 (update_max_bb_count): ... this one.
3273 (expensive_function_p): Use counts.
3274 (estimate_bb_frequencies): Update.
3275 (rebuild_frequencies): Update.
3276 * predict.h (counts_to_freqs): Rename to ...
3277 (update_max_bb_count): ... this one.
3278 * profile.c (compute_branch_probabilities): Add debug info
3279 * tree-inline.c (expand_call_inline): Update debug info.
3280 (optimize_inline_calls): Use update_max_bb_count..
3281 (tree_function_versioning): Use update_max_bb_count..
3282 * value-prof.c (gimple_value_profile_transformations):
3283 Do not use update_max_bb_count.
3285 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3287 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
3288 always use frequencies.
3290 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3292 * bb-reorder.c: Remove frequencies from comments.
3293 (better_edge_p): Use profile counts.
3294 (find_traces): Dump profile counts.
3295 (rotate_loop): Use profile counts.
3296 (find_traces_1_round): Likewise.
3297 (connect_better_edge_p): Use counts instead of probabilities for
3299 (copy_bb_p): Drop early check for non-0 frequency.
3300 (sanitize_hot_paths): Update comments.
3302 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3304 * ipa-split.c (struct split_point): Add count.
3305 (consider_split): Do not compute incoming frequency; compute incoming
3306 count and store it to split_point.
3307 (split_function): Set count of the call to split part correctly.
3309 2017-11-13 Carl Love <cel@us.ibm.com>
3311 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
3313 2017-11-13 Tom Tromey <tom@tromey.com>
3315 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
3317 2017-11-13 Carl Love <cel@us.ibm.com>
3319 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3320 Add support for builtins:
3321 unsigned int vec_first_{,miss}_match_{,or_eos}index,
3322 vector {un,}signed {char,int,short},
3323 vector {un,}signed {char,int,short}) arguments.
3324 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
3325 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
3326 Add BU_P9V_AV_2 expansions for the builtins.
3327 * config/rs6000/altivec.h (vec_first_match_index,
3328 vec_first_mismatch_index, vec_first_match_or_eos_index,
3329 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
3330 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
3331 new extern declaration.
3332 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
3333 * config/rs6000/vsx.md (first_match_index_<mode>,
3334 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
3335 first_mismatch_or_eos_index_<mode>): Add define expand.
3336 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
3337 * doc/extend.texi: Update the built-in documenation file for the new
3340 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
3342 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
3343 into the min/max operations for _Float<N> and _Float<N>X types.
3345 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3348 * dwarf2out.c (do_eh_frame): New static variable.
3349 (dwarf2out_begin_prologue): Set it.
3350 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
3352 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3354 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
3356 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
3358 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3361 * debug.h (dwarf2out_do_eh_frame): Declare.
3362 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
3363 (dwarf2out_do_frame): Use it.
3364 (dwarf2out_do_cfi_asm): Likewise.
3365 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
3366 (dwarf2out_assembly_start): Likewise.
3367 (dwarf2out_begin_prologue): Fix comment.
3368 * toplev.c (compile_file): Always call dwarf2out_frame_finish
3369 if the target needs either debug or unwind DWARF2 info.
3370 * lto-opts.c (lto_write_options): Do not save -fexceptions,
3371 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
3372 -ftrapping-math, -ftrapv and -fwrapv.
3374 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3376 * cgraph.c (cgraph_edge::sreal_frequency): New function.
3377 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
3378 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
3379 (estimate_edge_size_and_time): Likewise.
3380 (ipa_merge_fn_summary_after_inlining): Likewise.
3381 * ipa-inline.c (cgraph_freq_base_rec): Remove.
3382 (compute_uninlined_call_time): Use sreal_frequency.
3383 (compute_inlined_call_time): Likewise.
3384 (ipa_inline): Do not initialize cgraph_freq_base_rec.
3385 * profile-count.c: Include sreal.h.
3386 (profile_count::to_sreal_scale): New.
3387 * profile-count.h: Forward declare sreal.
3388 (profile_count::to_sreal_scale): Declare.
3390 2017-11-13 Nathan Sidwell <nathan@acm.org>
3392 * diagnostic.c (maybe_line_and_column): New.
3393 (diagnostic_get_location_text): Use it.
3394 (diagnostic_report_current_module): Likewise.
3395 (test_diagnostic_get_location_text): Add tests.
3397 2017-11-13 Luis Machado <luis.machado@linaro.org>
3399 * doc/md.texi (Specifying processor pipeline description): Fix
3400 incorrect latency for the div instruction example.
3402 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3404 PR tree-optimization/78821
3405 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
3406 that bit_not_p is the same.
3407 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
3408 (split_group): Count precisely bit_not_p bits in each statement.
3409 (invert_op): New function.
3410 (imm_store_chain_info::output_merged_store): Use invert_op to
3411 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
3412 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
3414 2017-11-13 Martin Liska <mliska@suse.cz>
3416 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
3417 (struct source_info): Likewise.
3418 (add_branch_counts): Likewise.
3419 (add_line_counts): Likewise.
3420 (function_summary): Likewise.
3421 (output_intermediate_line): Likewise.
3422 (generate_results): Likewise.
3424 2017-11-13 Martin Liska <mliska@suse.cz>
3426 * gcov.c (struct block_info): Remove typedef for block_t.
3427 (struct line_info): Likewise.
3428 (line_info::has_block): Likewise.
3429 (EXIT_BLOCK): Likewise.
3430 (unblock): Likewise.
3431 (circuit): Likewise.
3432 (get_cycles_count): Likewise.
3433 (process_file): Likewise.
3434 (read_graph_file): Likewise.
3435 (solve_flow_graph): Likewise.
3436 (find_exception_blocks): Likewise.
3437 (add_line_counts): Likewise.
3438 (accumulate_line_info): Likewise.
3439 (output_line_details): Likewise.
3441 2017-11-13 Martin Liska <mliska@suse.cz>
3443 * gcov.c (struct arc_info): Remove typedef for arc_t.
3444 (struct line_info): Likewise.
3445 (add_branch_counts): Likewise.
3446 (output_branch_count): Likewise.
3447 (function_info::~function_info): Likewise.
3448 (circuit): Likewise.
3449 (output_intermediate_line): Likewise.
3450 (read_graph_file): Likewise.
3451 (solve_flow_graph): Likewise.
3452 (find_exception_blocks): Likewise.
3453 (add_line_counts): Likewise.
3454 (accumulate_line_info): Likewise.
3455 (output_line_details): Likewise.
3456 (output_function_details): Likewise.
3458 2017-11-13 Martin Liska <mliska@suse.cz>
3460 * gcov.c (struct function_info): Remove typedef for function_t.
3461 (struct source_info): Likewise.
3462 (source_info::get_functions_at_location): Likewise.
3463 (solve_flow_graph): Likewise.
3464 (find_exception_blocks): Likewise.
3465 (add_line_counts): Likewise.
3466 (output_intermediate_file): Likewise.
3467 (process_file): Likewise.
3468 (generate_results): Likewise.
3469 (release_structures): Likewise.
3470 (read_graph_file): Likewise.
3471 (read_count_file): Likewise.
3472 (accumulate_line_counts): Likewise.
3473 (output_lines): Likewise.
3475 2017-11-13 Martin Liska <mliska@suse.cz>
3477 * gcov.c (function_info::function_info): Remove num_counts
3478 and add vector<gcov_type>.
3479 (function_info::~function_info): Use the vector.
3480 (process_file): Likewise.
3481 (read_graph_file): Likewise.
3482 (read_count_file): Likewise.
3483 (solve_flow_graph): Likewise.
3485 2017-11-13 Martin Liska <mliska@suse.cz>
3487 * gcov.c (function_info::is_artificial): New function.
3488 (process_file): Erase all artificial early.
3489 (generate_results): Skip as all artificial are already
3492 2017-11-13 Martin Liska <mliska@suse.cz>
3494 * gcov.c (read_graph_file): Store to global vector of functions.
3495 (read_count_file): Iterate the vector.
3496 (process_file): Likewise.
3497 (generate_results): Likewise.
3498 (release_structures): Likewise.
3500 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3502 PR tree-optimization/82954
3503 * gimple-ssa-store-merging.c
3504 (imm_store_chain_info::coalesce_immediate_stores): If
3505 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
3507 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
3509 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
3510 Upddate call to ENDIAN_LANE_N.
3511 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
3512 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
3513 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
3514 and use aarch64_endian_lane_rtx.
3515 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
3517 2017-11-12 Tom de Vries <tom@codesourcery.com>
3519 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
3521 2017-11-12 Tom de Vries <tom@codesourcery.com>
3523 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
3525 2017-11-12 Tom de Vries <tom@codesourcery.com>
3527 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
3529 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3530 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3531 * defaults.h (ASM_OUTPUT_LABELREF): Same.
3533 2017-11-11 Martin Sebor <msebor@redhat.com>
3536 * doc/extend.texi (attribute nonstring): Remove spurious argument.
3539 * prefic.c (translate_name): Replace strncpy with memcpy to
3540 avoid -Wstringop-truncation.
3542 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3544 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
3546 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3548 * predict.c (maybe_hot_frequency_p): Do not use cfun.
3550 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3552 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
3555 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
3557 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
3558 enable generating XXBRH if the value is in a vector register.
3559 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
3560 value is in a vector register.
3561 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
3562 register to register bswap64's instead of doing the GPR sequence
3563 used on previous machines.
3564 (bswapdi2_xxbrd): New insn.
3565 (bswapdi2_reg): Disallow on ISA 3.0.
3566 (register to register bswap64 splitter): Do not split the insn on
3567 ISA 3.0 systems that use XXBRD.
3569 2017-11-10 Martin Sebor <msebor@redhat.com>
3572 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
3574 (find_subframework_file): Same.
3576 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3578 * auto-profile.c (afdo_indirect_call): Drop frequency.
3579 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
3580 (cgraph_node::create_edge): Drop frequency argument.
3581 (cgraph_node::create_indirect_edge): Drop frequency argument.
3582 (cgraph_edge::make_speculative): Drop frequency arguments.
3583 (cgraph_edge::resolve_speculation): Do not update frequencies
3584 (cgraph_edge::dump_edge_flags): Do not dump frequency.
3585 (cgraph_node::dump): Check consistency in IPA mode.
3586 (cgraph_edge::maybe_hot_p): Use IPA counter.
3587 (cgraph_edge::verify_count_and_frequency): Rename to ...
3588 (cgraph_edge::verify_count): ... this one; drop frequency checking.
3589 (cgraph_node::verify_node): Update.
3590 * cgraph.h (struct cgraph_edge): Drop frequency.
3591 (cgraph_edge::frequency): New function.
3592 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
3594 (cgraph_edge::rebuild_edges): Likewise.
3595 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
3596 (duplicate_thunk_for_node): Do not pass frequency.
3597 (cgraph_node::create_clone): Scale only counts.
3598 (cgraph_node::create_virtual_clone): Do not pass frequency.
3599 (cgraph_node::create_edge_including_clones): Do not pass frequency.
3600 (cgraph_node::create_version_clone): Do not pass frequency.
3601 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
3602 (cgraph_node::expand_thunk): Do not pass frequency.
3603 (cgraph_node::create_wrapper): Do not pass frequency.
3604 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
3606 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
3607 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
3608 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
3609 (gather_caller_stats): Use frequency function.
3610 (ipcp_cloning_candidate_p): Use frequency function.
3611 (ipcp_propagate_stage): Use frequency function.
3612 (get_info_about_necessary_edges): Use frequency function.
3613 (update_profiling_info): Update only IPA profile.
3614 (update_specialized_profile): Use frequency functoin.
3615 (perhaps_add_new_callers): Update only IPA profile.
3616 * ipa-devirt.c (ipa_devirt): Use IPA profile.
3617 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
3618 (dump_ipa_call_summary): Use frequency function.
3619 (estimate_edge_size_and_time): Use frequency function.
3620 (ipa_merge_fn_summary_after_inlining): Use frequency function.
3621 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
3622 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
3623 (update_noncloned_counts): ... ths one; scale counts only.
3624 (clone_inlined_nodes): Do not scale frequency.
3625 (inline_call): Do not pass frequency.
3626 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
3627 (compute_inlined_call_time): Use IPA profile.
3628 (want_inline_small_function_p): Use IPA profile.
3629 (want_inline_self_recursive_call_p): Use IPA profile.
3630 (edge_badness): Use IPA profile.
3631 (lookup_recursive_calls): Use IPA profile.
3632 (recursive_inlining): Do not pass frequency.
3633 (resolve_noninline_speculation): Do not update frequency.
3634 (inline_small_functions): Collect max of IPA profile.
3635 (dump_overall_stats): Dump IPA porfile.
3636 (dump_inline_stats): Dump IPA porfile.
3637 (ipa_inline): Collect IPA stats.
3638 * ipa-inline.h (clone_inlined_nodes): Update prototype.
3639 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
3640 (ipa_propagate_frequency): Use frequency function.
3641 (ipa_profile): Cleanup.
3642 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
3643 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
3644 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
3645 (input_node): Do not stream frequency.
3646 (input_edge): Do not stream frequency.
3647 (merge_profile_summaries): Scale only IPA profiles.
3648 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
3649 * predict.c (drop_profile): Do not recompute frequency.
3650 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
3651 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
3652 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
3653 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
3654 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
3655 * tree-inline.c (copy_bb): Do not scale frequency.
3656 (expand_call_inline): Do not scale frequency.
3657 (tree_function_versioning): Do not scale frequency.
3658 * ubsan.c (ubsan_create_edge): Do not pass frequency.
3660 2017-11-10 Julia Koval <julia.koval@intel.com>
3662 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
3663 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
3664 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
3665 (_mm256_maskz_gf2p8affine_epi64_epi8)
3666 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
3667 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
3668 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
3669 (__builtin_ia32_vgf2p8affineqb_v32qi)
3670 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
3671 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
3673 2017-11-10 Tamar Christina <tamar.christina@arm.com>
3677 (arm_option_override): Refactor.
3678 (arm_option_reconfigure_globals): New.
3679 (arm_options_perform_arch_sanity_checks): New.
3680 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
3682 (arm_options_perform_arch_sanity_checks): Likewise
3684 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
3686 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
3687 (power9-qpmul): New.
3688 * rs6000/rs6000.md ("type" attr): Add qmul.
3689 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
3690 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
3691 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
3693 2017-11-10 Martin Sebor <msebor@redhat.com>
3696 * builtins.c (compute_objsize): Handle arrays that
3697 compute_builtin_object_size likes to fail for. Make extern.
3698 * builtins.h (compute_objsize): Declare.
3699 (check_strncpy_sizes): New function.
3700 (expand_builtin_strncpy): Call check_strncpy_sizes.
3701 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
3702 -Wstringop-truncation.
3703 (gimple_fold_builtin_strncat): Same.
3704 * gimple.c (gimple_build_call_from_tree): Set call location.
3705 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
3706 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
3707 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
3708 (handle_builtin_strlen): Use strlen_to_stridx.
3709 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
3711 Use strlen_to_stridx.
3712 (pass_strlen::execute): Release strlen_to_stridx.
3713 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
3714 (-Wstringop-truncation): Document new option.
3716 2017-11-10 Martin Liska <mliska@suse.cz>
3718 PR gcov-profile/82702
3719 * gcov.c (main): Handle intermediate files in a different
3721 (get_gcov_intermediate_filename): New function.
3722 (output_gcov_file): Remove support of intermediate files.
3723 (generate_results): Allocate intermediate file.
3724 (release_structures): Clean-up properly fn_end.
3725 (output_intermediate_file): Start iterating with line 1.
3727 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3729 PR tree-optimization/82929
3730 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3731 ops_swapped_p non-static data member.
3732 (store_immediate_info::store_immediate_info): Clear it.
3733 (imm_store_chain_info::coalesce_immediate_stores): If swapping
3734 ops set ops_swapped_p.
3735 (count_multiple_uses): Handle ops_swapped_p.
3737 2017-11-10 Martin Liska <mliska@suse.cz>
3739 * coverage.c (coverage_init): Stream information about
3740 support of has_unexecuted_blocks.
3741 * doc/gcov.texi: Document that.
3742 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
3743 * gcov.c (read_graph_file): Likewise.
3744 (output_line_beginning): Fix a small issue with
3747 2017-11-10 Bin Cheng <bin.cheng@arm.com>
3749 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
3750 reference of trivial component.
3752 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3755 * gimple-ssa-store-merging.c
3756 (pass_store_merging::terminate_all_aliasing_chains): For
3757 gimple_store_p stmts also call refs_output_dependent_p.
3759 PR rtl-optimization/82913
3760 * compare-elim.c (try_merge_compare): Punt if def_insn is not
3763 2017-11-09 Jeff Law <law@redhat.com>
3765 * vr-values.h: New file with vr_values class.
3766 * tree-vrp.c: Include vr-values.h
3767 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
3768 data objects into the vr_values class.
3769 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
3770 (get_value_range): Make it a member function within vr_values class.
3771 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
3772 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
3773 (op_with_constant_singleton_value_range): Likewise.
3774 (extract_range_for_var_from_comparison_expr): Likewise.
3775 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
3776 (extract_range_from_binary_expr): Likewise.
3777 (extract_range_from_unary_expr): Likewise.
3778 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
3779 (check_for_binary_op_overflow, extract_range_basic): Likewise.
3780 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
3781 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
3782 (compare_name_with_value, compare_names): Likewise.
3783 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
3784 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
3785 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
3786 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
3787 (extract_range_from_phi_node): Likewise.
3788 (simplify_truth_ops_using_ranges): Likewise.
3789 (simplify_div_or_mod_using_ranges): Likewise.
3790 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
3791 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
3792 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
3793 (simplify_float_conversion_using_ranges): Likewise.
3794 (simplify_internal_call_using_ranges): Likewise.
3795 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
3796 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
3797 poitner via x_vr_values for calls into gimple folder.
3798 (vrp_initialize_lattice): Make this the vr_values ctor.
3799 (vrp_free_lattice): Make this the vr_values dtor.
3800 (set_vr_value): New function.
3801 (class vrp_prop): Add vr_values data member. Add various member
3802 functions as well as member functions that delegate to vr_values.
3803 (check_array_ref): Make a member function within vrp_prop class.
3804 (search_for_addr_array, vrp_initialize): Likewise.
3805 (vrp_finalize): Likewise. Revamp to avoid direct access to
3806 vr_value, values_propagated, etc.
3807 (check_array_bounds): Extract vrp_prop class instance pointer from
3808 walk info structure. Use it to call member functions.
3809 (check_all_array_refs): Make a member function within vrp_prop class.
3810 Smuggle class instance pointer via walk info structure.
3811 (x_vr_values): New local static.
3812 (vrp_valueize): Use x_vr_values to get class instance.
3813 (vr_valueize_1): Likewise.
3814 (class vrp_folder): Add vr_values data member. Add various member
3815 functions as well as member functions that delegate to vr_values.
3816 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
3817 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
3818 class instance from vr_values. Use it to call member functions.
3819 (vrp_dom_walker): Add vr_values data member.
3820 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
3821 instance via x_vr_values.
3822 (identify_jump_threads): Accept vr_values as argument. Store
3823 it into the walker structure.
3824 (evrp_dom_walker): Add vr_values class data member. Add various
3826 (evrp_dom_walker::try_find_new_range): Use vr_values data
3827 member to access the memory allocator.
3828 (evrp_dom_walker::before_dom_children): Store vr_values class
3829 instance into the vrp_folder class.
3830 (evrp_dom_walker::push_value_range): Rework to avoid direct
3831 access to num_vr_values and vr_value.
3832 (evrp_dom_walker::pop_value_range): Likewise.
3833 (execute_early_vrp): Remove call to vrp_initialize_lattice.
3834 Use vr_values to get to dump_all_value_ranges member function.
3835 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
3836 and simplify_cond_using_ranges_2 via vrp_prop class instance.
3837 Pass vr_values class instance down to identify_jump_threads.
3838 Remove call to vrp_free_lattice.
3839 (debug_all_value_ranges): Remove.
3841 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
3842 (vrp_folder): Likewise.
3844 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
3845 Get it from the existing bitmap instead.
3846 (vrp_intersect_ranges_1): Likewise.
3848 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3850 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3852 (store_immediate_info::store_immediate_info): Add bitnotp argument,
3853 set bit_not_p to it.
3854 (imm_store_chain_info::coalesce_immediate_stores): Break group
3855 if bit_not_p is different.
3856 (count_multiple_uses, split_group,
3857 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
3858 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
3859 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
3860 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
3861 to store_immediate_info ctor.
3863 2017-11-09 Jim Wilson <jimw@sifive.com>
3865 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
3866 (scan_prog_file): Likewise.
3868 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3870 * bb-reorder.c (max_entry_frequency): Remove.
3871 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
3872 connect_traces, push_to_next_round_p): Remove prototypes.
3873 (find_traces_1_round): Use counts only.
3874 (push_to_next_round_p): Likewise.
3875 (find_traces): Likewise.
3876 (rotate_loop): Likewise.
3877 (find_traces_1_round): Likewise.
3878 (connect_traces): Likewise.
3879 (edge_order): Likewise.
3881 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
3883 * config/arm/arm.c (output_return_instruction): Add comments to
3884 indicate requirement for cmse_nonsecure_entry return to account
3885 for the size of clearing instruction output here.
3886 (thumb_exit): Likewise.
3887 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
3888 return in hardfloat mode.
3890 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
3892 * config/rs6000/rs6000.c (machine_function): Add a bool,
3893 "toc_is_wrapped_separately".
3894 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
3895 if it wasn't explicitly set or unset, we are optimizing for speed, and
3896 doing separate shrink-wrapping.
3897 (rs6000_get_separate_components): Enable the TOC component if
3898 saving the TOC register in the prologue.
3899 (rs6000_components_for_bb): Handle the TOC component.
3900 (rs6000_emit_prologue_components): Store the TOC register where needed.
3901 (rs6000_set_handled_components): Mark TOC as handled, if handled.
3902 (rs6000_emit_prologue): Don't save the TOC if that is already done.
3904 2017-11-09 Martin Jambor <mjambor@suse.cz>
3906 * ipa-param-manipulation.c: New file.
3907 * ipa-param-manipulation.h: Likewise.
3908 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
3909 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
3910 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
3911 (ipa_parm_adjustment): Likewise.
3912 (ipa_parm_adjustment_vec): Likewise.
3913 (ipa_get_vector_of_formal_parms): Moved declaration to
3914 ipa-param-manipulation.h.
3915 (ipa_get_vector_of_formal_parm_types): Likewise.
3916 (ipa_modify_formal_parameters): Likewise.
3917 (ipa_modify_call_arguments): Likewise.
3918 (ipa_combine_adjustments): Likewise.
3919 (ipa_dump_param_adjustments): Likewise.
3920 (ipa_modify_expr): Likewise.
3921 (ipa_get_adjustment_candidate): Likewise.
3922 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
3923 ipa-param-manipulation.c.
3924 (ipa_get_vector_of_formal_parm_types): Likewise.
3925 (ipa_modify_formal_parameters): Likewise.
3926 (ipa_modify_call_arguments): Likewise.
3927 (ipa_modify_expr): Likewise.
3928 (get_ssa_base_param): Likewise.
3929 (ipa_get_adjustment_candidate): Likewise.
3930 (index_in_adjustments_multiple_times_p): Likewise.
3931 (ipa_combine_adjustments): Likewise.
3932 (ipa_dump_param_adjustments): Likewise.
3933 * tree-sra.c: Also include ipa-param-manipulation.h
3934 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
3937 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3938 Alan Hayward <alan.hayward@arm.com>
3939 David Sherwood <david.sherwood@arm.com>
3941 * doc/sourcebuild.texi (vect_masked_store): Document.
3943 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3944 Alan Hayward <alan.hayward@arm.com>
3945 David Sherwood <david.sherwood@arm.com>
3947 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
3949 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3950 Alan Hayward <alan.hayward@arm.com>
3951 David Sherwood <david.sherwood@arm.com>
3953 * doc/sourcebuild.texi (vect_variable_length): Document.
3955 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3956 Alan Hayward <alan.hayward@arm.com>
3957 David Sherwood <david.sherwood@arm.com>
3959 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
3961 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3962 Alan Hayward <alan.hayward@arm.com>
3963 David Sherwood <david.sherwood@arm.com>
3965 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
3967 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3968 Alan Hayward <alan.hayward@arm.com>
3969 David Sherwood <david.sherwood@arm.com>
3971 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
3972 previously undocumented selectors.
3973 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
3975 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3977 * doc/rtl.texi (const_vector): Say that elements can be
3978 const_wide_ints too.
3979 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
3980 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
3981 (gen_vec_duplicate): Use it instead of CONSTANT_P.
3982 * optabs.c (expand_vector_broadcast): Likewise.
3984 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3985 Alan Hayward <alan.hayward@arm.com>
3986 David Sherwood <david.sherwood@arm.com>
3988 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
3989 scaled index even if the unscaled address was invalid.
3990 Don't increase the complexity of using a scale in that case.
3992 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3993 Alan Hayward <alan.hayward@arm.com>
3994 David Sherwood <david.sherwood@arm.com>
3996 * doc/rtl.texi: Rewrite the subreg rules so that they partition
3997 the inner register into REGMODE_NATURAL_SIZE bytes rather than
3998 UNITS_PER_WORD bytes.
3999 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4000 based on REGMODE_NATURAL_SIZE of the inner mode.
4001 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4002 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4003 * expmed.c (lowpart_bit_field_p): Divide the value up into
4004 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4005 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4006 whether something is likely to occupy more than one register.
4008 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4011 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4012 profile_count::adjust_for_ipa_scaling.
4013 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4014 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4016 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4018 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4020 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4021 (split_group): Add total_orig and total_new arguments, estimate the
4022 number of statements related to the store group without store merging
4023 and with store merging.
4024 (imm_store_chain_info::output_merged_store): Adjust split_group
4025 callers, punt if estimated number of statements with store merging
4026 is not smaller than estimated number of statements without it.
4028 (handled_load): Remove has_single_use checks.
4029 (pass_store_merging::process_store): Likewise.
4031 2017-11-09 Richard Biener <rguenther@suse.de>
4033 PR tree-optimization/82902
4034 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4036 2017-11-09 Martin Liska <mliska@suse.cz>
4039 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4042 2017-11-09 Martin Liska <mliska@suse.cz>
4044 PR tree-optimization/82669
4045 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4047 2017-11-09 Martin Liska <mliska@suse.cz>
4049 PR gcov-profile/48463
4050 * coverage.c (coverage_begin_function): Output also end locus
4051 of a function and information whether the function is
4053 * gcov-dump.c (tag_function): Parse and print the information.
4054 * gcov.c (INCLUDE_MAP): Add include.
4055 (INCLUDE_SET): Likewise.
4056 (struct line_info): Move earlier in the source file because
4057 of vector<line_info> in function_info structure.
4058 (line_info::line_info): Likewise.
4059 (line_info::has_block): Likewise.
4060 (struct source_info): Add new member index.
4061 (source_info::get_functions_at_location): New function.
4062 (function_info::group_line_p): New function.
4063 (output_intermediate_line): New function.
4064 (output_intermediate_file): Use the mentioned function.
4065 (struct function_start): New.
4066 (struct function_start_pair_hash): Likewise.
4067 (process_file): Add code that identifies group functions.
4068 Assign lines either to global or function scope.
4069 (generate_results): Skip artificial functions.
4070 (find_source): Assign index for each source file.
4071 (read_graph_file): Read new flag artificial and end_line.
4072 (add_line_counts): Assign it either to global of function scope.
4073 (accumulate_line_counts): Isolate core of the function to
4074 accumulate_line_info and call it for both function and global
4076 (accumulate_line_info): New function.
4077 (output_line_beginning): Fix GNU coding style.
4078 (print_source_line): New function.
4079 (output_line_details): Likewise.
4080 (output_function_details): Likewise.
4081 (output_lines): Iterate both source (global) scope and function
4083 (struct function_line_start_cmp): New class.
4084 * doc/gcov.texi: Reflect changes in documentation.
4086 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4089 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4090 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4091 and similarly for not instead of neg.
4093 2017-11-08 Andi Kleen <ak@linux.intel.com>
4095 * config/i386/i386.opt: Add -mforce-indirect-call.
4096 * config/i386/predicates.md: Check for flag_force_indirect_call.
4097 * doc/invoke.texi: Document -mforce-indirect-call
4099 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4101 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4103 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4104 config/riscv/riscv.c (predict.h): New include.
4105 (riscv_slow_unaligned_access_p): No longer static.
4106 (riscv_block_move_straight): Add require.
4107 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4109 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4112 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4113 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4114 nonimmediate_operand predicate for operand 1 instead of
4117 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4119 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4121 * config/aarch64/constraints.md (Uml): New constraint.
4122 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4125 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4127 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4128 of two vec_duplicates into a vec_concat.
4130 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4132 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4133 Simplify vec_merge of vec_duplicate and vec_concat.
4134 * config/aarch64/constraints.md (Utq): New constraint.
4135 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4138 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4140 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4141 Simplify vec_merge of vec_duplicate and const_vector.
4142 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4144 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4145 mode iterator. Update predicate on operand 1 to
4146 handle non-const_vec constants. Delete constraints.
4147 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4149 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4151 PR tree-optimization/78821
4152 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4154 (store_operand_info::store_operand_info): Initialize it to false.
4155 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4156 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4157 store in the group, and if chain_info is non-NULL, to ignore altogether
4159 (compatible_load_p): Fail if bit_not_p does not match.
4160 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4161 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4162 (pass_store_merging::process_store): Adjust
4163 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4164 call terminate_all_aliasing_chains newly when adding a store into
4165 a chain with non-NULL chain_info.
4167 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4169 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4171 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4173 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4175 (aarch64_layout_frame): Initialise emit_frame_chain.
4176 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4177 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4179 2017-11-08 Martin Liska <mliska@suse.cz>
4181 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4182 of gimple_call_internal_p.
4184 2017-11-07 Tom de Vries <tom@codesourcery.com>
4186 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4188 2017-11-07 Tom de Vries <tom@codesourcery.com>
4190 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4193 2017-11-08 Martin Liska <mliska@suse.cz>
4196 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4198 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4200 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4202 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4205 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4206 (*cmp<mode>_ccz_1): New insn with $k alternative.
4209 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4210 RTX_COMM_COMPARE as commutative as well.
4211 (ix86_binary_operator_ok): Formatting fix.
4212 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
4213 *<code><mode>3<mask_name><round_saeonly_name>,
4214 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
4215 *<s>mul<mode>3_highpart<mask_name>,
4216 *vec_widen_umult_even_v16si<mask_name>,
4217 *vec_widen_umult_even_v8si<mask_name>,
4218 *vec_widen_umult_even_v4si<mask_name>,
4219 *vec_widen_smult_even_v16si<mask_name>,
4220 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
4221 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
4222 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
4223 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
4224 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
4225 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
4226 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
4227 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
4228 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
4229 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
4230 ix86_binary_operator_ok. Formatting fixes.
4231 (*<plusminus_insn><mode>3<mask_name><round_name>,
4232 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
4235 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4237 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
4238 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
4239 eq and ne if TARGET_ISEL.
4240 (cmp): New code_iterator.
4241 (UNS, UNSU_, UNSIK): New code_attrs.
4242 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
4243 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
4245 ("eq<mode>3"): ... this.
4246 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
4248 ("ne<mode>3"): ... this.
4250 2017-11-07 Julia Koval <julia.koval@intel.com>
4253 * common/config/i386/i386-common.c
4254 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
4255 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
4256 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
4257 * config/i386/i386.opt: Ditto.
4258 * config/i386/i386.c (ix86_target_string): Ditto.
4259 (ix86_option_override_internal): Ditto.
4260 (ix86_init_mpx_builtins): Move MPX to args2.
4261 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
4262 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4263 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4264 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4265 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4266 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4267 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
4269 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
4272 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
4273 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
4274 (zero-extendsidi peephole2): Remove peephole.
4276 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4279 * stor-layout.c: Include attribs.h.
4280 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
4281 explicit lookup of "aligned" attribute.
4283 2017-11-07 Andrew Waterman <andrew@sifive.com>
4285 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
4286 (riscv_expand_block_move): Likewise.
4287 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
4289 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
4290 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
4291 * config/riscv/riscv.c (riscv_block_move_straight): New function.
4292 (riscv_adjust_block_mem): Likewise.
4293 (riscv_block_move_loop): Likewise.
4294 (riscv_expand_block_move): Likewise.
4295 * config/riscv/riscv.md (movmemsi): New pattern.
4297 2017-11-07 Michael Clark <michaeljclark@mac.com>
4299 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
4300 (MUSL_DYNAMIC_LINKER): Likewise.
4302 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
4304 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
4305 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
4308 2017-11-07 Richard Biener <rguenther@suse.de>
4310 * match.pd: Fix build.
4312 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
4313 Jackson Woodruff <jackson.woodruff@arm.com>
4315 PR tree-optimization/71026
4316 * match.pd: Canonicalize negate in division.
4318 2017-11-07 Sudakshina Das <sudi.das@arm.com>
4321 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
4323 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4325 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
4326 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
4328 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4330 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
4331 non-scalar integral types.
4332 * match.pd (negate_expr_p): Handle MINUS_EXPR.
4333 (-(A-B), -(~A)): New transformations.
4335 2017-11-07 Tom de Vries <tom@codesourcery.com>
4337 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
4338 semicolon after "do {} while (0)".
4339 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4340 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4341 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4342 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4343 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4345 2017-11-07 Tom de Vries <tom@codesourcery.com>
4347 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
4348 after "do {} while (0)".
4349 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4350 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4351 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4352 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4353 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4355 2017-11-07 Tom de Vries <tom@codesourcery.com>
4358 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
4360 (arm_rtx_costs_internal): Add missing semicolon after
4361 HANDLE_NARROW_SHIFT_ARITH call.
4363 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4365 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4366 disable isel if it was not set explicitly.
4368 2017-11-06 James Bowman <james.bowman@ftdichip.com>
4370 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
4371 (add_type_attribute) likewise.
4373 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
4375 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
4376 of struct ix86_frame.
4377 (ix86_initial_elimination_offset): Likewise.
4378 (ix86_expand_split_stack_prologue): Likewise.
4380 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
4382 * tree-vrp.h (enum value_range_type): Update stale comment.
4384 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4385 Alan Hayward <alan.hayward@arm.com>
4386 David Sherwood <david.sherwood@arm.com>
4388 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
4389 (aarch64_expand_vec_perm_const): Take the number of units too.
4390 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
4391 (aarch64_expand_vec_perm_const): Likewise.
4392 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
4393 (vec_perm<mode>): Update accordingly.
4395 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4396 Alan Hayward <alan.hayward@arm.com>
4397 David Sherwood <david.sherwood@arm.com>
4399 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
4400 Take the number of units too.
4401 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
4402 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
4403 but check for a vector mode before rather than after the call.
4404 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
4405 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
4406 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
4407 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
4408 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
4409 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
4410 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
4411 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
4412 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
4413 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
4414 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
4415 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
4416 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
4417 (aarch64_sqdmull2_n<mode>): Update accordingly.
4419 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4420 Alan Hayward <alan.hayward@arm.com>
4421 David Sherwood <david.sherwood@arm.com>
4423 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
4424 the number of units too.
4425 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
4426 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
4427 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
4428 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
4429 (vec_store_lanesxi<mode>): Update accordingly.
4431 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4432 Alan Hayward <alan.hayward@arm.com>
4433 David Sherwood <david.sherwood@arm.com>
4435 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
4436 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
4437 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
4438 of units rather than the mode.
4439 * config/aarch64/iterators.md (nunits): New mode attribute.
4440 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4441 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
4442 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
4443 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
4444 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
4445 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
4446 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
4447 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
4448 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
4449 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
4450 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4451 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
4452 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
4453 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
4454 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
4455 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
4456 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
4457 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
4458 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
4459 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
4460 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
4461 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
4462 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
4463 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4464 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
4465 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
4466 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4467 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
4468 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
4469 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
4470 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
4471 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
4472 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
4473 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
4474 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
4475 (aarch64_simd_vec_setv2di): Likewise.
4477 2017-11-06 Carl Love <cel@us.ibm.com>
4479 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
4481 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
4482 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
4484 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
4486 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
4487 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
4488 Add power 8 macro expansions.
4489 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
4490 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
4491 power 8 instructions. (VSX_XXBR): Add iterator.
4493 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
4495 * config/arm/arm.md (predicable_short_it): Change default to "no",
4496 improve documentation, remove uses that are identical to the default.
4497 (enabled_for_depr_it): Rename to enabled_for_short_it.
4498 * gcc/config/arm/arm-fixed.md (predicable_short_it):
4499 Remove default uses.
4500 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
4501 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
4502 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
4503 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
4505 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4508 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
4509 float128 helper macros, which are no longer used after deleting
4510 the old 'q' built-in functions, and moving the round to odd
4511 built-in functions to being special built-in functions.
4512 (BU_FLOAT128_2): Likewise.
4513 (BU_FLOAT128_1_HW): Likewise.
4514 (BU_FLOAT128_2_HW): Likewise.
4515 (BU_FLOAT128_3_HW): Likewise.
4516 (FABSQ): Delete old 'q' built-in functions.
4517 (COPYSIGNQ): Likewise.
4518 (SQRTF128_ODD): Move round to odd built-in functions to be
4519 special built-in functions, so that we can handle
4520 -mabi=ieeelongdouble.
4521 (TRUNCF128_ODD): Likewise.
4522 (ADDF128_ODD): Likewise.
4523 (SUBF128_ODD): Likewise.
4524 (MULF128_ODD): Likewise.
4525 (DIVF128_ODD): Likewise.
4526 (FMAF128_ODD): Likewise.
4527 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
4528 built-in names to 'f128'.
4529 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
4530 old 'q' built-in functions, as the machine independent code for
4531 'f128' built-in functions handles this.
4532 (rs6000_expand_builtin): Add expansion for float128 round to odd
4533 functions, keying off on -mabi=ieeelongdouble of whether to use
4534 the KFmode or TFmode variant.
4535 (rs6000_init_builtins): Initialize the _Float128 round to odd
4537 * doc/extend.texi (PowerPC Built-in Functions): Document the old
4538 _Float128 'q' built-in functions are now mapped into the new
4539 'f128' built-in functions.
4541 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
4543 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
4544 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
4546 2017-11-06 David Malcolm <dmalcolm@redhat.com>
4549 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
4550 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
4551 * toplev.c: Include "ipa-fnsummary.h".
4552 (toplev::finalize): Call ipa_fnsummary_c_finalize.
4554 2017-11-06 Jakub Jelinek <jakub@redhat.com>
4556 PR tree-optimization/82838
4557 * gimple-ssa-store-merging.c
4558 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
4559 on a separate gimple_seq which is then appended to seq.
4561 2017-11-06 Jeff Law <law@redhat.com>
4564 * config/i386/i386.c (PROBE_INTERVAL): Remove.
4565 (get_probe_interval): New functions.
4566 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
4567 (ix86_adjust_stack_and_probe): Likewise.
4568 (output_adjust_stack_and_probe): Likewise.
4569 (ix86_emit_probe_stack_range): Likewise.
4570 (ix86_expand_prologue): Likewise.
4572 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4574 PR tree-optimization/82816
4575 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
4576 if the modes of the two types are the same.
4577 (convert_plusminus_to_widen): Likewise.
4579 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4581 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
4583 (usadv16qi): New define_expand.
4584 (usadv8hi): New define_expand.
4586 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
4589 * ipa-inline-transform.c (update_noncloned_frequencies): Always
4591 (inline_transform): Likewise.
4592 * predict.c (counts_to_freqs): Remove useless conditional.
4593 * profile-count.h (profile_count::apply_scale): Move sanity check.
4594 * tree-inline.c (copy_bb): Always scale.
4595 (copy_cfg_body): Likewise.
4597 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
4600 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
4602 (*cmp_ite0): Add enabled_for_depr_it attribute.
4603 (*cmp_ite1): Likewise.
4605 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4607 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
4610 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4612 * tree-vrp.c (vrp_int_const_binop): Return true on success and
4613 return the value by pointer.
4614 (extract_range_from_multiplicative_op_1): Update accordingly.
4615 Return as soon as an operation fails.
4617 2017-11-05 Tom de Vries <tom@codesourcery.com>
4620 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
4621 (DEF_SANITIZER_BUILTIN): ... here.
4622 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
4623 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
4625 2017-11-05 Tom de Vries <tom@codesourcery.com>
4628 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
4630 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
4631 ASM_OUTPUT_BEFORE_CASE_LABEL call.
4632 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
4634 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4635 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4636 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4638 2017-11-05 Tom de Vries <tom@codesourcery.com>
4641 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
4644 2017-11-04 Michael Clark <michaeljclark@mac.com>
4646 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
4647 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
4649 (*addsi3_extended): Likewise.
4650 (*addsi3_extended2): Likewise.
4651 (<optab>si3): Likewise.
4652 (<optab>di3): Likewise.
4653 (<optab><mode>3): Likewise.
4654 (<*optabe>si3_internal): Likewise.
4655 (zero_extendqi<SUPERQI:mode>2): Likewise.
4656 (*add<mode>hi3): Likewise.
4657 (*xor<mode>hi3): Likewise.
4658 (<optab>di3): Likewise.
4659 (*<optab>si3_extend): Likewise.
4660 (*sge<u>_<X:mode><GPR:mode>): Likewise.
4661 (*slt<u>_<X:mode><GPR:mode>): Likewise.
4662 (*sle<u>_<X:mode><GPR:mode>): Likewise.
4664 2017-11-04 Andrew Waterman <andrew@sifive.com>
4666 * config/riscv/riscv.c (riscv_option_override): Conditionally set
4667 TARGET_STRICT_ALIGN based upon -mtune argument.
4669 2017-11-04 Andrew Waterman <andrew@sifive.com>
4671 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
4673 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
4675 * config/i386/i386.c (choose_basereg): Use optional scratch
4676 register and add assertion.
4677 (x86_emit_outlined_ms2sysv_save): Use scratch register when
4678 needed, and don't allocate stack.
4679 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
4680 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
4681 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
4683 2017-11-03 Jeff Law <law@redhat.com>
4685 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
4686 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
4687 to probe at the start of a noreturn function.
4689 2017-11-03 Jakub Jelinek <jakub@redhat.com>
4691 PR tree-optimization/78821
4692 * gimple-ssa-store-merging.c: Update the file comment.
4693 (MAX_STORE_ALIAS_CHECKS): Define.
4694 (struct store_operand_info): New type.
4695 (store_operand_info::store_operand_info): New constructor.
4696 (struct store_immediate_info): Add rhs_code and ops data members.
4697 (store_immediate_info::store_immediate_info): Add rhscode, op0r
4698 and op1r arguments to the ctor, initialize corresponding data members.
4699 (struct merged_store_group): Add load_align_base and load_align
4701 (merged_store_group::merged_store_group): Initialize them.
4702 (merged_store_group::do_merge): Update them.
4703 (merged_store_group::apply_stores): Pick the constant for
4704 encode_tree_to_bitpos from one of the two operands, or skip
4705 encode_tree_to_bitpos if neither operand is a constant.
4706 (class pass_store_merging): Add process_store method decl. Remove
4707 bool argument from terminate_all_aliasing_chains method decl.
4708 (pass_store_merging::terminate_all_aliasing_chains): Remove
4709 var_offset_p argument and corresponding handling.
4710 (stmts_may_clobber_ref_p): New function.
4711 (compatible_load_p): New function.
4712 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
4713 if there is overlap and rhs_code is not INTEGER_CST. For
4714 non-overlapping stores terminate group if rhs is not mergeable.
4715 (get_alias_type_for_stmts): Change first argument from
4716 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
4717 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
4718 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
4720 (get_location_for_stmts): Change first argument from
4721 auto_vec<gimple *> & to vec<gimple *> &.
4722 (struct split_store): Remove orig_stmts data member, add orig_stores.
4723 (split_store::split_store): Create orig_stores rather than orig_stmts.
4724 (find_constituent_stmts): Renamed to ...
4725 (find_constituent_stores): ... this. Change second argument from
4726 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
4727 to info structures rather than the statements.
4728 (split_group): Rename ALLOW_UNALIGNED argument to
4729 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
4730 it. Adjust find_constituent_stores caller.
4731 (imm_store_chain_info::output_merged_store): Handle rhs_code other
4732 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
4733 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
4734 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
4735 (mem_valid_for_store_merging): New function.
4736 (handled_load): New function.
4737 (pass_store_merging::process_store): New method.
4738 (pass_store_merging::execute): Use process_store method. Adjust
4739 terminate_all_aliasing_chains caller.
4741 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4743 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4744 Return true for more constants, symbols and label references.
4745 (aarch64_valid_floating_const): Remove unused function.
4747 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4750 * config/aarch64/aarch64.c (aarch64_layout_frame):
4751 Undo forcing of LR at bottom of frame.
4753 2017-11-03 Jeff Law <law@redhat.com>
4756 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
4757 for int_registers_saved.
4759 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
4760 extracted from tree-ssa-dom.c.
4761 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
4762 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
4763 (record_equivalences_from_incoming_edge): Add additional argument
4764 to single_pred_edge_ignoring_loop_edges call.
4765 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
4766 (uncprop_dom_walker::before_dom_children): Add additional argument
4767 to single_pred_edge_ignoring_loop_edges call.
4768 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
4769 single_pred_edge_ignoring_loop_edges rather than open coding.
4770 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
4772 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
4774 * match.pd (-(-A)): Rewrite.
4776 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4778 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
4779 (rs6000_emit_int_cmove): New declaration.
4780 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
4781 (rs6000_emit_sISEL): Delete.
4782 (rs6000_emit_int_cmove): Make non-static.
4783 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
4784 instead of rs6000_emit_sISEL.
4786 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
4788 * asan.c (create_cond_insert_point): Maintain profile.
4789 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
4791 * basic-block.h (struct basic_block_def): Remove frequency.
4792 (EDGE_FREQUENCY): Use to_frequency
4793 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
4795 (find_traces): Update to use to_frequency.
4796 (find_traces_1_round): Likewise; use only IPA counts.
4797 (bb_to_key): Likewise.
4798 (connect_traces): Use IPA counts only.
4799 (copy_bb_p): Update to use to_frequency.
4800 (fix_up_crossing_landing_pad): Likewise.
4801 (sanitize_hot_paths): Likewise.
4802 * bt-load.c (basic_block_freq): Likewise.
4803 * cfg.c (init_flow): Set count_max to uninitialized.
4804 (check_bb_profile): Remove frequencies; check counts.
4805 (dump_bb_info): Do not dump frequencies.
4806 (update_bb_profile_for_threading): Update counts only.
4807 (scale_bbs_frequencies_int): Likewise.
4808 (MAX_SAFE_MULTIPLIER): Remove.
4809 (scale_bbs_frequencies_gcov_type): Update counts only.
4810 (scale_bbs_frequencies_profile_count): Update counts only.
4811 (scale_bbs_frequencies): Update counts only.
4812 * cfg.h (struct control_flow_graph): Add count-max.
4813 (update_bb_profile_for_threading): Update prototype.
4814 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
4815 (find_many_sub_basic_blocks): Likewise.
4816 * cfgcleanup.c (try_forward_edges): Likewise.
4817 (try_crossjump_to_edge): Likewise.
4818 * cfgexpand.c (expand_gimple_cond): Likewise.
4819 (expand_gimple_tailcall): Likewise.
4820 (construct_init_block): Likewise.
4821 (construct_exit_block): Likewise.
4822 * cfghooks.c (verify_flow_info): Check consistency of counts.
4823 (dump_bb_for_graph): Do not dump frequencies.
4824 (split_block_1): Do not update frequencies.
4825 (split_edge): Do not update frequencies.
4826 (make_forwarder_block): Do not update frequencies.
4827 (duplicate_block): Do not update frequencies.
4828 (account_profile_record): Do not update frequencies.
4829 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
4830 for global heuristics.
4831 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
4832 (expected_loop_iterations_unbounded): Use counts only.
4833 * cfgloopmanip.c (scale_loop_profile): Simplify.
4834 (create_empty_loop_on_edge): Simplify
4836 (duplicate_loop_to_header_edge): Simplify
4837 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
4838 (update_br_prob_note): Take care of removing note when profile
4840 (relink_block_chain): Do not dump frequency.
4841 (rtl_account_profile_record): Use to_frequency.
4842 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
4843 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
4844 (cgraph_update_edges_for_call_stmt_node): Likewise.
4845 (cgraph_edge::verify_count_and_frequency): Update.
4846 (cgraph_node::verify_node): Temporarily disable frequency verification.
4847 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
4848 to_cgraph_frequency.
4849 (cgraph_edge::rebuild_edges): Convert to ipa counts.
4850 * cgraphunit.c (init_lowered_empty_function): Do not initialize
4852 (cgraph_node::expand_thunk): Update profile.
4853 * except.c (dw2_build_landing_pads): Do not update frequency.
4854 * final.c (compute_alignments): Use to_frequency.
4855 (dump_basic_block_info): Do not dump frequency.
4856 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
4857 (dump_gimple_bb_header): Do not dump frequency.
4858 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
4860 * gimple-streamer-in.c (input_bb): Do not stream frequency.
4861 * gimple-streamer-out.c (output_bb): Do not stream frequency.
4862 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
4863 (init_before_recovery): Do not update frequency.
4864 (sched_create_recovery_edges): Do not update frequency.
4865 * hsa-gen.c (convert_switch_statements): Do not update frequency.
4866 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
4867 (ipa_cp_c_finalize): Set max_count to uninitialized.
4868 * ipa-fnsummary.c (get_minimal_bb): Use counts.
4869 (param_change_prob): Use counts.
4870 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
4872 * ipa-split.c (consider_split): Use to_frequency.
4873 (split_function): Use to_frequency.
4874 * ira-build.c (loop_compare_func): Likewise.
4875 (mark_loops_for_removal): Likewise.
4876 (mark_all_loops_for_removal): Likewise.
4877 * loop-doloop.c (doloop_modify): Do not update frequency.
4878 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
4880 * lto-streamer-in.c (input_function): Update count_max.
4881 * omp-expand.c (expand_omp_taskreg): Update count_max.
4882 * omp-simd-clone.c (simd_clone_adjust): Update profile.
4883 * predict.c (maybe_hot_frequency_p): Use to_frequency.
4884 (maybe_hot_count_p): Use ipa counts only.
4885 (maybe_hot_bb_p): Simplify.
4886 (maybe_hot_edge_p): Simplify.
4887 (probably_never_executed): Do not take frequency argument.
4888 (probably_never_executed_bb_p): Do not pass frequency.
4889 (probably_never_executed_edge_p): Likewise.
4890 (combine_predictions_for_bb): Check that profile is nonzero.
4891 (propagate_freq): Do not set frequency.
4892 (drop_profile): Simplify.
4893 (counts_to_freqs): Simplify.
4894 (expensive_function_p): Use to_frequency.
4895 (propagate_unlikely_bbs_forward): Simplify.
4896 (determine_unlikely_bbs): Simplify.
4897 (estimate_bb_frequencies): Add hack to silence graphite issues.
4898 (compute_function_frequency): Use ipa counts.
4899 (pass_profile::execute): Update.
4900 (rebuild_frequencies): Use counts only.
4901 (force_edge_cold): Use counts only.
4902 * profile-count.c (profile_count::dump): Dump new count types.
4903 (profile_count::differs_from_p): Check compatiblity.
4904 (profile_count::to_frequency): New function.
4905 (profile_count::to_cgraph_frequency): New function.
4906 * profile-count.h (struct function): Declare.
4907 (enum profile_quality): Add profile_guessed_local and
4908 profile_guessed_global0.
4909 (class profile_proability): Decrease number of bits to 29;
4910 update from_reg_br_prob_note and to_reg_br_prob_note.
4911 (class profile_count: Update comment; decrease number of bits
4912 to 61. Check compatibility.
4913 (profile_count::compatible_p): New private member function.
4914 (profile_count::ipa_p): New member function.
4915 (profile_count::operator<): Handle global zero correctly.
4916 (profile_count::operator>): Handle global zero correctly.
4917 (profile_count::operator<=): Handle global zero correctly.
4918 (profile_count::operator>=): Handle global zero correctly.
4919 (profile_count::nonzero_p): New member function.
4920 (profile_count::force_nonzero): New member function.
4921 (profile_count::max): New member function.
4922 (profile_count::apply_scale): Handle IPA scalling.
4923 (profile_count::guessed_local): New member function.
4924 (profile_count::global0): New member function.
4925 (profile_count::ipa): New member function.
4926 (profile_count::to_frequency): Declare.
4927 (profile_count::to_cgraph_frequency): Declare.
4928 * profile.c (OVERLAP_BASE): Delete.
4929 (compute_frequency_overlap): Delete.
4930 (compute_branch_probabilities): Do not use compute_frequency_overlap.
4931 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
4932 * sched-ebb.c (rank): Use counts only.
4933 * shrink-wrap.c (handle_simple_exit): Use counts only.
4934 (try_shrink_wrapping): Use counts only.
4935 (place_prologue_for_one_component): Use counts only.
4936 * tracer.c (find_best_predecessor): Use to_frequency.
4937 (find_trace): Use to_frequency.
4938 (tail_duplicate): Use to_frequency.
4939 * trans-mem.c (expand_transaction): Do not update frequency.
4940 * tree-call-cdce.c: Do not update frequency.
4941 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
4942 (gimple_merge_blocks): Likewise.
4943 (gimple_split_edge): Likewise.
4944 (gimple_duplicate_sese_region): Likewise.
4945 (gimple_duplicate_sese_tail): Likewise.
4946 (move_sese_region_to_fn): Likewise.
4947 (gimple_account_profile_record): Likewise.
4948 (insert_cond_bb): Likewise.
4949 * tree-complex.c (expand_complex_div_wide): Likewise.
4950 * tree-eh.c (lower_resx): Update profile.
4951 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
4953 (initialize_cfun): Do not initialize frequencies
4954 (freqs_to_counts): Delete.
4955 (copy_cfg_body): Ignore count parameter.
4956 (copy_body): Update.
4957 (expand_call_inline): Update count_max.
4958 (optimize_inline_calls): Update count_max.
4959 (tree_function_versioning): Update count_max.
4960 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
4961 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
4963 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
4964 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
4965 (try_peel_loop): Likewise.
4966 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
4968 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
4969 (tree_transform_and_unroll_loop): Do not use frequencies
4970 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
4971 Use reliable prediction only.
4972 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
4973 * tree-ssa-sink.c (select_best_block): Use to_frequency.
4974 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
4975 probability scaling.
4976 * tree-ssa-threadupdate.c (create_block_for_threading): Do
4977 not update frequency
4978 (any_remaining_duplicated_blocks): Likewise.
4979 (update_profile): Likewise.
4980 (estimated_freqs_path): Delete.
4981 (freqs_to_counts_path): Delete.
4982 (clear_counts_path): Delete.
4983 (ssa_fix_duplicate_block_edges): Likewise.
4984 (duplicate_thread_path): Likewise.
4985 * tree-switch-conversion.c (gen_inbound_check): Use counts.
4986 * tree-tailcall.c (decrease_profile): Do not update frequency.
4987 (eliminate_tail_call): Likewise.
4988 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
4989 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
4990 (optimize_mask_stores): Likewise.
4991 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4992 * ubsan.c (ubsan_expand_null_ifn): Update profile.
4993 (ubsan_expand_ptr_ifn): Update profile.
4994 * value-prof.c (gimple_ic): Simplify.
4995 * value-prof.h (gimple_ic): Update prototype.
4996 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
4997 * ipa-inline.c (compute_uninlined_call_time): Be sure that
4999 (want_inline_self_recursive_call_p): Likewise.
5000 (resolve_noninline_speculation): Only cummulate defined counts.
5001 (inline_small_functions): Use nonzero_p.
5002 (ipa_inline): Do not access freed node.
5004 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5006 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5007 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5009 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5011 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5012 non-legitimate address.
5014 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5016 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5017 (*lt0_<mode>di, *lt0_<mode>si): New.
5019 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5021 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5022 TARGET_PAIRED_FLOAT.
5024 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5025 Jim Wilson <jim.wilson@linaro.org>
5027 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5028 * config/aarch64/aarch64-tune.md: Regenerated.
5029 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5030 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5032 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5034 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5035 (arc_expand_prologue): Restore blink for millicode.
5036 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5038 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5041 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5042 gen_vec_duplicate after forcing the scalar into a register.
5044 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5046 * combine (try_combine): Print the insns input to try_combine to the
5049 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5052 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5053 Remove second argument from aarch64_process_target_attr call.
5054 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5056 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5058 (aarch64_handle_attr_arch): Remove second argument.
5059 (aarch64_handle_attr_cpu): Ditto.
5060 (aarch64_handle_attr_tune): Ditto.
5061 (aarch64_handle_attr_isa_flags): Ditto.
5062 (aarch64_process_one_target_attr): Ditto.
5063 (aarch64_process_target_attr): Ditto.
5064 (aarch64_option_valid_attribute_p): Remove second argument.
5065 on aarch64_process_target_attr call.
5067 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5069 * diagnostic.c: Include "selftest-diagnostic.h".
5070 (selftest::assert_location_text): New function.
5071 (selftest::test_diagnostic_get_location_text): New function.
5072 (selftest::diagnostic_c_tests): Call it.
5074 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5076 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5077 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5078 (class selftest::test_diagnostic_context): Move to...
5079 * selftest-diagnostic.c: New file.
5080 * selftest-diagnostic.h: New file.
5082 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5084 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5085 offset range for FT32B.
5086 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5087 * config/ft32/ft32.md: Add TARGET_NOPM.
5088 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5089 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5091 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5093 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5095 2017-11-02 Jeff Law <law@redhat.com>
5097 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5098 virtual keyword on FINAL OVERRIDE members.
5100 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5101 virtuals together. Add virtual destructor.
5102 (substitute_and_fold_engine): Similarly.
5104 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5106 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5108 2017-11-02 Richard Biener <rguenther@suse.de>
5110 PR tree-optimization/82795
5111 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5113 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5115 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5117 (gcc_GAS_CHECK_FEATURE): Remove.
5118 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5119 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5120 * configure: Regenerate.
5122 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5124 * config/arc/arc.c (hwloop_optimize): Account for empty
5127 2017-11-02 Richard Biener <rguenther@suse.de>
5130 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5131 Truncate ARRAY_REF index and element size.
5133 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5135 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5137 2017-11-01 Jeff Law <law@redhat.com>
5139 * tree-ssa-ccp.c (ccp_folder): New class derived from
5140 substitute_and_fold_engine.
5141 (ccp_folder::get_value): New member function.
5142 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5143 (ccp_fold_stmt): Remove prototype.
5144 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5145 * tree-ssa-copy.c (copy_folder): New class derived from
5146 substitute_and_fold_engine.
5147 (copy_folder::get_value): Renamed from get_value.
5148 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5149 * tree-vrp.c (vrp_folder): New class derived from
5150 substitute_and_fold_engine.
5151 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5152 (vrp_folder::get_value): New member function.
5153 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5154 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5155 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5156 provide a class interface to folder/substitute routines.
5157 (ssa_prop_fold_stmt_fn): Remove typedef.
5158 (ssa_prop_get_value_fn): Likewise.
5159 (subsitute_and_fold): Remove prototype.
5160 (replace_uses_in): Likewise.
5161 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5162 Renamed from replace_uses_in. Call the virtual member function
5163 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5164 (substitute_and_fold_dom_walker): Remove initialization of
5165 data member entries for calbacks. Add substitute_and_fold_engine
5166 member and initialize it.
5167 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5168 member functions for get_value, replace_phi_args_in c
5169 replace_uses_in, and fold_stmt calls.
5170 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5171 substitute_and_fold. Remove assert. Update ctor call.
5173 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5174 (ssa_prop_visit_phi_fn): Likewise.
5175 (class ssa_propagation_engine): New class to provide an interface
5177 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5179 (ssa_prop_visit_phi): Likewise.
5180 (ssa_propagation_engine::simulate_stmt): Moved into class.
5181 Call visit_phi/visit_stmt from the class rather than via
5182 file scoped static variables.
5183 (ssa_propagation_engine::simulate_block): Moved into class.
5184 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5185 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5186 set file scoped statics for the visit_stmt/visit_phi callbacks.
5187 * tree-complex.c (complex_propagate): New class derived from
5188 ssa_propagation_engine.
5189 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5190 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5191 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5193 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5194 ssa_propagation_engine.
5195 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5196 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5197 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5198 * tree-ssa-copy.c (copy_prop): New class derived from
5199 ssa_propagation_engine.
5200 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5201 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5202 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5203 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5204 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5205 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5206 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5208 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5210 PR rtl-optimization/82778
5211 PR rtl-optimization/82597
5212 * compare-elim.c (struct comparison): Add in_a_setter field.
5213 (find_comparison_dom_walker::before_dom_children): Remove killed
5214 bitmap and df_simulate_find_defs call, instead walk the defs.
5215 Compute last_setter and initialize in_a_setter. Merge definitions
5216 with first initialization for a few variables.
5217 (try_validate_parallel): Use insn_invalid_p instead of
5218 recog_memoized. Return insn rather than just the pattern.
5219 (try_merge_compare): Fix up comment. Don't uselessly test if
5220 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
5222 (execute_compare_elim_after_reload): Remove df_chain_add_problem
5225 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5226 Alan Hayward <alan.hayward@arm.com>
5227 David Sherwood <david.sherwood@arm.com>
5229 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
5230 aarch64_hard_regno_nregs to get the number of registers
5233 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5234 Alan Hayward <alan.hayward@arm.com>
5235 David Sherwood <david.sherwood@arm.com>
5237 * config/aarch64/constraints.md (Upl): Rename to...
5239 * config/aarch64/aarch64.md
5240 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
5243 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5244 Alan Hayward <alan.hayward@arm.com>
5245 David Sherwood <david.sherwood@arm.com>
5247 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
5248 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
5251 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5252 Alan Hayward <alan.hayward@arm.com>
5253 David Sherwood <david.sherwood@arm.com>
5255 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
5256 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
5257 (aarch64_evpc_dup): Generate rtl direcly, rather than using
5259 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
5260 of a single element.
5261 * config/aarch64/iterators.md: Add a comment above the permute
5262 unspecs to say that they are generated directly by
5263 aarch64_expand_vec_perm_const.
5264 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
5266 2017-11-01 Nathan Sidwell <nathan@acm.org>
5268 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
5270 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5272 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
5273 and medany code models, and describe what they do.
5275 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5277 Revert accidental duplicate:
5279 * combine.c (can_change_dest_mode): Reject changes in
5280 REGMODE_NATURAL_SIZE.
5282 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
5284 PR rtl-optimization/64682
5285 PR rtl-optimization/69567
5286 PR rtl-optimization/69737
5287 PR rtl-optimization/82683
5288 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
5289 register mentioned in the note, drop the note, unless it came from I3,
5290 in which case it should go to I3 again.
5292 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5294 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
5295 and return false if not.
5296 (clear_bytes_written_by, live_bytes_read): Update accordingly.
5298 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5300 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
5301 range is known to be empty.
5303 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5304 Alan Hayward <alan.hayward@arm.com>
5305 David Sherwood <david.sherwood@arm.com>
5307 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
5308 and CONST_VECTOR_NUNITS instead of computing the number of units from
5309 the byte sizes of the vector and element.
5310 (simplify_binary_operation_1): Likewise.
5311 (simplify_const_binary_operation): Likewise.
5312 (simplify_ternary_operation): Likewise.
5314 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5315 Alan Hayward <alan.hayward@arm.com>
5316 David Sherwood <david.sherwood@arm.com>
5318 * var-tracking.c (INT_MEM_OFFSET): Replace with...
5319 (int_mem_offset): ...this new function.
5320 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
5321 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
5322 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
5325 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5326 Alan Hayward <alan.hayward@arm.com>
5327 David Sherwood <david.sherwood@arm.com>
5329 * lower-subreg.c (interesting_mode_p): New function.
5330 (compute_costs, find_decomposable_subregs, decompose_register)
5331 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
5332 (resolve_clobber, dump_choices): Use it.
5334 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5335 Alan Hayward <alan.hayward@arm.com>
5336 David Sherwood <david.sherwood@arm.com>
5338 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
5340 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5341 Alan Hayward <alan.hayward@arm.com>
5342 David Sherwood <david.sherwood@arm.com>
5344 * alias.c (find_base_value, find_base_term): Only process integer
5345 truncations. Check the precision rather than the size.
5347 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5348 Alan Hayward <alan.hayward@arm.com>
5349 David Sherwood <david.sherwood@arm.com>
5351 * machmode.h (is_narrower_int_mode): New function
5352 * optabs.c (expand_float, expand_fix): Use it.
5353 * dwarf2out.c (rotate_loc_descriptor): Likewise.
5355 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5356 Alan Hayward <alan.hayward@arm.com>
5357 David Sherwood <david.sherwood@arm.com>
5359 * rtl.h (narrower_subreg_mode): New function.
5360 * ira-color.c (update_costs_from_allocno): Use it.
5362 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5363 Alan Hayward <alan.hayward@arm.com>
5364 David Sherwood <david.sherwood@arm.com>
5366 * optabs-query.h (convert_optab_p): New function, split out from...
5367 (convert_optab_handler): ...here.
5368 (widening_optab_handler): Delete.
5369 (find_widening_optab_handler): Remove permit_non_widening parameter.
5370 (find_widening_optab_handler_and_mode): Likewise. Provide an
5371 override that operates on mode class wrappers.
5372 * optabs-query.c (widening_optab_handler): Delete.
5373 (find_widening_optab_handler_and_mode): Remove permit_non_widening
5374 parameter. Assert that the two modes are the same class and that
5375 the "from" mode is narrower than the "to" mode. Use
5376 convert_optab_handler instead of widening_optab_handler.
5377 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
5378 instead of widening_optab_handler.
5379 * expr.c (expand_expr_real_2): Update calls to
5380 find_widening_optab_handler.
5381 * optabs.c (expand_widen_pattern_expr): Likewise.
5382 (expand_binop_directly): Take the insn_code as a parameter.
5383 (expand_binop): Only call find_widening_optab_handler for
5384 conversion optabs; use optab_handler otherwise. Update calls
5385 to find_widening_optab_handler and expand_binop_directly.
5386 Use convert_optab_handler instead of widening_optab_handler.
5387 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
5388 find_widening_optab_handler and use scalar_mode rather than
5390 (convert_plusminus_to_widen): Likewise.
5392 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5393 Alan Hayward <alan.hayward@arm.com>
5394 David Sherwood <david.sherwood@arm.com>
5396 * machmode.h (fixed_size_mode): New class.
5397 * rtl.h (get_pool_mode): Return fixed_size_mode.
5398 * gengtype.c (main): Add fixed_size_mode.
5399 * target.def (get_raw_result_mode): Return a fixed_size_mode.
5400 (get_raw_arg_mode): Likewise.
5401 * doc/tm.texi: Regenerate.
5402 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
5403 * targhooks.c (default_get_reg_raw_mode): Likewise.
5404 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
5405 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
5406 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
5407 (msp430_get_raw_result_mode): Likewise.
5408 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
5409 * dbxout.c (dbxout_parms): Require fixed-size modes.
5410 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
5411 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5412 * omp-low.c (lower_oacc_reductions): Likewise.
5413 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
5414 (simplify_subreg): Update accordingly.
5415 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
5416 (force_const_mem): Update accordingly. Return NULL_RTX for modes
5417 that aren't fixed-size.
5418 (get_pool_mode): Return a fixed_size_mode.
5419 (output_constant_pool_2): Take a fixed_size_mode.
5421 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5422 Alan Hayward <alan.hayward@arm.com>
5423 David Sherwood <david.sherwood@arm.com>
5425 * doc/rtl.texi (vec_series): Document.
5426 (const): Say that the operand can be a vec_series.
5427 * rtl.def (VEC_SERIES): New rtx code.
5428 * rtl.h (const_vec_series_p_1): Declare.
5429 (const_vec_series_p): New function.
5430 * emit-rtl.h (gen_const_vec_series): Declare.
5431 (gen_vec_series): Likewise.
5432 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
5433 (gen_vec_series): Likewise.
5434 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
5435 * simplify-rtx.c (simplify_unary_operation): Handle negations
5437 (simplify_binary_operation_series): New function.
5438 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
5439 (test_vector_ops_series): New function.
5440 (test_vector_ops): Call it.
5441 * config/powerpcspe/altivec.md (altivec_lvsl): Use
5442 gen_const_vec_series.
5443 (altivec_lvsr): Likewise.
5444 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
5446 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5447 Alan Hayward <alan.hayward@arm.com>
5448 David Sherwood <david.sherwood@arm.com>
5450 * doc/rtl.texi (const): Update description of address constants.
5451 Say that vector constants are allowed too.
5452 * common.md (E, F): Use CONSTANT_P instead of checking for
5454 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
5455 checking for CONST_VECTOR.
5456 * expmed.c (make_tree): Use build_vector_from_val for a CONST
5458 * expr.c (expand_expr_real_2): Check for vector modes instead
5459 of checking for CONST_VECTOR.
5460 * rtl.h (const_vec_p): New function.
5461 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
5462 (unwrap_const_vec_duplicate): Handle them here too.
5464 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5465 David Malcolm <dmalcolm@redhat.com>
5466 Alan Hayward <alan.hayward@arm.com>
5467 David Sherwood <david.sherwood@arm.com>
5469 * rtl.h (vec_duplicate_p): New function.
5470 * selftest-rtl.c (assert_rtx_eq_at): New function.
5471 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
5472 (assert_rtx_eq_at): Declare.
5473 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
5474 * selftest-run-tests.c (selftest::run_tests): Call it.
5475 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
5476 (simplify_unary_operation_1): Recursively handle vector duplicates.
5477 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
5479 (simplify_subreg): Handle subregs of vector duplicates.
5480 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
5481 (selftest::simplify_rtx_c_tests): New functions.
5483 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5484 Alan Hayward <alan.hayward@arm.com>
5485 David Sherwood <david.sherwood@arm.com>
5487 * emit-rtl.h (gen_const_vec_duplicate): Declare.
5488 (gen_vec_duplicate): Likewise.
5489 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
5491 (gen_const_vector): ...here.
5492 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
5493 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
5494 whose elements are all equal.
5495 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
5496 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
5497 (simplify_relational_operation): Likewise.
5498 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5500 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
5501 (aarch64_expand_vector_init): Likewise.
5502 * config/arm/arm.c (neon_vdup_constant): Likewise.
5503 (neon_expand_vector_init): Likewise.
5504 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
5505 (arm_block_set_unaligned_vect): Likewise.
5506 (arm_block_set_aligned_vect): Likewise.
5507 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
5508 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
5509 (expand_vec_perm_even_odd_pack): Likewise.
5510 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
5511 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
5512 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
5513 gen_const_vec_duplicate.
5514 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
5515 * config/mips/mips.c (mips_gen_const_int_vector): Use
5516 gen_const_vec_duplicate.
5517 (mips_expand_vector_init): Use CONST0_RTX.
5518 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
5519 (define_split): Use gen_const_vec_duplicate.
5520 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
5521 (define_split): Use gen_const_vec_duplicate.
5522 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
5523 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
5524 * config/spu/spu.c (spu_const): Likewise.
5526 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5527 Alan Hayward <alan.hayward@arm.com>
5528 David Sherwood <david.sherwood@arm.com>
5530 * combine.c (can_change_dest_mode): Reject changes in
5531 REGMODE_NATURAL_SIZE.
5533 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
5535 * configure.ac (--enable-libssp): New.
5536 (gcc_cv_libc_provides_ssp): Check for explicit setting before
5537 trying to determine target-specific default. Adjust indentation.
5538 * configure: Regenerated.
5539 * doc/install.texi (Configuration): Expand --disable-libssp
5542 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
5544 config/i386/i386.c (ix86_expand_epilogue): Correct stack
5547 2017-10-31 Martin Jambor <mjambor@suse.cz>
5550 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
5552 2017-10-31 David Malcolm <dmalcolm@redhat.com>
5554 * auto-profile.c (autofdo_source_profile::read): Use
5555 UNKNOWN_LOCATION rather than 0.
5556 * diagnostic-core.h (warning_at_rich_loc): Rename to...
5557 (warning_at): ...this overload.
5558 (warning_at_rich_loc_n): Rename to...
5559 (warning_n): ...this overload.
5560 (error_at_rich_loc): Rename to...
5561 (error_at): ...this overload.
5562 (pedwarn_at_rich_loc): Rename to...
5563 (pedwarn): ...this overload.
5564 (permerror_at_rich_loc): Rename to...
5565 (permerror): ...this overload.
5566 (inform_at_rich_loc): Rename to...
5567 (inform): ...this overload.
5568 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
5569 (diagnostic_n_impl_richloc): Rename to...
5570 (diagnostic_n_impl): ...this rich_location *-based decl.
5571 (inform_at_rich_loc): Rename to...
5572 (inform): ...this, and add an assertion.
5573 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
5574 (warning_at_rich_loc): Rename to...
5575 (warning_at): ...this, and add an assertion.
5576 (warning_at_rich_loc_n): Rename to...
5577 (warning_n): ...this, and add an assertion.
5578 (warning_n): Update location_t-based implementation for removal of
5579 location_t-based diagnostic_n_impl.
5580 (pedwarn_at_rich_loc): Rename to...
5581 (pedwarn): ...this, and add an assertion.
5582 (permerror_at_rich_loc): Rename to...
5583 (permerror): ...this, and add an assertion.
5584 (error_n): Update for removal of location_t-based diagnostic_n_impl.
5585 (error_at_rich_loc): Rename to...
5586 (error_at): ...this, and add an assertion.
5587 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
5588 (driver::do_spec_on_infiles): Likewise.
5589 * substring-locations.c (format_warning_va): Update for renaming
5590 of inform_at_rich_loc.
5592 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
5594 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
5595 _Float<N>X built-in functions so that the variant without the
5596 "__builtin_" prefix is only enabled for the GNU C and Objective C
5597 languages when they are in non-strict ANSI/ISO mode.
5598 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
5599 * target.def (floatn_builtin_p): Add a target hook to control
5600 whether _Float<N> and _Float<N>X built-in functions without the
5601 "__builtin_" prefix are enabled, and return true for C and
5602 Objective C in the default hook. Include langhooks.h in
5604 * targhooks.h (default_floatn_builtin_p): Likewise.
5605 * targhooks.c (default_floatn_builtin_p): Likewise.
5606 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
5607 floatn_builtin_p target hook.
5608 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
5610 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
5611 Eric Botcazou <ebotcazou@adacore.com>
5613 PR rtl-optimization/81803
5614 * lra-constraints.c (curr_insn_transform): Also reload the whole
5615 register for a strict subreg no wider than a word if this is for
5616 a WORD_REGISTER_OPERATIONS target.
5618 2017-10-31 Jason Merrill <jason@redhat.com>
5620 * gdbinit.in: Skip over inlines from timevar.h.
5622 2017-10-31 Martin Liska <mliska@suse.cz>
5624 * doc/gcov.texi: Document new option.
5625 * gcov.c (print_usage): Likewise print it.
5626 (process_args): Support the argument.
5627 (format_count): New function.
5628 (format_gcov): Use the function.
5630 2017-10-31 Martin Liska <mliska@suse.cz>
5632 * gcov.c (struct name_map): do not use typedef.
5633 Define operator== and operator<.
5634 (name_search): Remove.
5635 (name_sort): Remove.
5636 (main): Do not allocate names.
5637 (process_file): Add vertical space.
5638 (generate_results): Use std::find.
5639 (release_structures): Do not release memory.
5640 (find_source): Use std::find.
5642 2017-10-31 Martin Liska <mliska@suse.cz>
5644 * gcov.c (struct line_info): Remove it's typedef.
5645 (line_info::line_info): Add proper ctor.
5646 (line_info::has_block): Do not use a typedef.
5647 (struct source_info): Do not use typedef.
5648 (circuit): Likewise.
5649 (get_cycles_count): Likewise.
5650 (output_intermediate_file): Iterate via vector iterator.
5651 (add_line_counts): Use std::vector methods.
5652 (accumulate_line_counts): Likewise.
5653 (output_lines): Likewise.
5655 2017-10-31 Martin Liska <mliska@suse.cz>
5657 * gcov.c (struct source_info): Remove typedef.
5658 (source_info::source_info): Add proper ctor.
5659 (accumulate_line_counts): Use struct, not it's typedef.
5660 (output_gcov_file): Likewise.
5661 (output_lines): Likewise.
5662 (main): Do not allocate an array.
5663 (output_intermediate_file): Use size of vector container.
5664 (process_file): Resize the vector.
5665 (generate_results): Do not preallocate, use newly added vector
5667 (release_structures): Do not release sources.
5668 (find_source): Use vector methods.
5669 (add_line_counts): Do not use typedef.
5671 2017-10-31 Martin Liska <mliska@suse.cz>
5673 * doc/gcov.texi: Document that.
5674 * gcov.c (add_line_counts): Mark lines with a non-executed
5676 (output_line_beginning): Handle such lines.
5677 (output_lines): Pass new argument.
5678 (output_intermediate_file): Print it in intermediate format.
5680 2017-10-31 Martin Liska <mliska@suse.cz>
5682 * color-macros.h: New file.
5683 * diagnostic-color.c: Factor out color related to macros to
5685 * doc/gcov.texi: Document -k option.
5686 * gcov.c (INCLUDE_STRING): Include string.h.
5687 (print_usage): Add -k option.
5688 (process_args): Parse it.
5689 (pad_count_string): New function.
5690 (output_line_beginning): Likewise.
5691 (DEFAULT_LINE_START): New macro.
5692 (output_lines): Support color output.
5694 2017-10-31 Martin Liska <mliska@suse.cz>
5696 PR gcov-profile/82633
5697 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
5698 their interaction with GCOV infrastructure.
5699 * configure.ac: Add -fkeep-{inline,static}-functions to
5701 * configure: Regenerate.
5703 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
5706 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
5708 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
5711 * config/rs6000/rs6000.md (allocate_stack): Force update interval
5712 into a register if it does not fit into an immediate offset field.
5714 2017-10-31 Olivier Hainque <hainque@adacore.com>
5716 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
5718 2017-10-31 Julia Koval <julia.koval@intel.com>
5720 * config.gcc: Add gfniintrin.h.
5721 * config/i386/gfniintrin.h: New.
5722 * config/i386/i386-builtin-types.def
5723 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5724 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5725 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5726 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5727 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5728 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
5729 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
5730 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
5731 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
5732 V64QI_FTYPE_V64QI_V64QI_INT): New types.
5733 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5734 * config/i386/immintrin.h: Include gfniintrin.h.
5735 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
5737 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
5739 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
5741 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5743 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
5744 (arm_ashldi3_1bit): Remove pattern.
5745 (ashrdi3): Remove shift by 1 expansion.
5746 (arm_ashrdi3_1bit): Remove pattern.
5747 (lshrdi3): Remove shift by 1 expansion.
5748 (arm_lshrdi3_1bit): Remove pattern.
5749 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
5750 cost of ashldi3 by 1.
5751 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
5752 (<shift>di3_neon): Likewise.
5754 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5756 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
5757 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
5758 and (*aarch64_simd_mov<VQ:mode>).
5759 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
5760 pattern alternative.
5761 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
5762 attributes to match pattern alternative.
5764 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
5766 * config.gcc (powerpc*-*-*): Add emmintrin.h.
5767 * config/rs6000/emmintrin.h: New file.
5768 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
5770 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5772 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
5773 * (movdi_vfp_cortexa8): Remove pattern.
5775 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5777 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
5779 (Specific, alpha*-dec-osf5.1): Remove.
5780 (Specific, mips-sgi-irix5): Remove.
5781 (Specific, mips-sgi-irix6): Remove.
5783 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5786 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
5787 arguments to clear_bit_region_be.
5789 2017-10-30 Jim Wilson <wilson@tuliptree.org>
5791 * gimplify.c: Include memmodel.h.
5793 2017-10-30 Martin Jambor <mjambor@suse.cz>
5795 * omp-grid.c (grid_attempt_target_gridification): Also insert a
5796 condition whether loop should be executed at all.
5798 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
5800 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5801 gimple folding of vec_madd() intrinsics.
5802 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
5803 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
5804 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
5806 2017-10-30 Richard Biener <rguenther@suse.de>
5808 PR tree-optimization/82762
5810 2017-10-23 Richard Biener <rguenther@suse.de>
5812 PR tree-optimization/82129
5814 2017-08-01 Richard Biener <rguenther@suse.de>
5816 PR tree-optimization/81181
5817 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5818 (compute_antic): ... end of iteration here.
5820 2017-10-30 Joseph Myers <joseph@codesourcery.com>
5822 * doc/invoke.texi (C Dialect Options): Document -std=c17,
5823 -std=iso9899:2017 and -std=gnu17.
5824 * doc/standards.texi (C Language): Document C17 support.
5825 * doc/cpp.texi (Overview): Mention -std=c17.
5826 (Standard Predefined Macros): Document C11 and C17 values of
5827 __STDC_VERSION__. Do not refer to C99 support as incomplete.
5828 * doc/extend.texi (Inline): Do not list individual options for
5829 standards newer than C99.
5830 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
5832 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
5835 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5837 * asan.c (asan_finish_file): Align asan globals array by shadow
5840 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5843 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
5844 (struct store_immediate_info): Add bitregion_start and bitregion_end
5846 (store_immediate_info::store_immediate_info): Add brs and bre
5847 arguments and initialize bitregion_{start,end} from those.
5848 (struct merged_store_group): Add bitregion_start, bitregion_end,
5849 align_base and mask fields. Drop unnecessary struct keyword from
5850 struct store_immediate_info. Add do_merge method.
5851 (clear_bit_region_be): Use memset instead of loop storing zeros.
5852 (merged_store_group::do_merge): New method.
5853 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
5854 stores as long as the surrounding bitregions have no gaps.
5855 (merged_store_group::merge_overlapping): Use do_merge.
5856 (merged_store_group::apply_stores): Test that bitregion_{start,end}
5857 is byte aligned, rather than requiring that start and width are
5858 byte aligned. Drop unnecessary struct keyword from
5859 struct store_immediate_info. Allocate and populate also mask array.
5860 Make start of the arrays relative to bitregion_start rather than
5861 start and size them according to bitregion_{end,start} difference.
5862 (struct imm_store_chain_info): Drop unnecessary struct keyword from
5863 struct store_immediate_info.
5864 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
5865 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
5866 struct keyword from struct store_immediate_info.
5867 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
5868 between stores as long as the surrounding bitregions have no gaps.
5870 (struct split_store): Add orig non-static data member.
5871 (split_store::split_store): Initialize orig to false.
5872 (find_constituent_stmts): Return store_immediate_info *, non-NULL
5873 if there is exactly a single original stmt. Change stmts argument
5874 to pointer from reference, if NULL, don't push anything to it. Add
5875 first argument, use it to optimize skipping over orig stmts that
5876 are known to be before bitpos already. Simplify.
5877 (split_group): Return unsigned int count how many stores are or
5878 would be needed rather than a bool. Add allow_unaligned argument.
5879 Change split_stores argument from reference to pointer, if NULL,
5880 only do a dry run computing how many stores would be produced.
5881 Rewritten algorithm to use both alignment and misalign if
5882 !allow_unaligned and handle bitfield stores with gaps.
5883 (imm_store_chain_info::output_merged_store): Set start_byte_pos
5884 from bitregion_start instead of start. Compute allow_unaligned
5885 here, if true, do 2 split_group dry runs to compute which one
5886 produces fewer stores and prefer aligned if equal. Punt if
5887 new count is bigger or equal than original before emitting any
5888 statements, rather than during that. Remove no longer needed
5889 new_ssa_names tracking. Replace num_stmts with
5890 split_stores.length (). Use 32-bit stack allocated entries
5891 in split_stores auto_vec. Try to reuse original store lhs/rhs1
5892 if possible. Handle bitfields with gaps.
5893 (pass_store_merging::execute): Ignore bitsize == 0 stores.
5894 Compute bitregion_{start,end} for the stores and construct
5895 store_immediate_info with that. Formatting fixes.
5897 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
5900 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
5901 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
5903 2017-10-29 Jim Wilson <wilson@tuliptree.org>
5905 * gimplify.c: Include tm_p.h.
5907 * common.opt (gcoff): Re-add as ignored option.
5908 (gcoff1, gcoff2, gcoff3): Likewise.
5910 * Makefile.in (OBJS): Delete sdbout.o.
5911 (GTFILES): Delete $(srcdir)/sdbout.c.
5912 * debug.h: Delete sdb_debug_hooks.
5913 * final.c: Delete sdbout.h include.
5914 (final_scan_insn): Delete SDB_DEBUG check.
5915 (rest_of_clean_state): Likewise.
5916 * output.h: Delete sdb_begin_function_line.
5919 * toplev.c: Delete sdbout.h include.
5920 (process_options): Delete SDB_DEBUG check.
5921 * tree-core.h (tree_type_common): Delete pointer field of
5923 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
5924 TYPE_SYMTAB_POINTER.
5925 * tree.h (TYPE_SYMTAB_POINTER): Delete.
5926 (TYPE_SYMTAB_IS_POINTER): Delete.
5927 (TYPE_SYMTAB_IS_DIE): Renumber.
5928 * xcoffout.c: Refer to former sdbout.c file.
5929 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
5931 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
5932 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
5933 (Debugging Options): Delete -gcoff.
5934 (-gstabs): Delete SDB reference.
5936 (-gcoff@var{level}): Delete.
5937 * doc/passes.texi (Debugging information output): Delete SDB and
5938 sdbout.c references.
5939 * doc/tm.texi: Regenerate.
5940 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
5941 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
5943 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
5944 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
5945 (SDB_DEBUGGING_INFO): Delete.
5946 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
5947 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
5948 * target.def (output_source_filename): Delete COFF reference.
5950 * common.opt (gcoff): Delete.
5951 (gxcoff+): Update Negative chain.
5952 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
5954 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
5955 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
5956 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
5958 (expand_function_start): Change sdb reference to past tense.
5959 (expand_function_end): Change sdb reference to past tense.
5960 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
5961 * opts.c (debug_type_names): Delete coff entry.
5962 (common_handle_option): Delete OPT_gcoff case.
5963 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
5965 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
5966 * config/cris/cris.h: Delete SDB reference in comment.
5967 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
5968 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
5969 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
5970 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
5972 (ix86_expand_prologue): Likewise.
5973 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
5974 * config/ia64/ia64.h: Likewise.
5975 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
5976 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
5978 * config/mmix/mmix.h: Likewise.
5979 * config/nds32/nds32.c: Likewise.
5980 * config/stormy/storym16.h: Likewise.
5981 * config/visium/visium.h: Likewise.
5982 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
5984 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
5986 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
5987 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
5988 FRAME_POINTER_REGNUM point at high end of local var area.
5990 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
5992 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
5993 Move comment around. Do not reset best_edge for a copiable
5994 destination if the copy would cause a partition change.
5995 (better_edge_p): Remove redundant check.
5997 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5999 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6001 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6003 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6004 for math functions that have _Float<N> and _Float<N>X variants.
6005 (mathfn_built_in_2): Add support for math functions that have
6006 _Float<N> and _Float<N>X variants.
6007 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6008 (expand_builtin_mathfn_ternary): Add support for fma with
6009 _Float<N> and _Float<N>X variants.
6010 (expand_builtin): Likewise.
6011 (fold_builtin_3): Likewise.
6012 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6013 create math function _Float<N> and _Float<N>X variants as external
6015 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6016 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6017 the __builtin_ prefix and if not strict ansi, without the prefix.
6018 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6019 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6020 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6021 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6022 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6023 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6024 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6025 function signatures for fma _Float<N> and _Float<N>X variants.
6026 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6027 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6028 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6029 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6030 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6031 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6032 * gencfn-macros.c (print_case_cfn): Add support for math functions
6033 that have _Float<N> and _Float<N>X variants.
6034 (print_define_operator_list): Likewise.
6035 (fltfn_suffixes): Likewise.
6037 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6038 for math functions that have _Float<N> and _Float<N>X variants.
6039 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6040 and _Float<N>X variants.
6041 (COPYSIGN): Likewise.
6044 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6045 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6047 (integer_valued_read_call_p): Likewise.
6048 * fold-const-call.c (fold_const_call_ss): Likewise.
6049 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6050 _Float<N> and _Float<N>X variants.
6051 (fold_const_call_ssss): Add support for fma _Float<N> and
6052 _Float<N>X variants.
6053 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6054 support for copysign and fma _Float<N> and _Float<N>X variants.
6055 (backprop::process_builtin_call_use): Likewise.
6056 * tree-call-cdce.c (can_test_argument_range); Add support for
6057 sqrt _Float<N> and _Float<N>X variants.
6058 (edom_only_function): Likewise.
6059 (get_no_error_domain): Likewise.
6060 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6061 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6062 copysign _Float<N> and _Float<N>X variants.
6063 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6064 handled by machine independent code.
6065 (FMAF128): Likewise.
6066 * doc/cpp.texi (Common Predefined Macros): Document defining
6067 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6068 fma _Float<N> and _Float<N>X variants.
6070 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6073 * config/i386/i386-modes.def (CCFPU): Remove definition.
6074 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6075 (ix86_cc_modes_compatible): Ditto.
6076 (ix86_expand_carry_flag_compare): Ditto.
6077 (ix86_expand_int_movcc): Ditto.
6078 (ix86_expand_int_addcc): Ditto.
6079 (ix86_reverse_condition): Ditto.
6080 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6081 Return true/false for unordered/ordered fp comparisons.
6082 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6083 (ix86_prepare_fp_compare_args): Update for rename.
6084 (ix86_expand_fp_compare): Update for rename. Generate unordered
6085 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6086 (ix86_expand_sse_compare_and_jump): Ditto.
6087 * config/i386/predicates.md (fcmov_comparison_operator):
6088 Remove CCFPU mode handling.
6089 (ix86_comparison_operator): Ditto.
6090 (ix86_carry_flag_operator): Ditto.
6091 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6092 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6093 (*cmpu<mode>_cc_i387): Ditto.
6094 (FPCMP): Remove mode iterator.
6095 (unord): Remove mode attribute.
6096 (unord_subst): New define_subst transformation
6097 (unord): New define_subst attribute.
6099 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6100 (*cmpi<unord>xf_i387): Ditto.
6101 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6102 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6103 using unord_subst transformation.
6104 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6105 (round_saeonly): Also handle CCFP mode.
6106 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6107 Remove UNSPEC_SAHF unspec handling.
6109 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6111 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6113 2017-10-27 Jeff Law <law@redhat.com>
6115 * gimple-ssa-sprintf.c: Include domwalk.h.
6116 (class sprintf_dom_walker): New class, derived from dom_walker.
6117 (sprintf_dom_walker::before_dom_children): New function.
6118 (struct call_info): Moved into sprintf_dom_walker class
6119 (compute_formath_length, handle_gimple_call): Likewise.
6120 (sprintf_length::execute): Call the dominator walker rather
6121 than walking the statements.
6123 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6124 gimple statement locations.
6125 (check_array_bounds): Corresponding changes. Get the statement's
6126 location directly from wi->stmt.
6128 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6131 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6133 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6135 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6136 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6138 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6141 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6142 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6143 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6144 maybe_get_pool_constant.
6145 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6148 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6150 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6151 2.26 caveat. Update gas and gld versions.
6152 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6155 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6157 * cgraph.h (set_malloc_flag): Declare.
6158 * cgraph.c (set_malloc_flag_1): New function.
6159 (set_malloc_flag): Likewise.
6160 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6161 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6163 (read_ipa_call_summary): Add support for reading is_return_callee.
6164 (write_ipa_call_summary): Stream is_return_callee.
6165 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6166 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6167 ipa-prop.h, ipa-fnsummary.h.
6168 (pure_const_names): Change to static.
6169 (malloc_state_e): Define.
6170 (malloc_state_names): Define.
6171 (funct_state_d): Add field malloc_state.
6172 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6173 (check_retval_uses): New function.
6174 (malloc_candidate_p): Likewise.
6175 (analyze_function): Add support for malloc attribute.
6176 (pure_const_write_summary): Stream malloc_state.
6177 (pure_const_read_summary): Add support for reading malloc_state.
6178 (dump_malloc_lattice): New function.
6179 (propagate_malloc): New function.
6180 (warn_function_malloc): New function.
6181 (ipa_pure_const::execute): Call propagate_malloc and
6182 ipa_free_fn_summary.
6183 (pass_local_pure_const::execute): Add support for malloc attribute.
6184 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6185 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6187 2017-10-27 Martin Liska <mliska@suse.cz>
6189 PR gcov-profile/82457
6190 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6191 for fork-like functions to be properly instrumented.
6193 2017-10-27 Richard Biener <rguenther@suse.de>
6196 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6197 info when we redirected EH.
6199 2017-10-26 Michael Collison <michael.collison@arm.com>
6201 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6203 (<optab>_trunchf<GPI:mode>2: New pattern.
6204 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6205 * config/aarch64/iterators.md (wv): New mode attribute.
6206 (vf, VF): New mode attributes.
6207 (vgp, VGP): New mode attributes.
6208 (s): Update attribute with SImode and DImode prefixes.
6210 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6212 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
6213 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
6214 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
6215 (nios2_option_overide): Initialize it. Don't allow R0-relative
6216 addressing with PIC.
6217 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
6218 (nios2_symbolic_constant_p): Likewise.
6219 (nios2_legitimate_address_p): Likewise.
6220 (nios2_r0rel_section_name_p): New.
6221 (nios2_symbol_ref_in_r0rel_data_p): New.
6222 (nios2_emit_move_sequence): Handle r0rel_constant_p.
6223 (r0rel_constant_p): New.
6224 (nios2_print_operand_address): Handle r0rel_constant_p.
6225 (nios2_cdx_narrow_form_p): Likewise.
6226 * config/nios2/nios2.opt (mr0rel-sec=): New option.
6227 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
6228 (Nios II Options): Document -mr0rel-sec.
6230 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6232 * config/nios2/nios2.c: Include xregex.h.
6233 (nios2_gprel_sec_regex): New.
6234 (nios2_option_overide): Initialize it. Don't allow GP-relative
6235 addressing with PIC.
6236 (nios2_small_section_name_p): Check for regex match.
6237 * config/nios2/nios2.opt (mgprel-sec=): New option.
6238 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
6239 (Nios II Options): Document -mgprel-sec.
6241 2017-10-26 Jim Wilson <wilson@tuliptree.org>
6243 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
6245 2017-10-26 Tom de Vries <tom@codesourcery.com>
6247 PR tree-optimization/82707
6248 * gimple.c (gimple_copy): Fix unsharing of
6249 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
6251 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
6253 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
6254 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
6255 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
6256 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
6257 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
6258 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
6259 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
6260 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
6261 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
6262 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
6263 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
6264 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
6265 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
6266 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
6267 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
6268 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
6269 _mm512_mask_cmpunord_ps_mask): New intrinsics.
6271 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6273 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
6275 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
6276 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
6277 warning to rs6000.c. Remove the Undocumented flag, since it has
6279 (-mabi=ibmlongdouble): Likewise.
6280 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
6281 already set, set the default format for long double.
6282 (rs6000_debug_reg_global): Print whether long double is IBM or
6284 (rs6000_option_override_internal): Rework setting long double
6285 format. Only warn if the user is changing the long double default
6286 and they did not use -Wno-psabi.
6287 * doc/invoke.texi (PowerPC options): Update the documentation for
6288 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
6290 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6291 Alan Hayward <alan.hayward@arm.com>
6292 David Sherwood <david.sherwood@arm.com>
6294 * rtl.h (wider_subreg_mode): New function.
6295 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
6296 rather than an unsigned int *.
6297 * ira-color.c (regno_max_ref_width): Replace with...
6298 (regno_max_ref_mode): ...this new variable.
6299 (coalesced_pseudo_reg_slot_compare): Update accordingly.
6300 Use wider_subreg_mode.
6301 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
6302 rather than an unsigned int *.
6303 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
6304 (process_alt_operands): Likewise.
6305 (invariant_p): Likewise.
6306 * lra-spills.c (assign_mem_slot): Likewise.
6307 (add_pseudo_to_slot): Likewise.
6308 * lra.c (collect_non_operand_hard_regs): Likewise.
6309 (add_regs_to_insn_regno_info): Likewise.
6310 * reload1.c (regno_max_ref_width): Replace with...
6311 (regno_max_ref_mode): ...this new variable.
6312 (reload): Update accordingly. Update call to
6313 ira_sort_regnos_for_alter_reg.
6314 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
6315 (init_eliminable_invariants): Update to use regno_max_ref_mode.
6316 (scan_paradoxical_subregs): Likewise.
6318 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6320 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
6321 (aarch64_frame): Add emit_frame_chain boolean.
6322 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6323 Move eh_return case to aarch64_layout_frame.
6324 (aarch64_layout_frame): Initialize emit_frame_chain.
6325 (aarch64_expand_prologue): Use emit_frame_chain.
6327 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6329 * config/aarch64/aarch64.c (aarch64_layout_frame):
6330 Ensure LR is always stored at the bottom of the callee-saves.
6331 Remove rarely used frame layout which saves callee-saves at top of
6332 frame, so the store of LR can be used as a valid probe in all cases.
6334 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6336 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
6337 Improve unaligned TImode/TFmode base/offset split.
6339 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6340 Alan Hayward <alan.hayward@arm.com>
6341 David Sherwood <david.sherwood@arm.com>
6343 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
6344 * combine.c (find_single_use_1): Likewise.
6345 (expand_field_assignment): Likewise.
6346 (move_deaths): Likewise.
6347 * lra-constraints.c (simplify_operand_subreg): Likewise.
6348 (curr_insn_transform): Likewise.
6349 * lra.c (collect_non_operand_hard_regs): Likewise.
6350 (add_regs_to_insn_regno_info): Likewise.
6351 * rtlanal.c (reg_referenced_p): Likewise.
6352 (covers_regno_no_parallel_p): Likewise.
6354 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6356 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
6357 Don't print any bits outside the precision of the value.
6358 * wide-int.cc (test_printing): Add some new tests.
6360 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6362 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
6363 supports -xbrace_comment option.
6364 * configure: Regenerate.
6365 * config.in: Regenerate.
6366 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
6367 (ASM_CPU_SPEC): Use it.
6369 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6371 * target.def (static_rtx_alignment): New hook.
6372 * targhooks.h (default_static_rtx_alignment): Declare.
6373 * targhooks.c (default_static_rtx_alignment): New function.
6374 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
6375 * doc/tm.texi: Regenerate.
6376 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
6377 instead of targetm.constant_alignment. Remove call to
6379 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6380 (cris_preferred_mininum_alignment): New function, split out from...
6381 (cris_constant_alignment): ...here.
6382 (cris_static_rtx_alignment): New function.
6383 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
6385 (ix86_constant_alignment): ...here.
6386 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6387 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6388 (mmix_static_rtx_alignment): New function.
6389 * config/spu/spu.c (spu_static_rtx_alignment): New function.
6390 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6392 2017-10-26 Tamar Christina <tamar.christina@arm.com>
6395 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
6396 Add flag_trapping_math and flag_fp_int_builtin_inexact.
6398 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
6400 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
6401 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
6402 mark as a sign-extending load.
6403 (local_pic_load_u): Define.
6405 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
6408 * fold-const.c (operand_equal_for_comparison_p): Also return true
6409 if ARG0 is a simple variant of ARG1 with narrower precision.
6410 (fold_ternary_loc): Always pass unstripped operands to the predicate.
6412 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
6414 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
6416 * i386.h (processor_costs): Add gather_static, gather_per_elt,
6417 scatter_static, scatter_per_elt.
6418 * x86-tune-costs.h: Add new cost entries.
6420 2017-10-25 Richard Biener <rguenther@suse.de>
6422 * tree-ssa-sccvn.h (vn_eliminate): Declare.
6423 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
6424 class pass_fre): Move to ...
6425 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
6426 class pass_fre): ... here and adjust for statistics.
6428 2017-10-25 Jakub Jelinek <jakub@redhat.com>
6431 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
6432 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
6433 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
6434 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
6437 2017-10-25 Richard Biener <rguenther@suse.de>
6439 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
6440 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
6441 eliminate_push_avail, eliminate_insert): Move inside...
6442 (class eliminate_dom_walker): ... this class in preparation
6444 (fini_eliminate): Remove by merging with ...
6445 (eliminate): ... this function. Adjust for class changes.
6446 (pass_pre::execute): Remove fini_eliminate call.
6447 (pass_fre::execute): Likewise.
6449 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6452 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
6453 (VPERMI2, VPERMI2I): New mode iterators.
6454 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
6455 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
6457 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
6458 mode iterator. Remove 3 old define_insn patterns.
6459 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
6460 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
6461 VPERMI2 mode iterator, remove the other two expanders.
6462 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
6463 to use VPERMI2 mode iterator, add another alternative for vpermi2*
6464 instructions, remove the other two patterns.
6465 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
6466 mode iterator, remove the other two patterns.
6467 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
6468 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
6469 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
6470 and adjust argument order accordingly.
6471 (ix86_expand_vec_perm): Adjust caller.
6472 (expand_vec_perm_1): Likewise.
6473 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
6474 (expand_vec_perm_vpermt2_vpshub2): ... this.
6475 (ix86_expand_vec_perm_const_1): Adjust caller.
6476 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
6479 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
6480 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
6481 (vec_shl_<mode>): Remove unused expander.
6482 (avx512bw_<shift_insn><mode>3): New define_insn.
6483 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
6484 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
6486 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
6489 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
6492 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6494 PR rtl-optimization/82396
6495 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
6496 (autopref_multipass_init): Simplify initialization.
6497 (autopref_rank_data): Simplify sort order.
6498 * gcc/sched-int.h (autopref_multipass_data_): Remove
6499 multi_mem_insn_p, min_offset and max_offset.
6501 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6504 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6505 Check special value of flag_omit_frame_pointer.
6506 (aarch64_can_eliminate): Likewise.
6507 (aarch64_override_options_after_change_1): Simplify handling of
6508 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
6510 2017-10-24 Richard Biener <rguenther@suse.de>
6512 PR tree-optimization/82697
6513 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
6514 zero for conditional load and unconditional store.
6516 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6518 * doc/install.texi: Document bootstrap-cet.
6520 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6523 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
6524 ENDBR instruction at function entrance if function is only
6527 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6530 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
6531 patterns to better describe from which operation the CF is computed.
6532 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
6533 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
6534 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
6535 is 0, use _0 suffixed expanders instead of emitting a comparison
6538 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
6540 * config/i386/i386.md(*movsf_internal, *movdf_internal):
6541 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
6543 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
6546 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
6547 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
6548 * loop-iv.c (iv_get_reaching_def): Likewise.
6549 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
6550 variable is promoted and the partition contains undefined values.
6552 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6554 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
6556 (nios2_address_cost): Define.
6557 (nios2_legitimize_address): Recognize (exp + constant) directly.
6558 (TARGET_ADDRESS_COST): Define.
6560 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6562 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
6563 (nios2_symbolic_memory_operand_p): Declare.
6564 (nios2_split_large_constant): Declare.
6565 (nios2_split_symbolic_memory_operand): Declare.
6566 * config/nios2/nios2.c: Adjust includes.
6567 (nios2_symbolic_constant_allowed): New.
6568 (nios2_symbolic_constant_p): New.
6569 (nios2_plus_symbolic_constant_p): New.
6570 (nios2_valid_addr_expr_p): Recognize addresses involving
6572 (nios2_legitimate_address_p): Likewise, also LO_SUM.
6573 (nios2_symbolic_memory_operand_p): New.
6574 (nios2_large_constant_p): New.
6575 (nios2_split_large_constant): New.
6576 (nios2_split_plus_large_constant): New.
6577 (nios2_split_symbolic_memory_operand): New.
6578 (nios2_legitimize_address): Code refactoring. Handle addresses
6579 involving symbolic constants.
6580 (nios2_emit_move_sequence): Likewise.
6581 (nios2_print_operand): Improve error output.
6582 (nios2_print_operand_address): Handle LO_SUM.
6583 (nios2_cdx_narrow_form_p): Likewise.
6584 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
6585 operands involving symbolic constants.
6586 (movhi_internal, movsi_internal): Likewise.
6587 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
6588 (extendhisi2, extendqi<mode>2): Likewise.
6590 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6592 * tree-pass.h (PROP_rtl_split_insns): Define.
6593 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
6595 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6597 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
6599 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6602 * target.def (const_not_ok_for_debug_p): Default to
6603 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
6604 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
6605 * targhooks.c (default_const_not_ok_for_debug_p): New function.
6606 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
6607 which targetm.const_not_ok_for_debug_p returned true.
6608 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
6610 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
6612 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
6613 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
6614 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
6616 (ix86_const_not_ok_for_debug_p): New function.
6617 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
6618 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
6620 2017-10-23 David Malcolm <dmalcolm@redhat.com>
6623 * system.h: Conditionally include "unique-ptr.h" if
6624 INCLUDE_UNIQUE_PTR is defined.
6625 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
6626 of defining INCLUDE_UNIQUE_PTR before including "system.h".
6628 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
6630 * config/rl78/rl78.md: New define_expand "subdi3".
6632 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
6635 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
6636 DF_REF_INSN if DF_REF_INSN_INFO is false.
6638 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
6640 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
6641 xmm_move instead of sse_move.
6642 (sse_store_index): New function.
6643 (ix86_register_move_cost): Be more sensible about mismatch stall;
6644 model AVX moves correctly; make difference between sse->integer and
6646 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
6647 moves; make difference between SSE and AVX.
6648 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
6649 and zmm_move. Increase size of sse load and store tables;
6650 add unaligned load and store tables; add ssemmx_to_integer.
6651 * x86-tune-costs.h: Update all entries according to real
6652 move latencies from Agner Fog's manual and chip documentation.
6654 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6657 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
6658 * config/i386/constraints.md (Wf): New constraint.
6659 * config/i386/i386.md (UNSPEC_SBB): New unspec.
6660 (cmp<dwi>_doubleword): Removed.
6661 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
6662 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
6663 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
6664 expand with cmp<dwi>_doubleword. For LTU and GEU use
6665 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
6667 * common.opt (gcolumn-info): Enable by default.
6668 * doc/invoke.texi (gcolumn-info): Document new default.
6670 2017-10-23 Richard Biener <rguenther@suse.de>
6672 PR tree-optimization/82672
6673 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
6674 Fold the stmt if we propagated into it.
6676 2017-10-23 Richard Biener <rguenther@suse.de>
6678 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
6679 (bitmap_remove_expr_from_set): ... this. All callers call this
6680 for non-constant values.
6681 (bitmap_set_subtract): Rename to...
6682 (bitmap_set_subtract_expressions): ... this. Adjust and
6684 (bitmap_set_contains_value): Remove superfluous check.
6685 (bitmap_set_replace_value): Inline into single caller ...
6686 (bitmap_value_replace_in_set): ... here and simplify.
6687 (dependent_clean): Merge into ...
6688 (clean): ... this using an overload. Adjust.
6689 (prune_clobbered_mems): Adjust.
6690 (compute_antic_aux): Likewise.
6691 (compute_partial_antic_aux): Likewise.
6693 2017-10-23 Richard Biener <rguenther@suse.de>
6695 PR tree-optimization/82129
6697 2017-08-01 Richard Biener <rguenther@suse.de>
6699 PR tree-optimization/81181
6700 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6701 (compute_antic): ... end of iteration here.
6703 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6705 * target.def (starting_frame_offset): New hook.
6706 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
6707 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
6708 * doc/tm.texi.in: Regenerate.
6709 * hooks.h (hook_hwi_void_0): Declare.
6710 * hooks.c (hook_hwi_void_0): New function.
6711 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
6712 STARTING_FRAME_OFFSET.
6713 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
6714 * reload1.c (reload): Likewise.
6715 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
6716 instead of STARTING_FRAME_OFFSET.
6717 * function.c (try_fit_stack_local): Likewise.
6718 (assign_stack_local_1): Likewise
6719 (instantiate_virtual_regs): Likewise.
6720 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
6721 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
6722 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
6723 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
6724 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
6725 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
6726 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
6727 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
6728 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
6729 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
6730 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
6731 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
6732 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
6733 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
6734 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
6735 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
6736 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
6737 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
6738 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
6739 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
6740 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
6741 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
6742 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
6743 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
6744 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
6745 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
6746 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
6747 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
6748 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
6749 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
6750 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
6751 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
6752 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
6753 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
6754 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
6755 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
6756 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
6757 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
6758 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
6759 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
6760 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
6761 * config/avr/avr.c (avr_starting_frame_offset): Make static and
6762 return a HOST_WIDE_INT.
6763 (avr_builtin_setjmp_frame_value): Use it instead of
6764 STARTING_FRAME_OFFSET.
6765 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6766 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
6767 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
6769 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6770 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
6771 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
6772 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6773 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
6774 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
6775 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6776 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
6777 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
6778 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6779 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
6780 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
6782 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6783 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
6784 * config/mips/mips.c (mips_compute_frame_info): Refer to
6785 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6786 (mips_starting_frame_offset): New function.
6787 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6788 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
6789 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
6790 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
6791 and return a HOST_WIDE_INT.
6792 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6793 (mmix_initial_elimination_offset): Refer to
6794 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6795 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
6796 * config/pa/pa.c (pa_starting_frame_offset): New function.
6797 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
6798 (pa_expand_prologue): Likewise.
6799 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6800 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
6801 !FRAME_GROWS_DOWNWARD handling to...
6802 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6803 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
6804 !FRAME_GROWS_DOWNWARD handling to...
6805 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6806 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
6807 !FRAME_GROWS_DOWNWARD handling to...
6808 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6809 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
6811 (rs6000_starting_frame_offset): New function.
6812 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
6813 !FRAME_GROWS_DOWNWARD handling to...
6814 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6815 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
6816 !FRAME_GROWS_DOWNWARD handling to...
6817 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6818 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
6819 !FRAME_GROWS_DOWNWARD handling to...
6820 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6821 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
6822 (rs6000_starting_frame_offset): New function.
6823 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
6824 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
6825 * config/vax/vax.c (vax_starting_frame_offset): New function.
6826 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
6827 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6828 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
6829 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
6830 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6831 * system.h (STARTING_FRAME_OFFSET): Poison.
6833 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6835 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
6836 SCALAR_TYPE_MODE instead of TYPE_MODE.
6838 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6839 Alan Hayward <alan.hayward@arm.com>
6840 David Sherwood <david.sherwood@arm.com>
6842 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
6844 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6845 Alan Hayward <alan.hayward@arm.com>
6846 David Sherwood <david.sherwood@arm.com>
6848 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
6850 2017-10-23 Richard Biener <rguenther@suse.de>
6852 PR tree-optimization/82129
6853 * tree-ssa-pre.c (bitmap_set_and): Remove.
6854 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
6855 canonicalizing expressions in the set to those with lowest
6856 ID rather than taking that from the first edge.
6858 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6860 * combine.c (rtx_equal_for_field_assignment_p): Use
6861 byte_lowpart_offset.
6863 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6864 Alan Hayward <alan.hayward@arm.com>
6865 David Sherwood <david.sherwood@arm.com>
6867 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
6868 to a SUBREG_PROMOTED_VAR.
6870 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6871 Alan Hayward <alan.hayward@arm.com>
6872 David Sherwood <david.sherwood@arm.com>
6874 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
6875 (expand_debug_source_expr): Likewise.
6876 * combine.c (combine_simplify_rtx): Likewise.
6877 * cse.c (fold_rtx): Likewise.
6878 * optabs.c (expand_float): Likewise.
6879 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6880 (simplify_binary_operation_1): Likewise.
6882 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6883 Alan Hayward <alan.hayward@arm.com>
6884 David Sherwood <david.sherwood@arm.com>
6886 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
6887 (record_promoted_value): Likewise.
6888 * expr.c (expand_expr_real_2): Likewise.
6889 * ree.c (update_reg_equal_equiv_notes): Likewise.
6890 (combine_set_extension): Likewise.
6891 * rtlanal.c (low_bitmask_len): Likewise.
6892 * simplify-rtx.c (neg_const_int): Likewise.
6893 (simplify_binary_operation_1): Likewise.
6895 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6896 Alan Hayward <alan.hayward@arm.com>
6897 David Sherwood <david.sherwood@arm.com>
6899 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
6900 * regcprop.c (maybe_mode_change): Likewise.
6901 * reload1.c (alter_reg): Likewise.
6903 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6905 * inchash.h (inchash::hash::add_wide_int): New function.
6906 * lto-streamer-out.c (hash_tree): Use it.
6908 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6910 * inchash.h (inchash::hash::add_wide_int): Rename to...
6911 (inchash::hash::add_hwi): ...this.
6912 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
6913 (polymorphic_call_target_hasher::hash): Likewise.
6914 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
6915 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
6916 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
6917 * lto-streamer-out.c (hash_tree): Likewise.
6918 * optc-save-gen.awk: Likewise.
6919 * tree.c (add_expr): Likewise.
6921 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6924 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
6925 for ordered inequality comparisons even with TARGET_IEEE_FP.
6927 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6930 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
6931 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6932 Expand with cmp<dwi>_doubleword.
6934 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6936 * extend.texi: Add x86 specific to 'nocf_check' attribute.
6937 List CET intrinsics.
6938 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
6939 specific to -fcf-protection option.
6941 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6943 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
6944 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
6945 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
6946 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
6947 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
6948 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
6949 (extra_objs): Add cet.o for Linux/x86 targets.
6950 (tmake_file): Add i386/t-cet for Linux/x86 targets.
6951 * config/i386/cet.c: New file.
6952 * config/i386/cetintrin.h: Likewise.
6953 * config/i386/t-cet: Likewise.
6954 * config/i386/cpuid.h (bit_SHSTK): New.
6955 (bit_IBT): Likewise.
6956 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
6957 pass IBT and SHSTK bits.
6958 * config/i386/i386-builtin-types.def
6959 (VOID_FTYPE_UNSIGNED_PVOID): New.
6960 (VOID_FTYPE_UINT64_PVOID): Likewise.
6961 * config/i386/i386-builtin.def: Add CET intrinsics.
6962 * config/i386/i386-c.c (ix86_target_macros_internal): Add
6963 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
6964 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
6965 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
6967 * config/i386/i386.c (rest_of_insert_endbranch): New.
6968 (pass_data_insert_endbranch): Likewise.
6969 (pass_insert_endbranch): Likewise.
6970 (make_pass_insert_endbranch): Likewise.
6971 (ix86_notrack_prefixed_insn_p): Likewise.
6972 (ix86_target_string): Add -mibt, -mshstk flags.
6973 (ix86_option_override_internal): Add flag_cf_protection
6975 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
6976 (ix86_print_operand): Add 'notrack' prefix output.
6977 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
6978 (ix86_expand_builtin): Expand CET intrinsics.
6979 (x86_output_mi_thunk): Add 'endbranch' instruction.
6980 * config/i386/i386.h (TARGET_IBT): New.
6981 (TARGET_IBT_P): Likewise.
6982 (TARGET_SHSTK): Likewise.
6983 (TARGET_SHSTK_P): Likewise.
6984 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
6985 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
6986 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
6987 (builtin_setjmp_setup): New pattern.
6988 (builtin_longjmp): Likewise.
6989 (rdssp<mode>): Likewise.
6990 (incssp<mode>): Likewise.
6991 (saveprevssp): Likewise.
6992 (rstorssp): Likewise.
6993 (wrss<mode>): Likewise.
6994 (wruss<mode>): Likewise.
6995 (setssbsy): Likewise.
6996 (clrssbsy): Likewise.
6997 (nop_endbr): Likewise.
6998 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7000 * config/i386/immintrin.h: Include <cetintrin.h>.
7001 * config/i386/linux-common.h
7002 (file_end_indicate_exec_stack_and_cet): New prototype.
7003 (TARGET_ASM_FILE_END): New.
7005 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7007 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7008 latencies instead of having separate table; make difference between
7009 integer and float costs.
7010 * i386.h (processor_costs): Remove scalar_stmt_cost,
7011 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7012 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7014 * x86-tune-costs.h: Remove entries which has been removed in
7015 procesor_costs from all tables; make cond_taken_branch_cost
7016 and cond_not_taken_branch_cost COST_N_INSNS based.
7018 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7020 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7022 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7024 * config/i386/i386.md (isa): Remove fma_avx512f.
7025 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7026 <avx512>_fmadd_<mode>_mask3<round_name>,
7027 <avx512>_fmsub_<mode>_mask<round_name>,
7028 <avx512>_fmsub_<mode>_mask3<round_name>,
7029 <avx512>_fnmadd_<mode>_mask<round_name>,
7030 <avx512>_fnmadd_<mode>_mask3<round_name>,
7031 <avx512>_fnmsub_<mode>_mask<round_name>,
7032 <avx512>_fnmsub_<mode>_mask3<round_name>,
7033 <avx512>_fmaddsub_<mode>_mask<round_name>,
7034 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7035 <avx512>_fmsubadd_<mode>_mask<round_name>,
7036 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7037 (*vec_widen_umult_even_v16si<mask_name>,
7038 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7039 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7041 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7043 * extend.texi: Add 'nocf_check' documentation.
7044 * gimple.texi: Add second parameter to
7045 gimple_build_call_from_tree.
7046 * invoke.texi: Add -fcf-protection documentation.
7047 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7049 2017-10-20 Richard Biener <rguenther@suse.de>
7051 PR tree-optimization/82473
7052 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7053 the largest input type.
7055 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7057 * c-attribs.c (handle_nocf_check_attribute): New function.
7058 (c_common_attribute_table): Add 'nocf_check' handling.
7059 * gimple-parser.c: Add second argument NULL to
7060 gimple_build_call_from_tree.
7061 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7062 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7064 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7065 * common.opt: Add fcf-protection flag.
7066 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7067 * flag-types.h: Add enum cf_protection_level.
7068 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7069 Add 'nocf_check' attribute propagation to gimple call.
7070 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7071 (gimple_build_call_from_tree): Update prototype.
7072 (gimple_call_nocf_check_p): New function.
7073 (gimple_call_set_nocf_check): Likewise.
7074 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7075 * ipa-icf.c: Add nocf_check attribute in statement hash.
7076 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7077 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7078 * toplev.c (process_options): Add flag_cf_protection handling.
7080 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7082 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7084 2017-10-20 Richard Biener <rguenther@suse.de>
7086 PR tree-optimization/82603
7087 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7088 remove false predicated stores.
7090 2017-10-20 Richard Biener <rguenther@suse.de>
7092 * graphite-isl-ast-to-gimple.c
7093 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7094 Remove return value and simplify, dump copied stmt after lhs
7096 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7097 Reduce dump verbosity.
7098 (gsi_insert_earliest): Likewise.
7099 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7100 * graphite.c (print_global_statistics): Adjust dumping.
7101 (print_graphite_scop_statistics): Likewise.
7102 (print_graphite_statistics): Do not dump loops here.
7103 (graphite_transform_loops): But here.
7105 2017-10-20 Nicolas Roche <roche@adacore.com>
7107 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7108 * configure: Regenerate.
7110 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7113 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7114 functions when optimizing replace GIMPLE_RETURN stmts with
7115 calls to __builtin_unreachable ().
7118 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7119 for -fsanitize=thread link of executables.
7120 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7121 link of executables.
7124 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7126 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7127 define_insns for logical vector shifts to use VI248_AVX512BW
7128 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7129 condition, useless isa and prefix attributes. Change the first
7130 2 of these define_insns to ...
7131 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7132 define_insn for avx512vl.
7133 (<shift_insn><mode>3): ... and this, new define_insn without
7134 masking for non-avx512vl.
7137 * config/i386/sse.md (*andnot<mode>3,
7138 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7139 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7140 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7141 not applied use empty suffix even for TARGET_AVX512VL.
7142 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7143 is applied, supply evex,evex or evex,evex,evex instead of just
7146 2017-10-20 Julia Koval <julia.koval@intel.com>
7148 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7149 (OPTION_MASK_ISA_GFNI_UNSET): New.
7150 (ix86_handle_option): Handle OPT_mgfni.
7151 * config/i386/cpuid.h (bit_GFNI): New.
7152 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7153 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7154 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7155 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7156 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7157 * config/i386/i386.opt: Add mgfni.
7159 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7161 * config/msp430/msp430.c (msp430_option_override): Disable
7162 -fdelete-null-pointer-checks.
7163 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7165 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7167 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7168 of x87 and SSE instructions.
7170 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7172 * asan.c (create_cond_insert_point): Do not update edge count.
7173 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7174 (afdo_propagate_circuit): Likewise.
7175 (afdo_calculate_branch_prob): Likewise.
7176 (afdo_annotate_cfg): Likewise.
7177 * basic-block.h (struct edge_def): Remove count.
7178 (edge_def::count): New accessor.
7179 * bb-reorder.c (rotate_loop): Update.
7180 (find_traces_1_round): Update.
7181 (connect_traces): Update.
7182 (sanitize_hot_paths): Update.
7183 * cfg.c (unchecked_make_edge): Update.
7184 (make_single_succ_edge): Update.
7185 (check_bb_profile): Update.
7186 (dump_edge_info): Update.
7187 (update_bb_profile_for_threading): Update.
7188 (scale_bbs_frequencies_int): Update.
7189 (scale_bbs_frequencies_gcov_type): Update.
7190 (scale_bbs_frequencies_profile_count): Update.
7191 (scale_bbs_frequencies): Update.
7192 * cfganal.c (connect_infinite_loops_to_exit): Update.
7193 * cfgbuild.c (compute_outgoing_frequencies): Update.
7194 (find_many_sub_basic_blocks): Update.
7195 * cfgcleanup.c (try_forward_edges): Update.
7196 (try_crossjump_to_edge): Update
7197 * cfgexpand.c (expand_gimple_cond): Update
7198 (expand_gimple_tailcall): Update
7199 (construct_exit_block): Update
7200 * cfghooks.c (verify_flow_info): Update
7201 (redirect_edge_succ_nodup): Update
7202 (split_edge): Update
7203 (make_forwarder_block): Update
7204 (duplicate_block): Update
7205 (account_profile_record): Update
7206 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7207 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7208 * cfgloopmanip.c (scale_loop_profile): Update.
7210 (lv_adjust_loop_entry_edge): Update.
7211 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7212 (force_nonfallthru_and_redirect): Update.
7213 (purge_dead_edges): Update.
7214 (rtl_flow_call_edges_add): Update.
7215 * cgraphunit.c (init_lowered_empty_function): Update.
7216 (cgraph_node::expand_thunk): Update.
7217 * gimple-pretty-print.c (dump_probability): Update.
7218 (dump_edge_probability): Update.
7219 * gimple-ssa-isolate-paths.c (isolate_path): Update.
7220 * haifa-sched.c (sched_create_recovery_edges): Update.
7221 * hsa-gen.c (convert_switch_statements): Update.
7222 * ifcvt.c (dead_or_predicable): Update.
7223 * ipa-inline-transform.c (inline_transform): Update.
7224 * ipa-split.c (split_function): Update.
7225 * ipa-utils.c (ipa_merge_profiles): Update.
7226 * loop-doloop.c (add_test): Update.
7227 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
7228 * lto-streamer-in.c (input_cfg): Update.
7229 (input_function): Update.
7230 * lto-streamer-out.c (output_cfg): Update.
7231 * modulo-sched.c (sms_schedule): Update.
7232 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7233 * predict.c (maybe_hot_edge_p): Update.
7234 (unlikely_executed_edge_p): Update.
7235 (probably_never_executed_edge_p): Update.
7236 (dump_prediction): Update.
7237 (drop_profile): Update.
7238 (propagate_unlikely_bbs_forward): Update.
7239 (determine_unlikely_bbs): Update.
7240 (force_edge_cold): Update.
7241 * profile.c (compute_branch_probabilities): Update.
7242 * reg-stack.c (better_edge): Update.
7243 * shrink-wrap.c (handle_simple_exit): Update.
7244 * tracer.c (better_p): Update.
7245 * trans-mem.c (expand_transaction): Update.
7246 (split_bb_make_tm_edge): Update.
7247 * tree-call-cdce.c: Update.
7248 * tree-cfg.c (gimple_find_sub_bbs): Update.
7249 (gimple_split_edge): Update.
7250 (gimple_duplicate_sese_region): Update.
7251 (gimple_duplicate_sese_tail): Update.
7252 (gimple_flow_call_edges_add): Update.
7253 (insert_cond_bb): Update.
7254 (execute_fixup_cfg): Update.
7255 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
7256 * tree-complex.c (expand_complex_div_wide): Update.
7257 * tree-eh.c (lower_resx): Update.
7258 (unsplit_eh): Update.
7259 (cleanup_empty_eh_move_lp): Update.
7260 * tree-inline.c (copy_edges_for_bb): Update.
7261 (freqs_to_counts): Update.
7262 (copy_cfg_body): Update.
7263 * tree-ssa-dce.c (remove_dead_stmt): Update.
7264 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7265 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7266 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7267 (unloop_loops): Update.
7268 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7269 * tree-ssa-loop-split.c (connect_loops): Update.
7270 (split_loop): Update.
7271 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7272 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7273 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7274 * tree-ssa-reassoc.c (branch_fixup): Update.
7275 * tree-ssa-tail-merge.c (replace_block_by): Update.
7276 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7277 (compute_path_counts): Update.
7278 (update_profile): Update.
7279 (recompute_probabilities): Update.
7280 (update_joiner_offpath_counts): Update.
7281 (estimated_freqs_path): Update.
7282 (freqs_to_counts_path): Update.
7283 (clear_counts_path): Update.
7284 (ssa_fix_duplicate_block_edges): Update.
7285 (duplicate_thread_path): Update.
7286 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7287 (case_bit_test_cmp): Update.
7288 (collect_switch_conv_info): Update.
7289 (gen_inbound_check): Update.
7290 (do_jump_if_equal): Update.
7291 (emit_cmp_and_jump_insns): Update.
7292 * tree-tailcall.c (decrease_profile): Update.
7293 (eliminate_tail_call): Update.
7294 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7295 (vect_do_peeling): Update.
7296 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7297 * ubsan.c (ubsan_expand_null_ifn): Update.
7298 (ubsan_expand_ptr_ifn): Update.
7299 * value-prof.c (gimple_divmod_fixed_value): Update.
7300 (gimple_mod_pow2): Update.
7301 (gimple_mod_subtract): Update.
7302 (gimple_ic): Update.
7303 (gimple_stringop_fixed_value): Update.
7305 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7308 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
7310 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
7312 PR rtl-optimization/82395
7313 * ira-color.c (allocno_priority_compare_func): Fix comparison step
7314 based on non_spilled_static_chain_regno_p.
7316 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7318 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
7319 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
7320 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
7322 2017-10-19 Martin Sebor <msebor@redhat.com>
7324 PR tree-optimization/82596
7325 * tree.c (array_at_struct_end_p): Handle STRING_CST.
7327 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7329 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
7330 (get_mem_refs_of_builtin_call): Likewise.
7331 * builtins.c (expand_builtin_apply): Adjust call to
7332 allocate_dynamic_stack_space.
7333 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
7334 the third argument to allocate_dynamic_stack_space, otherwise -1.
7335 (expand_builtin): Deal with all alloca variants.
7336 (is_inexpensive_builtin): Likewise.
7337 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
7338 * calls.c (special_function_p): Deal with all alloca variants.
7339 (initialize_argument_information): Adjust call to
7340 allocate_dynamic_stack_space.
7341 (expand_call): Likewise.
7342 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
7343 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
7344 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
7345 use it for the stack usage computation.
7346 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
7347 * function.c (gimplify_parameters): Call build_alloca_call_expr.
7348 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
7349 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
7350 (in_loop_p): Remove first argument and useless check.
7351 (pass_walloca::execute): Remove useless test and adjust call to above.
7352 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
7353 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
7354 (gimplify_call_expr): Deal with all alloca variants.
7355 * hsa-gen.c (gen_hsa_alloca): Likewise.
7356 (gen_hsa_insns_for_call): Likewise.
7357 * ipa-pure-const.c (special_builtin_state): Likewise.
7358 * tree-chkp.c (chkp_build_returned_bound): Likewise.
7359 * tree-object-size.c (alloc_object_size): Likewise.
7360 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
7361 (call_may_clobber_ref_p_1): Likewise.
7362 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
7363 (ccp_fold_stmt): Likewise.
7364 (optimize_stack_restore): Likewise.
7365 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
7366 (mark_all_reaching_defs_necessary_1): Likewise.
7367 (propagate_necessity): Likewise.
7368 (eliminate_unnecessary_stmts): Likewise.
7369 * tree.c (build_common_builtin_nodes): Build
7370 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
7371 (build_alloca_call_expr): New function.
7372 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
7373 (CASE_BUILT_IN_ALLOCA): Likewise.
7374 (build_alloca_call_expr): Declare.
7375 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
7377 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7380 * dwarf2out.c (new_die_raw): New static inline function.
7381 (new_die): Use it to create the DIE.
7382 (add_AT_external_die_ref): Likewise.
7383 (clone_die): Likewise.
7384 (clone_as_declaration): Likewise.
7385 (dwarf2out_vms_debug_main_pointer): Likewise.
7386 (base_type_die): Likewise. Remove early return for corner cases.
7387 Do not call add_pubtype on the DIE here.
7388 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
7389 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
7390 typedefs for base types with DW_AT_endianity. Make sure a DIE with
7391 native order exists for base types, attach the DIE manually and call
7392 add_pubtype on it. Do not equate a reverse order DIE to the type.
7394 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
7396 * config/arm/arm.c (align_ok_ldrd_strd): New function.
7397 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
7399 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
7401 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7403 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
7404 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
7405 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
7406 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
7407 * opts.c (sanitizer_opts): Add builtin.
7408 * ubsan.c (instrument_builtin): New function.
7409 (pass_ubsan::execute): Call it.
7410 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
7411 * doc/invoke.texi: Document -fsanitize=builtin.
7413 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
7414 builtins, store max (log2 (align), 0) into uchar field instead of
7415 align into uptr field.
7416 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
7417 store uchar 0 field instead of uptr 0 field.
7418 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
7419 instead of passing one address of struct with 2 locations pass
7420 two addresses of structs with 1 location each.
7421 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
7422 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
7423 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
7424 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
7425 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
7426 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
7427 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
7428 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
7430 2017-10-19 Martin Liska <mliska@suse.cz>
7433 * file-find.c (remove_prefix): Remove.
7434 * file-find.h (remove_prefix): Likewise.
7435 * gcc-ar.c: Remove smartness of lookup.
7437 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
7439 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
7440 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
7441 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
7443 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7446 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
7447 (setcc + and to xor + setcc): New peephole2.
7449 2017-10-19 Tom de Vries <tom@codesourcery.com>
7451 * doc/sourcebuild.texi (Test Directives, Variants of
7452 dg-require-support): Add dg-require-stack-size.
7454 2017-10-19 Martin Liska <mliska@suse.cz>
7457 * gimplify.c (gimplify_decl_expr): Do not instrument variables
7458 that have a large alignment.
7459 (gimplify_target_expr): Likewise.
7461 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
7463 PR rtl-optimization/82602
7464 * ira.c (rtx_moveable_p): Return false for volatile asm.
7466 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
7469 * config/i386/i386-modes.def (CCGZ): New CC mode.
7470 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
7471 * config/i386/predicates.md (ix86_comparison_operator):
7473 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7474 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
7475 with double-word subtraction.
7476 (put_condition_code): Handle CCGZmode.
7478 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
7480 * wide-int.cc (debug (const wide_int &)): New.
7481 (debug (const wide_int *)): New.
7482 (debug (const widest_int &)): New.
7483 (debug (const widest_int *)): New.
7485 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
7488 * lra-constraints.c (curr_insn_transform): Use non-input operand
7489 instead of output one for matched reload.
7491 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7493 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
7494 (tree-ssa-loop-ivopts.h): New header file.
7495 (struct builtin_info): New fields.
7496 (classify_builtin_1): Compute and record base and offset parts for
7497 memset builtin partition by calling strip_offset.
7498 (offset_cmp, fuse_memset_builtins): New functions.
7499 (finalize_partitions): Fuse adjacent memset partitions by calling
7501 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
7502 Expose the interface.
7503 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
7505 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7507 PR tree-optimization/82574
7508 * tree-loop-distribution.c (find_single_drs): New parameter. Check
7509 that data reference must be executed exactly once per iteration
7510 against the outermost loop in nest.
7511 (classify_partition): Update call to above function.
7513 2017-10-18 Richard Biener <rguenther@suse.de>
7515 PR tree-optimization/82591
7516 * graphite.c (graphite_transform_loops): Move code gen message
7518 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7519 Here. Handle scop_to_isl_ast failing.
7520 (scop_to_isl_ast): Limit the number of ISL operations.
7522 2017-10-18 Richard Biener <rguenther@suse.de>
7524 * graphite-isl-ast-to-gimple.c
7525 (translate_isl_ast_to_gimple::set_rename): Simplify.
7526 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
7527 (graphite_copy_stmts_from_block): ... here.
7528 (copy_bb_and_scalar_dependences): Simplify.
7529 (add_parameters_to_ivs_params): Canonicalize.
7530 (generate_entry_out_of_ssa_copies): Simplify.
7531 * graphite-sese-to-poly.c (extract_affine_name): Simplify
7532 by passing in ISL dimension.
7533 (parameter_index_in_region_1): Rename to ...
7534 (parameter_index_in_region): ... this.
7535 (extract_affine): Adjust assert, pass down parameter index.
7536 (add_param_constraints): Use range-info when available.
7537 (build_scop_context): Adjust.
7538 * sese.c (new_sese_info): Adjust.
7539 (free_sese_info): Likewise.
7540 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
7541 Remove unused typedefs.
7542 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
7544 2017-10-18 Martin Liska <mliska@suse.cz>
7546 * combine.c (simplify_compare_const): Add gcc_fallthrough.
7548 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7550 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
7551 (s390_sched_init): Do not reset s390_sched_state if we entered the
7552 current basic block via a fallthru edge and all others are unlikely.
7554 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7556 * config/s390/s390.c (NUM_SIDES): New variable.
7557 (LONGRUNNING_THRESHOLD): New variable.
7558 (LATENCY_FACTOR): New variable.
7559 (s390_sched_score): Decrease score for long-running instructions on
7561 (s390_sched_variable_issue): Perform bookkeeping for long-running
7564 2017-10-18 Richard Biener <rguenther@suse.de>
7566 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
7567 Simplify with removal of the parameter rename map.
7568 (set_rename): Likewise.
7569 (should_copy_to_new_region): Likewise.
7570 (graphite_copy_stmts_from_block): Likewise.
7571 (copy_bb_and_scalar_dependences): Remove initialization of
7572 unused copied_bb_map.
7574 (copy_internal_parameters): Likewise.
7575 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
7576 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
7577 Use INTEGRAL_TYPE_P.
7578 (parameter_index_in_region_1): Rename to ...
7579 (assign_parameter_index_in_region): ... this. Assert we have
7580 a parameter we handle.
7581 (scan_tree_for_params): Adjust.
7582 * sese.h (parameter_rename_map_t): Remove.
7583 (struct sese_info_t): Remove unused parameter_rename_map and
7584 copied_bb_map members.
7585 * sese.c (new_sese_info): Adjust.
7586 (free_sese_info): Likewise.
7588 2017-10-18 Martin Liska <mliska@suse.cz>
7591 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
7592 on an abnormal edge.
7594 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
7596 * doc/invoke.texi (ffunction-sections and fdata-sections):
7599 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7601 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
7602 the use statement can throw internally.
7604 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7606 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
7607 any RTX present on the RHS of a SET.
7608 * compare-elim.c (try_eliminate_compare): Restore comment.
7610 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7612 * langhooks.h (struct lang_hooks): Document that tree_size langhook
7613 may be also called on tcc_type nodes.
7614 * langhooks.c (lhd_tree_size): Likewise.
7616 2017-10-17 David Malcolm <dmalcolm@redhat.com>
7618 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
7619 format_warning_at_substring.
7620 (maybe_warn): Convert source_range * param to a location_t. Pass
7621 UNKNOWN_LOCATION rather than NULL to fmtwarn.
7622 (format_directive): Remove code to extract source_ranges and
7623 source_range * in favor of just a location_t.
7624 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
7626 * substring-locations.c (format_warning_va): Convert
7627 source_range * param to a location_t.
7628 (format_warning_at_substring): Likewise.
7629 * substring-locations.h (format_warning_va): Likewise.
7630 (format_warning_at_substring): Likewise.
7632 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
7634 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
7636 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
7637 and scatter/gather ops.
7639 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
7640 vec_gather_load and vec_scatter_store.
7641 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
7642 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
7643 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
7644 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
7645 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
7646 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
7648 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
7650 * reg-stack.c (compare_for_stack_reg): Add bool argument.
7651 Detect FTST instruction and handle its register pops. Only pop
7652 second operand if can_pop_second_op is true.
7653 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
7654 set can_pop_second_op to false in the compare_for_stack_reg call.
7656 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
7657 output_fp_compare for stack register operands.
7658 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
7659 instructions here. Do not emit stack register pops here. Assert
7660 that FCOMPP pops next to top stack register. Rewrite function.
7662 2017-10-17 Nathan Sidwell <nathan@acm.org>
7665 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
7666 use DECL_ASSEMBLER_NAME_RAW.
7669 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
7672 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
7673 Wilco Dijkstra <wilco.dijkstra@arm.com>
7675 * builtins.c (expand_builtin_update_setjmp_buf): Add a
7676 converstion to Pmode from the buf_addr.
7678 2017-10-17 Richard Biener <rguenther@suse.de>
7680 * graphite-dependences.c (scop_get_reads_and_writes): Change
7681 output parameters to references.
7683 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
7685 PR 71026/tree-optimization
7686 * fold-const.c (distribute_real_division): Removed.
7687 (fold_binary_loc): Remove calls to distribute_real_divison.
7689 2017-10-17 Richard Biener <rguenther@suse.de>
7691 * graphite-scop-detection.c
7692 (scop_detection::stmt_has_simple_data_refs_p): Always use
7693 the full nest as region.
7694 (try_generate_gimple_bb): Likewise.
7695 * sese.c (scalar_evolution_in_region): Simplify now that
7696 SCEV can handle instantiation in regions.
7697 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
7698 in the non-loop part of a function if requested.
7700 2017-10-17 Richard Biener <rguenther@suse.de>
7702 PR tree-optimization/82563
7703 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
7705 (graphite_regenerate_ast_isl): Call it.
7706 * graphite-scop-detection.c (build_scops): Remove entry edge split.
7708 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7710 PR tree-optimization/82549
7711 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
7712 Formatting fixes. Instead of calling make_bit_field_ref with negative
7715 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
7717 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
7718 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
7719 _mm_maskz_reduce_ss): New.
7720 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
7721 __builtin_ia32_reducess_mask): Ditto..
7722 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
7723 * config/i386/sse.md (reduces<mode>): Renamed to ...
7724 (reduces<mode><mask_scalar_name>): ... this.
7725 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
7727 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
7728 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
7730 2017-10-16 David Malcolm <dmalcolm@redhat.com>
7732 * Makefile.in (OBJS): Add unique-ptr-tests.o.
7733 * selftest-run-tests.c (selftest::run_tests): Call
7734 selftest::unique_ptr_tests_cc_tests.
7735 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
7736 * unique-ptr-tests.cc: New file.
7738 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
7741 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7743 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7744 (make_hard_regno_born, make_hard_regno_dead): Update
7745 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
7747 2017-10-16 Jeff Law <law@redhat.com>
7749 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7751 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
7753 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
7755 2017-10-16 Olivier Hainque <hainque@adacore.com>
7757 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
7758 with_cpu if we were configured for an e500v2 target cpu name.
7760 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
7762 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
7763 * doc/invoke.texi: Document +nodsp as a valid extension for
7766 2017-10-16 Martin Liska <mliska@suse.cz>
7768 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
7769 (test_set_range): Likewise.
7770 (test_range_functions): Rename to ...
7771 (test_bit_in_range): ... this.
7772 (sbitmap_c_tests): Add new test.
7774 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7776 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
7778 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
7779 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
7781 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7783 * config/aarch64/aarch64-builtins.c
7784 (aarch64_types_quadopu_lane_qualifiers): New.
7785 (TYPES_QUADOPU_LANE): New.
7786 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
7787 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
7788 (aarch64_<sur>dot_laneq<vsi2qi>): New.
7789 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
7790 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
7791 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
7792 (Vdottype, DOTPROD): New.
7793 (sur): Add SDOT and UDOT.
7795 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7797 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
7798 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
7799 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
7801 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
7802 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
7803 Enable TARGET_DOTPROD.
7804 (cortex-a75.cortex-a55): Likewise.
7805 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
7807 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7809 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
7810 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
7812 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
7814 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
7815 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
7816 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
7817 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
7818 * config/arm/types.md (neon_dot, neon_dot_q): New.
7819 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
7821 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7823 * config/arm/arm.h (TARGET_DOTPROD): New.
7824 * config/arm/arm.c (arm_arch_dotprod): New.
7825 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
7826 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
7827 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
7828 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
7829 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
7830 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
7831 * doc/invoke.texi (armv8.2-a): Document dotprod
7833 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7835 * i386.c (ix86_vec_cost): New function.
7836 (ix86_rtx_costs): Handle vector operations better.
7837 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
7838 * x86-tune-costs.h: Add new costs to all tables.
7840 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7842 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
7844 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
7845 divsd, sqrtss and sqrtsd
7846 * x86-tune-costs.h: Add new entries to all costs.
7847 (znver1_cost): Fix to match real instruction latencies.
7849 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7850 Michael Collison <michael.collison@arm.com>
7852 * compare-elim.c: Include emit-rtl.h.
7853 (can_merge_compare_into_arith): New function.
7854 (try_validate_parallel): Likewise.
7855 (try_merge_compare): Likewise.
7856 (try_eliminate_compare): Call the above when no previous clobber
7858 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
7861 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7865 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
7866 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
7870 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
7871 to be any operand_equal_p operands. For & (B - 1) require
7872 B to be power of 2. Recognize
7873 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
7875 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7878 * optabs.c (expand_memory_blockage): Fix call of
7879 targetm.have_memory_blockage.
7881 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7884 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
7885 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
7886 objects to extra_objs instead of overwriting it.
7888 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7890 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
7891 Use any_fp_register_operand as operand[3] predicate. Simplify
7892 equality test for operands[2] and operands[4] memory location.
7893 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
7894 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
7895 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
7896 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
7897 any_fp_register_operand as operand[1] predicate. Simplify
7898 equality test for operands[0] and operands[3] memory location.
7899 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
7900 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
7901 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
7903 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7905 * target-insns.def: Add memory_blockage.
7906 * optabs.c (expand_memory_blockage): New function.
7907 (expand_asm_memory_barrier): Rename ...
7908 (expand_asm_memory_blockage): ... to this.
7909 (expand_mem_thread_fence): Call expand_memory_blockage
7910 instead of expand_asm_memory_barrier.
7911 (expand_mem_singnal_fence): Ditto.
7912 (expand_atomic_load): Ditto.
7913 (expand_atomic_store): Ditto.
7914 * doc/md.texi (Standard Pattern Names For Generation):
7915 Document memory_blockage instruction pattern.
7917 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
7919 * config/rl78/rl78.c (rl78_emit_libcall): New function.
7920 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
7921 * config/rl78/rl78.md: New define_expand "adddi3".
7923 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7925 * cfghooks.c (verify_flow_info): Disable check that all probabilities
7928 2017-10-13 Jeff Law <law@redhat.com>
7930 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
7932 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7935 * internal-fn.c (expand_mul_overflow): If both operands have
7936 the same highpart of -1 or 0 and the topmost bit of lowpart
7937 is different, overflow is if res <= 0 rather than res < 0.
7939 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
7941 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
7942 TARGET_P9_VECTOR code for unaligned_load case.
7944 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7946 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
7948 2017-10-13 Nathan Sidwell <nathan@acm.org>
7950 * tree-core.h (tree_contains_struct): Make bool.
7951 * tree.c (tree_contains_struct): Likewise.
7952 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
7953 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
7954 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
7955 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
7956 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
7958 2017-10-13 Richard Biener <rguenther@suse.de>
7960 * graphite-isl-ast-to-gimple.c
7961 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
7962 parameters and dominance check.
7963 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
7964 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
7965 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
7966 Do not update SSA form here or do intermediate IL verification.
7967 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
7968 (graphite_initialize): Remove check on the number of loops in
7969 the function and inline into graphite_transform_loops.
7970 (graphite_finalize): Inline into graphite_transform_loops.
7971 (graphite_transform_loops): Perform SSA update and IL verification
7973 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
7975 2017-10-13 Richard Biener <rguenther@suse.de>
7977 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
7978 graphite_expression_type_precision): Avoid global constructor
7980 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
7981 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
7982 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
7983 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
7984 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
7985 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
7987 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
7990 * config/i386/i386.h (ix86_red_zone_size): New.
7991 * config/i386/i386.md (push peephole2s): Replace
7992 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
7994 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7995 Alan Hayward <alan.hayward@arm.com>
7996 David Sherwood <david.sherwood@arm.com>
7998 * combine.c (can_change_dest_mode): Reject changes in
7999 REGMODE_NATURAL_SIZE.
8001 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8002 Alan Hayward <alan.hayward@arm.com>
8003 David Sherwood <david.sherwood@arm.com>
8005 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8006 (expand_debug_source_expr): Likewise.
8007 * combine.c (combine_simplify_rtx): Likewise.
8008 * cse.c (fold_rtx): Likewise.
8009 * fwprop.c (canonicalize_address): Likewise.
8010 * targhooks.c (default_shift_truncation_mask): Likewise.
8012 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8013 Alan Hayward <alan.hayward@arm.com>
8014 David Sherwood <david.sherwood@arm.com>
8016 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8017 (widened_mode): Likewise.
8018 (expand_unop): Likewise.
8019 * ree.c (transform_ifelse): Likewise.
8020 (merge_def_and_ext): Likewise.
8021 (combine_reaching_defs): Likewise.
8022 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8024 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8025 Alan Hayward <alan.hayward@arm.com>
8026 David Sherwood <david.sherwood@arm.com>
8028 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8029 * combine.c (gen_lowpart_for_combine): Likewise.
8030 * dwarf2out.c (rtl_for_decl_location): Likewise.
8031 * final.c (alter_subreg): Likewise.
8032 * rtlhooks.c (gen_lowpart_general): Likewise.
8033 (gen_lowpart_if_possible): Likewise.
8035 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8036 Alan Hayward <alan.hayward@arm.com>
8037 David Sherwood <david.sherwood@arm.com>
8039 * calls.c (expand_call): Use subreg_lowpart_offset.
8040 * cse.c (cse_insn): Likewise.
8041 * regcprop.c (copy_value): Likewise.
8042 (copyprop_hardreg_forward_1): Likewise.
8044 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8047 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8048 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8049 =Q constraints to +Q and into insn condition add check
8050 that operands[0] and operands[1] are equal.
8051 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8052 =Q constraints to +Q and into insn condition add check
8053 that operands[0] is equal to either operands[1] or operands[2].
8056 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8057 instead of handling MINUS_EXPR twice (once for each argument),
8058 canonicalize operand order and handle just once, use rtype where
8059 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8062 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8063 any values of __C while still being pattern recognizable as a simple
8066 2017-10-13 Richard Biener <rguenther@suse.de>
8068 PR tree-optimization/82451
8070 2017-10-02 Richard Biener <rguenther@suse.de>
8072 PR tree-optimization/82355
8073 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8074 a mapping for the enclosing loop but avoid generating one for
8076 (copy_bb_and_scalar_dependences): Remove premature codegen
8077 error on PHIs in blocks duplicated into multiple places.
8078 * graphite-scop-detection.c
8079 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8080 in the region use it as loop and nest to analyze the DR in.
8081 (try_generate_gimple_bb): Likewise.
8082 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8083 (add_loop_constraints): For blocks in a loop not in the region
8084 create a dimension with a single iteration.
8085 * sese.h (gbb_loop_at_index): Remove assert.
8087 * cfgloop.c (loop_preheader_edge): For the loop tree root
8088 return the single successor of the entry block.
8089 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8090 Reset the SCEV hashtable and niters.
8091 * graphite-scop-detection.c
8092 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8093 assert that we only have POLYNOMIAL_CHREC that vary in loops
8094 contained in the region.
8095 (scop_detection::graphite_can_represent_expr): Adjust.
8096 (scop_detection::stmt_has_simple_data_refs_p): For loops
8097 not in the region set loop to NULL. The nest is now the
8098 entry edge to the region.
8099 (try_generate_gimple_bb): Likewise.
8100 * sese.c (scalar_evolution_in_region): Adjust for
8101 instantiate_scev change.
8102 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8103 Make nest parameter the edge into the region.
8104 (create_data_ref): Likewise.
8105 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8106 entry edge into a region and adjust instantiate_scev calls.
8107 (create_data_ref): Likewise.
8108 (graphite_find_data_references_in_stmt): Likewise.
8109 (find_data_references_in_stmt): Pass the loop preheader edge
8110 from the nest argument.
8111 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8112 parameter the edge into the region.
8113 (instantiate_parameters): Use the loop preheader edge as entry.
8114 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8116 (get_instantiated_value_entry): Make instantiate_below parameter
8117 the edge into the region.
8118 (instantiate_scev_name): Likewise. Adjust dominance checks,
8119 when we cannot use loop-based instantiation instantiate by
8120 walking use-def chains.
8121 (instantiate_scev_poly): Adjust.
8122 (instantiate_scev_binary): Likewise.
8123 (instantiate_scev_convert): Likewise.
8124 (instantiate_scev_not): Likewise.
8125 (instantiate_array_ref): Remove.
8126 (instantiate_scev_3): Likewise.
8127 (instantiate_scev_2): Likewise.
8128 (instantiate_scev_1): Likewise.
8129 (instantiate_scev_r): Do not blindly handle N-operand trees.
8130 Do not instantiate array-refs. Handle all constants and invariants.
8131 (instantiate_scev): Make instantiate_below parameter
8132 the edge into the region.
8133 (resolve_mixers): Use the loop preheader edge for the region
8134 parameter to instantiate_scev_r.
8135 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8137 2017-10-13 Richard Biener <rguenther@suse.de>
8139 PR tree-optimization/82525
8140 * graphite-isl-ast-to-gimple.c
8141 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8143 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8144 Fail code generation when we cannot represent the isl integer.
8145 (binary_op_to_tree): Elide modulo operations that are no-ops
8146 in the type we code generate. Remove now superfluous code
8149 2017-10-13 Richard Biener <rguenther@suse.de>
8151 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8152 (scop_detection::harmful_loop_in_region): Remove premature
8153 IV type restriction.
8154 (scop_detection::graphite_can_represent_scev): We can handle
8155 pointer IVs just fine.
8157 2017-10-13 Alan Modra <amodra@gmail.com>
8159 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8160 "Clobbers and Scratch Registers". Add paragraph on
8161 alternative to clobbers for scratch registers and OpenBLAS
8164 2017-10-13 Alan Modra <amodra@gmail.com>
8166 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8167 example of a memory input for a string of known length. Move
8168 commentary out of table. Add a number of new examples
8169 covering array memory inputs.
8171 2017-10-12 Martin Liska <mliska@suse.cz>
8173 PR tree-optimization/82493
8174 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8175 (test_range_functions): New function.
8176 (sbitmap_c_tests): Likewise.
8177 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8178 * selftest.h (sbitmap_c_tests): New function.
8180 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8182 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8184 * config/rs6000/amo.h: Fix spacing issue.
8186 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8189 * config/i386/i386.md (*ashl<mode>3_mask_1,
8190 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8191 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8194 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8196 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8197 (profile_probability): Set max_probability
8198 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8200 * profile-count.c (profile_probability::differs_from_p): Do not
8201 rely on max_probaiblity == 10000
8203 2017-10-12 Jeff Law <law@redhat.com>
8205 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8208 2017-10-12 Martin Sebor <msebor@redhat.com>
8212 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
8213 (handle_alias_pairs): Call it.
8214 * common.opt (-Wattribute-alias): New option.
8215 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
8216 * doc/invoke.texi (-Wattribute-alias): Document.
8218 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
8221 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8223 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8225 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8226 (make_hard_regno_born, make_hard_regno_dead): Update
8227 bb_killed_pseudos and bb_gen_pseudos.
8229 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8231 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
8233 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
8235 * config/alpha/alpha.c (alpha_split_conditional_move):
8236 Use std::swap instead of manually swapping.
8237 (alpha_stdarg_optimize_hook): Ditto.
8238 (alpha_canonicalize_comparison): Ditto.
8240 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8242 * tree-loop-distribution.c (struct builtin_info): New struct.
8243 (struct partition): Refactor fields into struct builtin_info.
8244 (partition_free): Free struct builtin_info.
8245 (build_size_arg_loc, build_addr_arg_loc): Delete.
8246 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
8247 information from struct builtin_info.
8248 (find_single_drs): New function refactored from classify_partition.
8249 Also moved builtin validity checks to this function.
8250 (compute_access_range, alloc_builtin): New functions.
8251 (classify_builtin_st, classify_builtin_ldst): New functions.
8252 (classify_partition): Refactor code into functions find_single_drs,
8253 classify_builtin_st and classify_builtin_ldst.
8254 (distribute_loop): Don't do runtime alias check when distributing
8256 (find_seed_stmts_for_distribution): New function.
8257 (pass_loop_distribution::execute): Refactor code finding seed
8258 stmts into above function. Support distribution for the innermost
8259 two-level loop nest. Adjust dump information.
8261 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8263 * tree-loop-distribution.c: Adjust the general comment.
8264 (NUM_PARTITION_THRESHOLD): New macro.
8265 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
8266 (classify_partition): Skip builtin pattern of loop nest's inner loop.
8267 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
8268 in call to build_partition_graph.
8269 (finalize_partitions): New parameter. Make loop distribution more
8270 conservative by fusing more partitions.
8271 (distribute_loop): Don't do runtime alias check in case of loop nest
8273 (find_seed_stmts_for_distribution): New function.
8274 (prepare_perfect_loop_nest): New function.
8275 (pass_loop_distribution::execute): Refactor code finding seed stmts
8276 and loop nest into above functions. Support loop nest distribution.
8277 Adjust dump information accordingly.
8279 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8281 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
8282 and set PTYPE_SEQUENTIAL for merged partition.
8284 2017-10-12 Richard Biener <rguenther@suse.de>
8286 PR tree-optimization/69728
8288 2017-09-19 Richard Biener <rguenther@suse.de>
8290 PR tree-optimization/69728
8291 * graphite-sese-to-poly.c (schedule_error): New global.
8292 (add_loop_schedule): Handle empty domain by failing the
8294 (build_original_schedule): Handle schedule_error.
8296 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
8297 domain by returning an unchanged schedule.
8299 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8301 * genrecog.c (validate_pattern): For VEC_SELECT verify that
8302 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
8304 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
8306 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
8309 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8312 * expr.c (store_field): Don't optimize away bitsize == 0 store
8313 from CALL_EXPR with addressable return type.
8315 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8317 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
8318 * config/rs6000/rs6000.md (sel): Delete mode attribute.
8319 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
8320 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
8321 TARGET_ISEL instead of TARGET_ISEL<sel>.
8323 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
8325 * config/rs6000/rs6000.c
8326 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
8328 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8330 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
8332 (reg_or_cint_operand): Fix comment.
8333 (reg_or_zero_operand): New predicate.
8334 * config/rs6000/rs6000-protos.h (output_isel): Delete.
8335 * config/rs6000/rs6000.c (output_isel): Delete.
8336 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
8337 instead of reg_or_cint_operand. Output instruction directly (not via
8339 (isel_unsigned_<mode>): Ditto.
8340 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
8341 gpc_reg_operand. Add an instruction alternative for this. Output
8342 instruction directly.
8343 (*isel_reversed_unsigned_<mode>): Ditto.
8345 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8347 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
8348 (TARGET_CANONICALIZE_COMPARISON): Define.
8350 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
8353 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8354 Check whether the dest is REG before adding REG_EQUIV note.
8356 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8359 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8361 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8362 (make_hard_regno_born, make_hard_regno_dead): Update
8363 bb_killed_pseudos and bb_gen_pseudos.
8365 2017-10-11 Nathan Sidwell <nathan@acm.org>
8367 * incpath.h (enum incpath_kind): Name enum, prefix values.
8368 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
8369 * incpath.c (heads, tails): Use INC_MAX.
8370 (add_env_var_paths, add_standard_paths): Use incpath_kind.
8371 (merge_include_chains, split_quote_chain,
8372 register_include_chains): Update incpath_kind names.
8373 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
8374 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
8376 (add_framework_path, darwin_register_objc_includes): Likewise.
8377 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
8379 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8381 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
8382 Do not use float_operator operator predicate.
8383 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
8384 * config/i386/predicates.md (float_operator): Remove predicate.
8386 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8388 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
8389 (*jccxf_i387): Ditto.
8390 (*jcc<mode>_i387): Ditto.
8391 (*jccu<mode>_i387): Ditto.
8392 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
8393 (*jcc_*_i387 splitters): Remove.
8394 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
8395 * config/i386/i386.c (ix86_split_fp_branch): Remove.
8396 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
8399 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8401 * profile-count.h (slow_safe_scale_64bit): New function.
8402 (safe_scale_64bit): New inline.
8403 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
8404 * profile-count.c: Include wide-int.h
8405 (slow_safe_scale_64bit): New.
8407 2017-10-11 Nathan Sidwell <nathan@acm.org>
8409 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
8410 HAS_DECL_ASSEMBLER_NAME_P.
8411 * gimple-expr.c (gimple_decl_printable_name: Check
8412 HAS_DECL_ASSEMBLER_NAME_P too.
8413 * ipa-utils.h (type_in_anonymous_namespace_p): Check
8414 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
8415 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
8416 * passes.c (rest_of_decl_compilation): Check
8417 HAS_DECL_ASSEMBLER_NAME_P too.
8418 * recog.c (verify_changes): Likewise.
8419 * tree-pretty-print.c (dump_decl_name): Likewise.
8420 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
8422 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
8423 (SET_DECL_ASSEMBLER_NAME): Use it.
8424 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
8425 (COPY_DECL_ASSEMBLER_NAME): Likewise.
8426 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
8428 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8430 * config.gcc (i386, x86_64): Add extra objects.
8431 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
8432 (ix86_min_insn_size): Declare.
8433 (ix86_issue_rate): Declare.
8434 (ix86_adjust_cost): Declare.
8435 (ia32_multipass_dfa_lookahead): Declare.
8436 (ix86_macro_fusion_p): Declare.
8437 (ix86_macro_fusion_pair_p): Declare.
8438 (ix86_bd_has_dispatch): Declare.
8439 (ix86_bd_do_dispatch): Declare.
8440 (ix86_core2i7_init_hooks): Declare.
8441 (ix86_atom_sched_reorder): Declare.
8442 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
8443 (COSTS_N_BYTES): Move to x86-tune-costs.h.
8444 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
8445 (rip_relative_addr_p): Rename to ...
8446 (ix86_rip_relative_addr_p): ... this one; export.
8447 (memory_address_length): Update.
8448 (ix86_issue_rate): Move to x86-tune-sched.c.
8449 (ix86_flags_dependent): Move to x86-tune-sched.c.
8450 (ix86_agi_dependent): Move to x86-tune-sched.c.
8451 (exact_dependency_1): Move to x86-tune-sched.c.
8452 (exact_store_load_dependency): Move to x86-tune-sched.c.
8453 (ix86_adjust_cost): Move to x86-tune-sched.c.
8454 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
8455 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
8456 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
8457 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
8458 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
8459 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
8460 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
8461 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
8462 (min_insn_size): Rename to ...
8463 (ix86_min_insn_size): ... this one; export.
8464 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
8465 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
8466 (core2i7_first_cycle_multipass_backtrack): Move to
8467 x86-tune-sched-core.c.
8468 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
8469 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
8470 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
8471 (ix86_avoid_jump_mispredicts): Update.
8472 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
8473 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
8474 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
8475 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
8476 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
8477 (MAX_INSN): Move to ix86-tune-sched-bd.c.
8478 (MAX_IMM): Move to ix86-tune-sched-bd.c.
8479 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
8480 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
8481 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
8482 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
8483 (MAX_STORE): Move to ix86-tune-sched-bd.c.
8484 (BIG): Move to ix86-tune-sched-bd.c.
8485 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
8486 (enum insn_path): Move to ix86-tune-sched-bd.c.
8487 (get_mem_group): Move to ix86-tune-sched-bd.c.
8488 (is_cmp): Move to ix86-tune-sched-bd.c.
8489 (dispatch_violation): Move to ix86-tune-sched-bd.c.
8490 (is_branch): Move to ix86-tune-sched-bd.c.
8491 (is_prefetch): Move to ix86-tune-sched-bd.c.
8492 (init_window): Move to ix86-tune-sched-bd.c.
8493 (allocate_window): Move to ix86-tune-sched-bd.c.
8494 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
8495 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
8496 (process_end_window): Move to ix86-tune-sched-bd.c.
8497 (allocate_next_window): Move to ix86-tune-sched-bd.c.
8498 (find_constant): Move to ix86-tune-sched-bd.c.
8499 (get_num_immediates): Move to ix86-tune-sched-bd.c.
8500 (has_immediate): Move to ix86-tune-sched-bd.c.
8501 (get_insn_path): Move to ix86-tune-sched-bd.c.
8502 (get_insn_group): Move to ix86-tune-sched-bd.c.
8503 (count_num_restricted): Move to ix86-tune-sched-bd.c.
8504 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
8505 (add_insn_window): Move to ix86-tune-sched-bd.c.
8506 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
8507 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
8508 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
8509 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
8510 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
8511 (do_dispatch): Move to ix86-tune-sched-bd.c.
8512 (has_dispatch): Move to ix86-tune-sched-bd.c.
8513 * config/i386/t-i386: Add new object files.
8514 * config/i386/x86-tune-costs.h: New file.
8515 * config/i386/x86-tune-sched-atom.c: New file.
8516 * config/i386/x86-tune-sched-bd.c: New file.
8517 * config/i386/x86-tune-sched-core.c: New file.
8518 * config/i386/x86-tune-sched.c: New file.
8520 2017-10-11 Liu Hao <lh_mouse@126.com>
8522 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
8523 the generic version below instead.
8524 (should_colorize): Recognize Windows consoles as terminals
8526 * pretty-print.c [__MINGW32__] (write_all): New function.
8527 [__MINGW32__] (find_esc_head): Likewise.
8528 [__MINGW32__] (find_esc_terminator): Likewise.
8529 [__MINGW32__] (eat_esc_sequence): Likewise.
8530 [__MINGW32__] (mingw_ansi_fputs): New function that handles
8532 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
8535 2017-10-11 Richard Biener <rguenther@suse.de>
8537 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
8538 Properly call analyze_scalar_evolution with the loop of the stmt.
8540 2017-10-11 Richard Biener <rguenther@suse.de>
8542 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
8543 * tree-core.h (tree_base): Add chrec_var union member.
8544 * tree.h (CHREC_VAR): Remove.
8545 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
8546 * tree-chrec.h (build_polynomial_chrec): Adjust.
8547 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
8548 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
8550 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
8552 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
8553 * match.pd: ... here.
8554 ((T) X == (T) Y): Relax condition.
8556 2017-10-11 Bin Cheng <bin.cheng@arm.com>
8558 PR tree-optimization/82472
8559 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
8561 (break_alias_scc_partitions): Update postorder number.
8563 2017-10-11 Martin Liska <mliska@suse.cz>
8566 * opts.c (parse_no_sanitize_attribute): Do not use error_value
8568 * opts.h (parse_no_sanitize_attribute): Remove last argument.
8570 2017-10-11 Martin Liska <mliska@suse.cz>
8572 * print-rtl.c (print_insn): Move declaration of idbuf
8573 to same scope as name.
8575 2017-10-11 Martin Liska <mliska@suse.cz>
8580 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8583 2017-10-11 Martin Liska <mliska@suse.cz>
8586 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8589 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
8591 * config.gcc (default_gnu_indirect_function): Default to yes for
8592 arm*-*-linux* with glibc.
8594 2017-10-11 Richard Biener <rguenther@suse.de>
8596 * tree-scalar-evolution.c (get_scalar_evolution): Handle
8597 default-defs and types we do not want to analyze.
8598 (interpret_loop_phi): Replace unreachable code with an assert.
8599 (compute_scalar_evolution_in_loop): Remove and inline ...
8600 (analyze_scalar_evolution_1): ... here, replacing condition with
8601 what makes the intent clearer. Remove handling of cases
8602 get_scalar_evolution now handles.
8604 2017-10-10 Jim Wilson <wilson@tuliptree.org>
8606 PR rtl-optimization/81434
8607 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
8608 comment for main loop. In sched_group_found if, also add checks for
8609 pass and min_cost_group.
8611 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
8613 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
8614 (rs6000_insn_cost): New function.
8615 * config/rs6000/rs6000.md (cost): New attribute.
8617 2017-10-10 Jakub Jelinek <jakub@redhat.com>
8618 H.J. Lu <hongjiu.lu@intel.com>
8622 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
8623 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
8624 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
8625 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
8626 to OPTION_MASK_ISA_AVX512VL - builtins that have both
8627 OPTION_MASK_ISA_MMX and some other bit set require both
8628 mmx and the ISAs without the mmx bit.
8629 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
8630 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
8631 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
8632 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
8633 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
8634 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
8635 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
8636 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
8637 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
8638 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
8639 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
8640 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
8641 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
8642 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
8643 Add OPTION_MASK_ISA_MMX.
8645 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
8647 * config.gcc (armv7*-*-freebsd*): New target.
8648 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
8650 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
8652 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
8653 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
8656 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
8658 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
8659 a template parameter.
8660 (WIDE_INT_REF_FOR): Update accordingly.
8661 * tree.h (wi::int_traits <const_tree>): Delete.
8662 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
8663 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
8664 (wi::tree_to_wide_ref): New typedef.
8665 (wi::to_wide): New function.
8666 * calls.c (get_size_range): Use wi::to_wide when operating on
8668 * cgraph.c (cgraph_node::create_thunk): Likewise.
8669 * config/i386/i386.c (ix86_data_alignment): Likewise.
8670 (ix86_local_alignment): Likewise.
8671 * dbxout.c (stabstr_O): Likewise.
8672 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
8673 * expr.c (const_vector_from_tree): Likewise.
8674 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
8675 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
8676 (fold_negate_expr_1, int_const_binop_1, const_binop)
8677 (fold_convert_const_int_from_real, optimize_bit_field_compare)
8678 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
8679 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
8680 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
8681 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
8682 (fold_not_const, round_up_loc): Likewise.
8683 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
8684 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
8685 (alloca_call_type): Likewise.
8686 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
8687 * godump.c (go_output_typedef): Likewise.
8688 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
8689 * internal-fn.c (get_min_precision): Likewise.
8690 * ipa-cp.c (ipcp_store_vr_results): Likewise.
8691 * ipa-polymorphic-call.c
8692 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
8693 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
8694 (ipa_modify_call_arguments): Likewise.
8695 * match.pd: Likewise.
8696 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
8697 * print-tree.c (print_node_brief, print_node): Likewise.
8698 * stmt.c (expand_case): Likewise.
8699 * stor-layout.c (layout_type): Likewise.
8700 * tree-affine.c (tree_to_aff_combination): Likewise.
8701 * tree-cfg.c (group_case_labels_stmt): Likewise.
8702 * tree-data-ref.c (dr_analyze_indices): Likewise.
8703 (prune_runtime_alias_test_list): Likewise.
8704 * tree-dump.c (dequeue_and_dump): Likewise.
8705 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
8706 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
8707 * tree-pretty-print.c (dump_generic_node): Likewise.
8708 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
8709 (simple_iv_with_niters): Likewise.
8710 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
8711 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
8712 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
8713 * tree-ssa-loop-niter.c (split_to_var_and_offset)
8714 (refine_value_range_using_guard, number_of_iterations_ne_max)
8715 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
8716 (get_cst_init_from_scev, record_nonwrapping_iv)
8717 (scev_var_range_cant_overflow): Likewise.
8718 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
8719 * tree-ssa-pre.c (compute_avail): Likewise.
8720 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
8721 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
8722 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
8723 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
8724 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
8725 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
8726 (dump_case_nodes, try_switch_expansion): Likewise.
8727 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
8728 (vect_do_peeling): Likewise.
8729 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
8730 * tree-vect-stmts.c (vectorizable_load): Likewise.
8731 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
8732 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
8733 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
8734 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
8735 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
8736 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
8737 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
8738 (evrp_dom_walker::before_dom_children): Likewise.
8739 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
8740 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
8741 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
8742 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
8743 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
8744 (get_range_pos_neg): Likewise.
8745 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
8746 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
8747 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
8748 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
8749 * config/avr/avr.c (avr_fold_builtin): Likewise.
8750 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
8751 * config/msp430/msp430.c (msp430_attr): Likewise.
8752 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
8753 * config/powerpcspe/powerpcspe-c.c
8754 (altivec_resolve_overloaded_builtin): Likewise.
8755 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
8756 (rs6000_expand_ternop_builtin): Likewise.
8757 * config/rs6000/rs6000-c.c
8758 (altivec_resolve_overloaded_builtin): Likewise.
8759 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
8760 (rs6000_expand_ternop_builtin): Likewise.
8761 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
8763 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8765 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
8766 when copying loop nest with only one inner loop.
8768 2017-10-10 Richard Biener <rguenther@suse.de>
8770 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
8771 blocks if SCEV is active.
8772 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
8774 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
8775 (scev_initialize): Assert we are not yet initialized.
8777 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8779 * tree-loop-distribution.c (generate_loops_for_partition): Remove
8780 inner loop's exit stmt by making it always exit the loop, otherwise
8781 we would generate an infinite empty loop.
8783 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8785 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
8786 renaming variables in new preheader if it's deleted.
8788 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8790 * tree-loop-distribution.c (struct partition): Remove unused field
8791 loops of the structure.
8792 (partition_alloc, partition_free): Ditto.
8793 (build_rdg_partition_for_vertex): Ditto.
8795 2017-10-09 Jeff Law <law@redhat.com>
8797 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
8798 return type to match prototype and documentation.
8800 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8802 * config/rs6000/rs6000.c (processor_costs): Move to ...
8803 * config/rs6000/rs6000.h: ... here.
8804 (rs6000_cost): Declare.
8806 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
8808 * except.c (setjmp_fn): New global variable.
8809 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
8810 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
8811 if DONT_USE_BUILTIN_SETJMP is defined.
8813 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8815 * target.def (insn_cost): New hook.
8816 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
8817 * doc/tm.texi: Regenerate.
8818 * rtlanal.c (insn_cost): Use the new hook.
8820 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8822 * combine.c (combine_validate_cost): Compute the new insn_cost,
8823 not just pattern_cost.
8824 (try_combine): Adjust comment.
8826 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8828 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
8830 * combine.c (uid_insn_cost): Adjust comment.
8831 (combine_validate_cost): Adjust comment. Use pattern_cost instead
8833 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
8834 * dse.c (find_shift_sequence): Ditto.
8835 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
8836 (bb_valid_for_noce_process_p): Use pattern_cost.
8837 * rtl.h (insn_rtx_cost): Delete.
8838 (pattern_cost): New prototype.
8839 (insn_cost): New prototype.
8840 * rtlanal.c (insn_rtx_cost): Rename to...
8841 (pattern_cost): ... this.
8844 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
8846 * config/i386/i386.md (*jcc_2): Remove insn pattern.
8847 (*jcc<mode>_0_r_i387): Ditto.
8848 (*jccxf_r_i387): Ditto.
8849 (*jcc<mode>_r_i387): Ditto.
8850 (*jccu<mode>_r_i387): Ditto.
8851 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
8852 (*jcc): Rename from *jcc_1.
8854 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8856 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
8857 deferred rescans after the lvx/stvx recombination pre-pass.
8859 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8861 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
8862 memory operation instruction support.
8863 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
8864 (rs6000-ibm-aix[789]*): Likewise.
8865 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
8866 Document new functions.
8868 2017-10-09 Richard Biener <rguenther@suse.de>
8870 PR tree-optimization/82397
8871 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
8872 equality only for semantically equal trees.
8874 2017-10-09 Richard Biener <rguenther@suse.de>
8876 PR tree-optimization/82449
8877 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
8878 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
8879 allow constant addresses.
8880 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
8883 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8885 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
8888 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8891 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
8894 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8897 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
8899 2017-10-09 Jakub Jelinek <jakub@redhat.com>
8902 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
8903 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
8905 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
8907 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
8908 (WI_BINARY_PREDICATE_RESULT): Likewise.
8909 (wi::binary_traits::operator_result): New type.
8910 (wi::binary_traits::predicate_result): Likewise.
8911 (generic_wide_int::operator~, unary generic_wide_int::operator-)
8912 (generic_wide_int::operator==, generic_wide_int::operator!=)
8913 (generic_wide_int::operator&, generic_wide_int::and_not)
8914 (generic_wide_int::operator|, generic_wide_int::or_not)
8915 (generic_wide_int::operator^, generic_wide_int::operator+
8916 (binary generic_wide_int::operator-, generic_wide_int::operator*):
8918 (operator~, unary operator-, operator==, operator!=, operator&)
8919 (operator|, operator^, operator+, binary operator-, operator*): New
8921 * expr.c (get_inner_reference): Use wi::bit_and_not.
8922 * fold-const.c (fold_binary_loc): Likewise.
8923 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
8924 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
8925 (bit_value_binop): Likewise.
8926 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
8927 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
8928 (extract_range_from_binary_expr_1): Likewise.
8929 (masked_increment): Likewise.
8930 (simplify_bit_ops_using_ranges): Likewise.
8932 2017-10-09 Martin Jambor <mjambor@suse.cz>
8935 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
8936 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
8937 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
8939 (hsa_fixup_mov_insn_type): New function.
8940 (hsa_op_with_type::get_in_type): Use it.
8941 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
8942 immediates in an assert.
8943 (hsa_op_with_type::extend_int_to_32bit): New method.
8944 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
8945 types. Convert to dest type if necessary.
8946 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
8947 (reg_for_gimple_ssa): Pass false as min32int to
8948 hsa_type_for_scalar_tree_type.
8949 (gen_hsa_addr): Fixup type when creating addresable temporary.
8950 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
8951 (gen_hsa_unary_operation): Extend operands and convert to dest type if
8952 necessary. Call hsa_fixup_mov_insn_type.
8953 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
8954 extend operands and convert to dest type if necessary.
8955 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
8956 to dest type if necessary.
8957 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
8958 if conversion nt necessary and size matches.
8959 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
8960 to dest type if necessary.
8961 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
8962 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
8964 (gen_hsa_clrsb): Likewise.
8965 (gen_hsa_ffs): Likewise.
8966 (gen_hsa_divmod): Extend operands and convert to dest type if
8968 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
8970 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8972 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
8973 Remove empty default arguments. Use a brace block as output
8975 (conditional return): Ditto.
8977 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
8978 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
8980 (group_ending_nop): Ditto.
8981 (doloop_end): Ditto.
8982 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
8983 (splitters for those): Ditto.
8985 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8987 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
8988 a conditional jump (and the compare for it) so that pc_rtx is the
8990 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
8991 for the deleted and renamed ctr<mode>_internal[234] patterns.
8992 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
8993 Delete second conditional return pattern.
8994 (ctr<mode>_internal2): Delete this second bdnz pattern.
8995 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
8996 (ctr<mode>_internal4): Delete this second bdz pattern.
8998 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9000 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9001 (always_initialized_rtx_for_ssa_name_p): New predicate.
9002 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9003 (finish_out_of_ssa): Free new field of SA.
9004 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9005 * tree-ssa-coalesce.c: Include tree-ssa.h.
9006 (get_parm_default_def_partitions): Remove extern keyword.
9007 (get_undefined_value_partitions): New function.
9008 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9009 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9011 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9013 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9015 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9017 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9019 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9020 for targets that preffer 128bit.
9022 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9024 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9026 2017-10-08 Olivier Hainque <hainque@adacore.com>
9028 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9029 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9031 (thumb_set_return_address): Likewise.
9033 2017-10-08 Olivier Hainque <hainque@adacore.com>
9035 * common/config/arm/arm-common.c (arm_except_unwind_info):
9036 Handle DWARF2_UNWIND_INFO.
9038 2017-10-07 Michael Collison <michael.collison@arm.com>
9040 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9043 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9045 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9046 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9047 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9048 defined, force the creation of a new block for a dispatch label.
9050 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9052 * invoke.texi (Wsuggest-attribute=cold): Document.
9053 * common.opt (Wsuggest-attribute=cold): New
9054 * ipa-pure-const.c (warn_function_cold): New function.
9055 * predict.c (compute_function_frequency): Use it.
9056 * predict.h (warn_function_cold): Declare.
9058 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9060 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9063 2017-10-06 Martin Liska <mliska@suse.cz>
9065 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9066 keyword for member functions.
9067 (struct sanopt_tree_couple): New struct.
9068 (struct sanopt_tree_couple_hash): New function.
9069 (struct sanopt_ctx): Add new hash_map.
9070 (has_dominating_ubsan_ptr_check): New function.
9071 (record_ubsan_ptr_check_stmt): Likewise.
9072 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9073 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9074 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9076 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9079 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9080 aarch64_simd_valid_immediate on CONST_VECTORs.
9081 (aarch64_reg_or_bic_imm): Likewise.
9083 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9085 PR rtl-optimization/82396
9086 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9088 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9090 * graphite-dependences.c (scop_get_reads): Move code to...
9091 (scop_get_must_writes): Move code to...
9092 (scop_get_may_writes): Move code to...
9093 (scop_get_reads_and_writes): ... here.
9094 (scop_get_dependences): Call scop_get_reads_and_writes.
9096 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9098 PR tree-optimization/82434
9099 * fold-const.h (can_native_encode_type_p,
9100 can_native_encode_string_p): Remove.
9101 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9102 don't encode anything, just return what would be otherwise returned.
9103 (native_encode_fixed, native_encode_complex, native_encode_vector):
9105 (native_encode_string): Likewise. Inline by hand
9106 can_native_encode_string_p.
9107 (can_native_encode_type_p): Remove.
9108 (can_native_encode_string_p): Remove.
9109 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9110 STRING_CSTs using can_native_encode_string_p, test all
9111 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9112 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9113 argument from native_encode_expr.
9114 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9115 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9118 2017-10-06 Richard Biener <rguenther@suse.de>
9120 PR tree-optimization/82397
9121 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9122 operand_equal_p but rely on data_ref_compare_tree for detecting
9124 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9125 to match up with dr_group_sort_cmp.
9127 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9130 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9132 * config/s390/s390-builtin-types.def: Regenerate.
9134 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9137 * config/s390/s390-builtin-types.def: Regenerate.
9138 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9139 Change flag from B_VXE to B_VX.
9140 (s390_vec_min_dbl): Remove B_VXE flag.
9142 2017-10-06 Richard Biener <rguenther@suse.de>
9144 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9145 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9146 translate_isl_ast_to_gimple::is_valid_rename,
9147 translate_isl_ast_to_gimple::get_rename,
9148 translate_isl_ast_to_gimple::get_def_bb_for_const,
9149 translate_isl_ast_to_gimple::get_new_name,
9150 translate_isl_ast_to_gimple::collect_all_ssa_names,
9151 translate_isl_ast_to_gimple::copy_loop_phi_args,
9152 translate_isl_ast_to_gimple::collect_all_ssa_names,
9153 translate_isl_ast_to_gimple::copy_loop_phi_args,
9154 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9155 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9156 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9157 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9158 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9159 translate_isl_ast_to_gimple::copy_cond_phi_args,
9160 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9161 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9162 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9163 translate_isl_ast_to_gimple::rename_uses,
9164 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9165 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9166 (set_rename_for_each_def): Likewise.
9167 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9168 here. Handle rewriting SCEV analyzable uses here.
9169 (copy_bb_and_scalar_dependences): Generate code for PHI
9171 (graphite_regenerate_ast_isl): Adjust.
9172 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9173 (add_write, add_read): New functions.
9174 (build_cross_bb_scalars_def): Use it and simplify.
9175 (build_cross_bb_scalars_use): Likewise.
9176 (graphite_find_cross_bb_scalar_vars): Inline into...
9177 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9178 simulating out-of-SSA. Compute liveout and add dependencies.
9179 (build_scops): Force an empty entry block.
9180 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9182 (sese_build_liveouts): Declare.
9183 (sese_trivially_empty_bb_p): Likewise.
9184 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9185 compute liveout and debug_liveout.
9186 (sese_bad_liveouts_use): Remove.
9187 (sese_reset_debug_liveouts_bb): Likewise.
9188 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9189 (sese_build_liveouts): Build liveout and debug_liveout and store
9191 (new_sese_info): Adjust.
9192 (free_sese_info): Likewise.
9193 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9194 do not build liveout here.
9195 (move_sese_in_condition): Adjust region entry.
9196 (scev_analyzable_p): Match up with chrec_apply requirements.
9197 (sese_trivially_empty_bb_p): New.
9198 * tree-into-ssa.c (get_reaching_def): Properly support generating
9199 default-defs for incremental rewrite of anonymous names.
9201 2017-10-06 Richard Biener <rguenther@suse.de>
9203 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9204 precision do not perform modulo reduction.
9206 2017-10-06 Richard Biener <rguenther@suse.de>
9208 PR tree-optimization/82436
9209 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9210 conservatively choose the vectorization factor when checking
9211 whether we can perform the required load permutation.
9212 (vect_transform_slp_perm_load): Assert when we may not fail.
9214 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
9216 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
9217 message for incompatible -msdata=* and -mcall-* options.
9219 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9221 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
9222 rate for post-reload scheduling.
9224 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9226 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
9228 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9230 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
9231 to improve monte carlo in scimark.
9233 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9235 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
9236 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
9237 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
9238 pentium4_cost, nocona_cost): Set reassociation width to 1.
9239 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
9240 width to 2 for fp operations and 1 otherwise.
9241 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
9243 (atom_cost): Set reassociation width to 2.
9244 (slm_cost, generic_cost): Set fp reassociation width
9245 to 2 and 1 otherwise.
9246 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
9247 (core_cost): Set fp reassociation width to 4 and vector to 2.
9248 (ix86_reassociation_width): Rewrite using cost table; special case
9249 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
9250 and TARGET_AVX128_OPTIMAL.
9251 * config/i386/i386.h (processor_costs): Add
9252 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
9253 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
9254 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
9255 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
9256 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
9257 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
9259 2017-10-05 Nathan Sidwell <nathan@acm.org>
9261 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
9263 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9265 * config/arm/arm.c (arm_test_fpu_data): New.
9266 (arm_run_selftests): Call arm_test_fpu_data.
9268 2017-10-04 Nathan Sidwell <nathan@acm.org>
9270 * toplev.c (toplev::main): Remove excess parens on pretty_printer
9272 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
9274 2017-10-04 Sudakshina Das <sudi.das@arm.com>
9276 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
9277 check type for aarch64_simd_valid_immediate.
9278 (aarch64_output_simd_mov_immediate): Update prototype.
9279 (aarch64_simd_valid_immediate): Update prototype.
9280 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
9281 support for ORR-immediate.
9282 (and<mode>3): modified pattern to add support for BIC-immediate.
9283 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
9284 now checks for valid immediate for BIC and ORR based on new enum
9286 (aarch64_output_simd_mov_immediate): Function now used to output
9287 BIC/ORR imm as well based on new enum argument.
9288 * config/aarch64/constraints.md (Do): New vector immediate constraint.
9290 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
9291 (aarch64_reg_or_bic_imm): Likewise.
9293 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9295 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
9296 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
9298 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9302 PR rtl-optimization/82396
9303 * haifa-sched.c (autopref_multipass_init): Simplify
9305 (autopref_rank_data): Simplify sort order.
9306 * sched-int.h (autopref_multipass_data_): Remove
9307 multi_mem_insn_p, min_offset and max_offset.
9309 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9311 * doc/sourcebuild.texi: Document vect_peeling_profitable.
9313 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9315 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
9318 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9320 * doc/sourcebuild.texi: Document vect_long_mult.
9322 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
9324 PR tree-optimization/82413
9325 * fold-const.c (build_range_check): Use widest_int when comparing
9326 the maximum ETYPE value with HIGH.
9328 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9330 PR rtl-optimization/82396
9331 * haifa-sched.c (autopref_multipass_init): Simplify
9333 (autopref_rank_data): Simplify sort order.
9334 * sched-int.h (autopref_multipass_data_): Remove
9335 multi_mem_insn_p, min_offset and max_offset.
9337 2017-10-04 Jakub Jelinek <jakub@redhat.com>
9339 PR tree-optimization/82381
9340 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
9341 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
9344 PR tree-optimization/82374
9345 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
9346 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
9347 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
9348 current_function_decl to the new decl.
9350 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9352 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
9353 helper macro for IEEE float128 hardware built-in functions.
9354 (SQRTF128_ODD): Add built-in functions with the round-to-odd
9356 (TRUNCF128_ODD): Likewise.
9357 (ADDF128_ODD): Likewise.
9358 (SUBF128_ODD): Likewise.
9359 (MULF128_ODD): Likewise.
9360 (DIVF128_ODD): Likewise.
9361 (FMAF128_ODD): Likewise.
9362 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
9363 UNSPEC_TRUNC_ROUND_TO_ODD.
9364 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
9365 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
9366 floating point round to odd instructions.
9367 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
9368 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
9369 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
9370 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
9371 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
9372 (trunc<mode>sf2_hw): Change the truncate with round to odd
9373 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
9374 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
9375 to odd hardware instructions.
9376 (sub<mode>3_odd): Likewise.
9377 (mul<mode>3_odd): Likewise.
9378 (div<mode>3_odd): Likewise.
9379 (sqrt<mode>2_odd): Likewise.
9380 (fma<mode>4_odd): Likewise.
9381 (fms<mode>4_odd): Likewise.
9382 (nfma<mode>4_odd): Likewise.
9383 (nfms<mode>4_odd): Likewise.
9384 (trunc<mode>df2_odd): Change the truncate with round to odd
9385 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
9387 * doc/extend.texi (PowerPC built-in functions): Update documentation
9388 for existing IEEE float128-bit built-in functions. Add built-in
9389 functions that generate the IEEE 128-bit floating point round to
9392 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
9394 PR rtl-optimization/77729
9395 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
9396 to (X&(C1&~C2))|C2 transformations.
9398 2017-10-03 Martin Jambor <mjambor@suse.cz>
9400 PR tree-optimization/82363
9401 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
9402 mismatch, mark lacc written regardless of racc.
9404 2017-10-03 Jakub Jelinek <jakub@redhat.com>
9406 PR tree-optimization/82381
9407 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
9408 stmt_to_insert nor wheather SSA_NAMEs are default defs.
9409 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
9410 fallthrough into reassoc_stmt_dominates_stmt_p.
9413 * combine.c (combine_instructions): Don't combine in unreachable
9416 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9419 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
9420 function to not use the have_cpu variable. Do not set cpu_index,
9421 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
9423 (rs6000_valid_attribute_p): Remove duplicate initializations of
9424 old_optimize and func_optimize.
9425 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
9426 (rs6000_activate_target_options): Make global.
9427 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
9430 2017-10-02 Jakub Jelinek <jakub@redhat.com>
9432 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
9433 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
9434 Set *poffset to 0 and *psize and *pmax_size to -1 if
9435 *poffset + *psize overflows in HOST_WIDE_INT.
9437 PR tree-optimization/82387
9438 PR tree-optimization/82388
9439 PR tree-optimization/82389
9440 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
9441 instead of live_bytes non-NULL.
9443 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
9446 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
9449 2017-10-02 Richard Biener <rguenther@suse.de>
9451 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
9452 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
9453 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
9455 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9457 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
9458 requested precision matches the type's.
9459 * calls.c (alloc_max_size): Calculate the new candidate size as
9460 a widest_int and use wi::to_widest when comparing it with the
9461 current candidate size.
9462 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
9463 zero rather than integer_zero_node.
9464 * match.pd: Check for a no-op conversion before using wi::add
9465 rather than after. Use tree_to_uhwi when summing small shift
9466 counts into an unsigned int.
9468 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9469 Alan Hayward <alan.hayward@arm.com>
9470 David Sherwood <david.sherwood@arm.com>
9473 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
9474 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
9475 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
9476 POINTER_AND_FP_REGS.
9478 2017-10-02 Richard Biener <rguenther@suse.de>
9480 PR tree-optimization/82355
9481 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9482 a mapping for the enclosing loop but avoid generating one for
9484 (copy_bb_and_scalar_dependences): Remove premature codegen
9485 error on PHIs in blocks duplicated into multiple places.
9486 * graphite-scop-detection.c
9487 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9488 in the region use it as loop and nest to analyze the DR in.
9489 (try_generate_gimple_bb): Likewise.
9490 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9491 (add_loop_constraints): For blocks in a loop not in the region
9492 create a dimension with a single iteration.
9493 * sese.h (gbb_loop_at_index): Remove assert.
9495 2017-10-01 Kevin Buettner <kevinb@redhat.com>
9497 * omp-expand.c (adjust_context_scope): New function.
9498 (expand_parallel_call): Call adjust_context_scope.
9500 2017-10-01 Jeff Law <law@redhat.com>
9502 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
9503 dom_opt_dom_walker class with direct access to private members.
9504 Add comments. Call test_for_singularity.
9505 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
9506 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
9507 m_dummy_cond anymore.
9508 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
9510 (pass_dominator:execute): Build the dummy_cond here and pass it
9511 to the dom_opt_dom_walker ctor.
9512 (test_for_singularity): New function.
9514 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
9515 Maya Rashish <coypu@sdf.org>
9517 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
9518 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
9519 (alpha*-*-netbsd*) Use nbsd_tm_file.
9520 (arm*-*-netbsdelf*) Likewise.
9521 (i[34567]86-*-netbsdelf*) Likewise.
9522 (x86_64-*-netbsd*) Likewise.
9523 (mips*-*-netbsd*) Likewise.
9524 (powerpc-*-netbsd*) Likewise.
9525 (sh*-*-netbsd*) Likewise.
9526 (sparc-*-netbsdelf*) Likewise.
9527 (sparc64-*-netbsd*) Likewise.
9528 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
9530 (vax-*-netbsdelf*) Likewise.
9531 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
9532 (UINT_FAST8_TYPE) Likewise.
9533 (INT_FAST16_TYPE) Check CHAR_FAST16.
9534 (UINT_FAST16_TYPE) Likewise.
9536 2017-09-30 Jakub Jelinek <jakub@redhat.com>
9539 * config/i386/i386.md
9540 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
9541 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
9542 *divmodsi4_zext_2): New define_insn_and_split.
9543 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
9544 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
9545 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
9546 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
9547 New define_insn_and_split.
9548 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
9549 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
9550 operands[1] having DImode when mode is SImode.
9552 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
9553 always SImode for DIV and MOD in REG_EQUAL notes.
9555 2017-09-29 Yury Gribov <tetra2005@gmail.com>
9558 * match.pd: Fix handling of NaNs in pattern.
9560 2017-09-29 Jeff Law <law@redhat.com>
9562 * sbitmap.c (bitmap_bit_in_range_p): New function.
9563 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
9564 * tree-ssa-dse.c (live_bytes_read): New function.
9565 (dse_classify_store): Ignore reads of dead bytes.
9567 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
9568 typos and whitespace errors.
9569 * config/i386/predicates.md (address_no_seg_operand): Likewise.
9570 * config/s390/s390.c (s390_emit_prologue): Likewise.
9572 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9575 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
9576 with a symbol for LRA.
9578 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9580 PR rtl-optimization/82338
9581 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
9583 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
9585 * genmodes.c (calc_wider_mode): Suppress qsort macro.
9586 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
9587 (qsort_chk): Declare.
9588 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
9589 (qsort_chk): New function.
9591 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9593 PR tree-optimization/82337
9594 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9595 phi definition if the PHI result appears in an abnormal PHI.
9596 (find_basis_for_base_expr): Don't record a basis if the LHS of the
9597 basis appears in an abnormal PHI.
9599 2017-09-29 Richard Biener <rguenther@suse.de>
9601 * graphite-isl-ast-to-gimple.c
9602 (translate_isl_ast_to_gimple::set_codegen_error): New function.
9603 (binary_op_to_tree): Use it.
9604 (get_rename_from_scev): Likewise.
9605 (copy_loop_phi_nodes): Likewise.
9606 (copy_bb_and_scalar_dependences): Likewise.
9607 (translate_pending_phi_nodes): Likewise.
9609 2017-09-29 Jakub Jelinek <jakub@redhat.com>
9612 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
9613 for movabsq $(i32 << shift), r64.
9615 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
9617 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
9618 index when encoding %esp as %rsp to avoid 0x67 prefix.
9620 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
9622 * config/i386/i386.md (*movsf_internal, *movdf_internal):
9623 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
9625 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
9627 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
9628 Extensions with more than 16 double VFP registers.
9629 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
9630 to_clear_mask and all code related to it. Replace the remaining
9631 entry by a sbitmap and adapt code accordingly.
9633 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
9635 * brig-builtins.def: Change pure attributes to const.
9637 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9639 * config.gcc (default_gnu_indirect_function): Default to yes for
9640 sparc*-*-linux* with glibc.
9642 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9644 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
9645 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
9646 when creating .init_array and .fini_array sections with priority
9649 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
9652 * config/aarch64/aarch64.c
9653 (aarch64_builtin_support_vector_misalignment): Always return false
9654 when misalignment is unknown.
9656 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9658 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
9659 this function to return false if the definition used by the swap
9660 instruction is artificial, or if the memory address from which the
9661 constant value is loaded is not represented by a base address held
9662 in a register or if the base address register is a frame or stack
9663 pointer. Additionally, return false if the base address of the
9664 loaded constant is a SYMBOL_REF but is not considered to be a
9666 (replace_swapped_load_constant): New function.
9667 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
9668 loaded constant vector with a load of a swapped constant vector.
9670 2017-09-27 Carl Love <cel@us.ibm.com>
9672 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
9673 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
9674 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
9677 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
9679 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
9680 first, always call autopref_rank_data otherwise.
9682 2017-09-27 Richard Biener <rguenther@suse.de>
9684 * graphite-scop-detection.c (find_scop_parameters): Move
9685 loop bound handling ...
9686 (gather_bbs::before_dom_children): ... here, avoiding the need
9687 to build scop_info->loop_nest.
9688 (record_loop_in_sese): Remove.
9689 * sese.h (sese_info_t::loop_nest): Remove.
9690 * sese.c (new_sese_info): Do not allocate loop_nest.
9691 (free_sese_info): Do not free loop_nest.
9693 2017-09-27 Jakub Jelinek <jakub@redhat.com>
9696 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
9697 lhs from calls if the lhs has addressable type.
9699 2017-09-27 Richard Biener <rguenther@suse.de>
9701 * graphite.h (scop::max_alias_set): New member.
9702 * graphite-scop-detection.c: Remove references to non-existing
9703 --param in comments.
9704 (build_alias_sets): Record the maximum alias set used for drs.
9705 (build_scops): Support zero as unlimited for
9706 --param graphite-max-arrays-per-scop.
9707 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
9709 (build_poly_sr_1): ... here. Compute alias set based on the
9710 maximum alias set used for drs rather than
9711 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
9713 2017-09-27 Richard Biener <rguenther@suse.de>
9715 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
9716 --param loop-block-tile-size=0 to disable tiling.
9718 2017-09-27 Richard Biener <rguenther@suse.de>
9720 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
9721 (graphite-max-nb-scop-params): Document special value zero.
9722 * domwalk.h (dom_walker::STOP): New symbolical constant.
9723 (dom_walker::dom_walker): Add optional parameter for bb to
9725 (dom_walker::~dom_walker): Declare.
9726 (dom_walker::before_dom_children): Document STOP return value.
9727 (dom_walker::m_user_bb_to_rpo): New member.
9728 (dom_walker::m_bb_to_rpo): Likewise.
9729 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
9730 mapping here if not provided by the user.
9731 (dom_walker::~dom_walker): Free bb to RPO mapping if not
9732 provided by the user.
9733 (dom_walker::STOP): Define.
9734 (dom_walker::walk): Do not compute bb to RPO mapping here.
9735 Support STOP return value from before_dom_children to stop
9737 * graphite-optimize-isl.c (optimize_isl): If the schedule
9738 is the same still generate code if -fgraphite-identity
9739 or -floop-parallelize-all are given.
9740 * graphite-scop-detection.c: Include cfganal.h.
9741 (gather_bbs::gather_bbs): Get and pass through bb to RPO
9743 (gather_bbs::before_dom_children): Return STOP for BBs
9745 (build_scops): Compute bb to RPO mapping and pass it to
9746 the domwalk. Treat --param graphite-max-nb-scop-params=0
9747 as not limiting the number of params.
9748 * graphite.c (graphite_initialize): Remove limit on the
9749 number of basic-blocks in a function.
9750 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
9751 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
9752 default value of 10.
9754 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9756 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
9757 Adjust code to eliminate needing to do the shift right 32-bits
9758 operation after XSCVDPSPN.
9760 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9762 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
9763 ((X / Y) != 0 -> X >= Y): Likewise.
9765 2017-09-26 Carl Love <cel@us.ibm.com>
9767 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
9768 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
9769 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
9770 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
9771 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
9772 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
9773 definitions and overloading.
9774 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
9775 statement for P9V_BUILTIN_XST_LEN_R.
9776 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
9777 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
9778 define_expand and define_insn for the instructions and builtins.
9779 * doc/extend.texi: Update the built-in documentation file for the new
9781 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
9782 define_insn for the instructions
9784 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
9787 * gcc/config/netbsd-protos.h: New file.
9788 * gcc/config/netbsd.c: New file.
9789 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
9790 * gcc/config/t-netbsd: New file.
9791 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
9792 (tmake_file) Add t-netbsd.
9793 (extra_objs) Add netbsd.o.
9795 2017-09-26 Janus Weil <janus@gcc.gnu.org>
9799 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
9801 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9803 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
9804 sign extension from a vector register to a GPR by doing a 32-bit
9805 direct move and then an EXTSW.
9806 (extendsi<mode>2 splitter): Likewise.
9807 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
9808 right or vector extract after doing XSCVDPSPN. Use
9809 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
9811 (movdi_from_sf_zero_ext): Likewise.
9812 (reload_gpr_from_vsxsf): Likewise.
9813 (p8_mfvsrd_4_disf): Delete, no longer used.
9814 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
9815 then needing to move the SFmode to a GPR to use the XSCVDPSP
9816 instruction instead of FRSP and XSCVDPSPN.
9817 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
9818 it is adjacent to the other XSCVSPDP insns.
9819 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
9820 SFmode to be in traditional Altivec registers.
9821 (vsx_xscvdpspn): Eliminate useless alternative constraint.
9822 (vsx_xscvspdpn): Likewise.
9823 (vsx_xscvspdpn_scalar): Likewise.
9825 2017-09-26 Martin Jambor <mjambor@suse.cz>
9827 * tree-sra.c (compare_access_positions): Put integral types first,
9828 stabilize sorting of integral types, remove conditions putting
9829 non-full-precision integers last.
9830 (sort_and_splice_var_accesses): Disable scalarization if a
9831 non-integert would be represented by a non-full-precision integer.
9833 2017-09-26 Joseph Myers <joseph@codesourcery.com>
9835 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
9836 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
9837 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
9838 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
9839 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
9840 conditionals inside the function instead of around it. Call
9841 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
9842 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
9844 2017-09-26 Richard Biener <rguenther@suse.de>
9846 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
9848 (scop_detection::build_scop_breadth): ... this. Removed.
9849 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
9850 (scop_detection::harmful_stmt_in_bb): Likewise.
9851 (scop_detection::graphite_can_represent_stmt): Likewise.
9852 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
9853 (scop_detection::can_represent_loop): Remove recursion, fold in ...
9854 (scop_detection::can_represent_loop_1): ... this. Removed.
9855 (scop_detection::harmful_loop_in_region): Simplify after inlining
9856 the above and remove more quadraticness.
9857 (build_scops): Adjust.
9858 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
9861 2017-09-26 Jakub Jelinek <jakub@redhat.com>
9864 * config/i386/i386.c (ix86_print_operand_address_as): Only test
9865 REGNO (base) == SP_REG if base is a REG.
9868 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
9869 if it is different SSA_NAME.
9870 (optimize_range_tests_cmp_bitwise): New function.
9871 (optimize_range_tests): Call it.
9873 2017-09-26 Richard Biener <rguenther@suse.de>
9875 PR tree-optimization/82321
9876 * graphite.c (canonicalize_loop_closed_ssa): Properly check
9877 for the def being inside the loop.
9879 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9881 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
9883 * config/s390/s390-builtins.def: Fix constraint on op4.
9885 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9887 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
9888 independent expanders.
9889 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
9890 ("vec_ordered", "vec_unordered"): New expanders.
9892 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9894 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
9897 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9899 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
9900 vec_unpacks_lo_v16qi.
9901 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
9903 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9905 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
9906 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
9907 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
9909 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9911 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
9913 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
9915 ("*vec_slb<mode>"): New insn pattern.
9916 ("vec_shr_<mode>"): New expander.
9917 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
9918 and force the shift count operand to V16QImode.
9919 ("vec_srb<mode>"): Set shift count mode to V16QI.
9921 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9923 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
9924 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
9925 ("vec_widen_smult_hi_<mode>"): New expander definitions.
9927 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
9930 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
9932 2017-09-26 Richard Biener <rguenther@suse.de>
9934 PR tree-optimization/82320
9935 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
9938 2017-09-25 Jeff Law <law@redhat.com>
9940 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
9941 prototype for new argument.
9942 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
9943 mostly extracted from rs6000_emit_allocate_stack.
9944 (rs6000_emit_probe_stack_range_stack_clash): New function.
9945 (rs6000_emit_allocate_stack): Call
9946 rs6000_emit_probe_stack_range_stack_clash as needed.
9947 (rs6000_emit_probe_stack_range): Add additional argument
9948 to call to gen_probe_stack_range{si,di}.
9949 (output_probe_stack_range): New.
9950 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
9951 (output_probe_stack_range_stack_clash): New.
9952 (rs6000_emit_prologue): Emit notes into dump file as requested.
9953 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
9954 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
9955 Add additional operand and pass it to output_probe_stack_range.
9957 2017-09-25 Bin Cheng <bin.cheng@arm.com>
9959 PR tree-optimization/82163
9960 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
9961 (checking_verify_loop_closed_ssa): New parameter.
9962 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
9963 (check_loop_closed_ssa_stmt): Delete.
9964 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
9965 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
9966 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
9969 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
9971 * brig-builtins.def: Treat HSAIL barrier builtins as
9972 setjmp/longjump style functions.
9974 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9976 * target.def (constant_alignment): New hook.
9977 * defaults.h (CONSTANT_ALIGNMENT): Delete.
9978 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
9979 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
9980 * doc/tm.texi: Regenerate.
9981 * targhooks.h (default_constant_alignment): Declare.
9982 (constant_alignment_word_strings): Likewise.
9983 * targhooks.c (default_constant_alignment): New function.
9984 (constant_alignment_word_strings): Likewise.
9985 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
9986 instead of CONSTANT_ALIGNMENT.
9987 * varasm.c (align_variable, get_variable_align, build_constant_desc)
9988 (force_const_mem): Likewise.
9989 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
9990 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
9991 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
9992 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9993 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
9995 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
9996 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9997 constant_alignment_word_strings.
9998 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
9999 (CONSTANT_ALIGNMENT): Likewise.
10000 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10001 (arm_constant_alignment): New function.
10002 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10003 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10004 constant_alignment_word_strings.
10005 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10006 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10007 constant_alignment_word_strings.
10008 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10009 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10010 (cris_constant_alignment): New function.
10011 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10012 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10013 (epiphany_constant_alignment): New function.
10014 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10015 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10016 constant_alignment_word_strings.
10017 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10018 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10019 constant_alignment_word_strings.
10020 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10021 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10022 constant_alignment_word_strings.
10023 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10024 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10025 * config/i386/i386.c (ix86_constant_alignment): Make static.
10026 Use the same interface as the target hook.
10027 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10028 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10029 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10030 constant_alignment_word_strings.
10031 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10032 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10033 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10034 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10035 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10036 constant_alignment_word_strings.
10037 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10038 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10039 constant_alignment_word_strings.
10040 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10041 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10042 constant_alignment_word_strings.
10043 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10044 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10046 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10047 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10048 * config/mips/mips.c (mips_constant_alignment): New function.
10049 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10050 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10051 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10052 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10053 (mmix_constant_alignment): Make static. Use the same interface
10054 as the target hook.
10055 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10056 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10057 constant_alignment_word_strings.
10058 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10059 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10060 constant_alignment_word_strings.
10061 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10062 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10063 constant_alignment_word_strings.
10064 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10065 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10066 (rs6000_constant_alignment): New function.
10067 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10068 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10069 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10070 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10071 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10072 (rs6000_constant_alignment): New function.
10073 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10074 * config/s390/s390.c (s390_constant_alignment): New function.
10075 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10076 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10077 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10078 constant_alignment_word_strings.
10079 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10080 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10081 (sparc_constant_alignment): New function.
10082 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10083 * config/spu/spu.c (spu_constant_alignment): New function.
10084 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10085 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10086 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10087 constant_alignment_word_strings.
10088 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10089 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10090 constant_alignment_word_strings.
10091 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10092 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10093 constant_alignment_word_strings.
10094 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10095 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10096 (visium_constant_alignment): New function.
10097 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10098 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10099 (xtensa_constant_alignment): New function.
10100 * system.h (CONSTANT_ALIGNMENT): Poison.
10102 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10104 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10105 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10106 (rs6000_builtin_valid_without_lhs): New helper function.
10107 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10108 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10110 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10112 * target.h (vec_perm_indices): Use unsigned short rather than
10114 (auto_vec_perm_indices): Likewise.
10115 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10116 Use unsigned int rather than unsigned char.
10117 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10119 2017-09-25 Richard Biener <rguenther@suse.de>
10121 * cfgloop.h (sort_sibling_loops): Declare.
10122 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10123 (sort_sibling_loops): New function sorting the sibling loop list
10125 * graphite.c (graphite_transform_loops): Sort sibling loops.
10127 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10129 * target.def (vec_perm_const_ok): Change sel parameter to
10131 * optabs-query.c (can_vec_perm_p): Update accordingly.
10132 * doc/tm.texi: Regenerate.
10133 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10134 auto_vec_perm_indices and remove separate nelt field.
10135 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10136 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10137 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10138 (aarch64_expand_vec_perm_const): Update accordingly.
10139 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10140 to vec_perm_indices.
10141 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10142 auto_vec_perm_indices and remove separate nelt field.
10143 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10144 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10145 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10147 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10148 to vec_perm_indices.
10149 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10150 sel to vec_perm_indices.
10151 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10152 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10153 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10155 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10158 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10161 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10162 on the FUNCTION_DECL function context if it has a DIE that is a
10165 2017-09-25 Richard Biener <rguenther@suse.de>
10167 PR tree-optimization/82285
10168 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10171 2017-09-25 Tom de Vries <tom@codesourcery.com>
10175 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10178 2017-09-25 Richard Biener <rguenther@suse.de>
10180 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10181 ISL errors other than isl_error_quota happen. Dump if the
10182 schedule is the same.
10183 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10184 errors instead of aborting inside ISL.
10186 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10189 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10190 of libgcc_eh for m64.
10191 * config/i386/darwin64.h: Likewise.
10193 2017-09-25 Richard Biener <rguenther@suse.de>
10195 PR middle-end/82144
10196 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10197 attribute for incomplete types nor twice for complete ones.
10199 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10202 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10203 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10206 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10209 * config/i386/i386.opt (mprefer-avx256): Use
10210 ix86_target_flags variable.
10211 * config/i386/i386.c (ix86_target_string): Move
10212 -mprefer-avx256 to flag2_opts.
10214 2017-09-22 Jakub Jelinek <jakub@redhat.com>
10216 PR middle-end/35691
10217 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
10218 and x != -1 | y != -1 into (x & y) != -1.
10220 2017-09-22 Steve Ellcey <sellcey@cavium.com>
10222 * config.gcc: Add new case statement to set
10223 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
10224 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
10225 s390x-*-linux* case statements. Added aarch64 to the list of
10226 supported architectures.
10228 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10230 PR tree-optimization/82289
10231 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
10232 STMT_VINFO_RELEVANT_P.
10234 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10235 Alan Hayward <alan.hayward@arm.com>
10236 David Sherwood <david.sherwood@arm.com>
10238 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
10239 for VR_RANGE only; don't allow VR_ANTI_RANGE.
10240 (extract_range_from_binary_expr_1): Don't call
10241 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
10243 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10244 Alan Hayward <alan.hayward@arm.com>
10245 David Sherwood <david.sherwood@arm.com>
10247 * target.def (preferred_vector_alignment): New hook.
10248 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
10250 * doc/tm.texi: Regenerate.
10251 * targhooks.h (default_preferred_vector_alignment): Declare.
10252 * targhooks.c (default_preferred_vector_alignment): New function.
10253 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
10255 (DR_TARGET_ALIGNMENT): New macro.
10256 (aligned_access_p): Update commentary.
10257 (vect_known_alignment_in_bytes): New function.
10258 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
10260 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
10261 Calculate the misalignment based on the target alignment rather than
10263 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
10264 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
10265 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
10266 the target alignment, rather than masking the element misalignment
10267 with the number of elements in a vector. Also use the target
10268 alignment when calculating the maximum number of peels.
10269 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
10270 instead of TYPE_ALIGN_UNIT.
10271 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
10272 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
10273 (vect_create_addr_base_for_vector_ref): Update call accordingly.
10274 (vect_create_data_ref_ptr): Likewise.
10275 (vect_setup_realignment): Realign by ANDing with
10276 -DR_TARGET_MISALIGNMENT.
10277 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
10278 the number of peels based on DR_TARGET_ALIGNMENT.
10279 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
10280 with the guaranteed alignment boundary when deciding whether
10282 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
10283 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
10284 (ensure_base_align): Remove stmt_info parameter. Get the
10285 target base alignment from DR_TARGET_ALIGNMENT.
10286 (vectorizable_store): Update call accordingly. Interpret
10287 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
10289 (vectorizable_load): Likewise.
10291 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10292 Alan Hayward <alan.hayward@arm.com>
10293 David Sherwood <david.sherwood@arm.com>
10295 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
10296 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
10297 (vect_enhance_data_refs_alignment): Likewise.
10299 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10301 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10302 error. Only quit immediately if parsing is complete.
10303 (BEGIN): Initialize fatal_err and parse_done.
10304 (begin fpu, end fpu): Check number of arguments.
10305 (begin arch, end arch): Likewise.
10306 (begin cpu, end cpu): Likewise.
10307 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10308 (optalias): Likewise.
10310 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10312 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10313 * config/arm/arm-isa.h: Delete. Move definitions to ...
10314 * arm-cpus.in: ... here. Use new feature and fgroup values.
10315 * config/arm/arm.c (arm_option_override): Use lower case for feature
10317 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10318 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10319 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10321 (print_isa_bits_for): New function.
10322 (gen_isa): New function.
10323 (gen_comm_data): Use print_isa_bits_for.
10324 (define feature): New keyword.
10325 (define fgroup): New keyword.
10326 * config/arm/t-arm (TM_H): Remove.
10327 (GTM_H): Add arm-isa.h.
10328 (arm-isa.h): Add rule to generate file.
10329 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10330 case for feature bit names.
10332 2017-09-22 Richard Biener <rguenther@suse.de>
10334 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
10336 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
10337 print of no dependency loops ...
10338 * graphite.c (graphite_transform_loops): ... here.
10339 (canonicalize_loop_closed_ssa_form): Work from inner to outer
10341 (same_close_phi_node, remove_duplicate_close_phi,
10342 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
10343 (canonicalize_loop_closed_ssa): ... here and simplify.
10344 * graphite-optimize-isl.c: Include tree-vectorizer.h.
10345 (optimize_isl): Use dump_printf_loc to tell when we stopped
10346 optimizing because of an ISL timeout.
10348 2017-09-22 Richard Biener <rguenther@suse.de>
10350 PR tree-optimization/82291
10351 * tree-if-conv.c (predicate_mem_writes): Make sure to
10352 remove writes in blocks predicated with false.
10354 2017-09-22 Richard Biener <rguenther@suse.de>
10356 * sese.c: Include cfganal.h.
10357 (if_region_set_false_region): Remove.
10358 (create_if_region_on_edge): Likewise.
10359 (move_sese_in_condition): Re-implement without destroying
10362 2017-09-22 Richard Biener <rguenther@suse.de>
10364 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
10365 Verify both BBs contain loop PHI nodes before dispatching to
10366 copy_loop_phi_args.
10367 (graphite_regenerate_ast_isl): Do not recompute dominators,
10368 do not verify three times. Restructure for clarity.
10369 * graphite-scop-detection.c (same_close_phi_node,
10370 remove_duplicate_close_phi, make_close_phi_nodes_unique,
10371 defined_in_loop_p, canonicalize_loop_closed_ssa,
10372 canonicalize_loop_closed_ssa_form): Simplify, remove excess
10373 checking and SSA rewrite, move to ...
10374 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
10375 (graphite_initialize): Do not pass in ctx, do not reset the
10376 SCEV cache, compute only dominators.
10377 (graphite_transform_loops): Allocate ISL ctx after
10378 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
10379 Maintain post-dominators only around build_scops.
10380 * sese.c (if_region_set_false_region): Make static. Free
10381 and recompute dominators.
10382 (move_sese_in_condition): Assert we don't get called with
10383 post-dominators computed.
10384 * sese.h (if_region_set_false_region): Remove.
10386 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
10388 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
10389 mode attribute for TARGET_AVX512VL.
10391 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
10393 * config/i386/i386.opt (mprefer-avx256): New option.
10394 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
10396 (ix86_preferred_simd_mode): Return 256-bit AVX modes
10397 for TARGET_PREFER_AVX256.
10398 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
10400 2017-09-21 Jeff Law <law@redhat.com>
10402 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
10403 Fix dump output if the only stack space is for pushed registers.
10405 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10407 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
10410 2017-09-21 Martin Sebor <msebor@redhat.com>
10413 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
10414 code (in C++) or code that triggers warnings.
10416 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
10418 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
10420 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
10422 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
10423 * sched-rgn.c: Ditto.
10424 * sel-sched-ir.c: Ditto.
10426 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
10428 * toplev.h (set_random_seed): Adjust return type.
10429 * toplev.c (init_local_tick): Move eager initialization of random_seed
10430 to get_random_seed. Adjust comment.
10431 (init_random_seed): Inline to get_random_seed, delete.
10432 (get_random_seed): Initialize random_seed lazily.
10433 (set_random_seed): Do not return previous value.
10434 (print_switch_value): Do not call get_random_seed.
10436 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
10438 * cgraph.c (delete_function_version): New, broken out from...
10439 (cgraph_node::delete_function_version): ...here. Rename to
10440 cgraph_node::delete_function_version_by_decl. Update all uses.
10441 (cgraph_node::remove): Call delete_function_version.
10443 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10446 * tree-inline.c (expand_call_inline): Emit clobber stmts for
10447 VAR_DECLs to which addressable non-volatile parameters are mapped
10448 and for id->retvar after the return value assignment. Clear
10449 id->retval and id->retbnd after inlining.
10451 2017-09-21 Richard Biener <rguenther@suse.de>
10453 PR tree-optimization/82276
10454 PR tree-optimization/82244
10455 * tree-vrp.c (build_assert_expr_for): Set
10456 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
10458 (remove_range_assertions): Revert earlier change.
10460 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
10463 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
10465 2017-09-21 Richard Biener <rguenther@suse.de>
10467 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10468 Restore valid IL after code generation errors.
10469 * graphite.c (graphite_transform_loops): Diagnose code
10470 generation issues as MSG_MISSED_OPTIMIZATION and continue
10471 with processing SCOPs.
10473 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10474 Alan Hayward <alan.hayward@arm.com>
10475 David Sherwood <david.sherwood@arm.com>
10477 * calls.c (compute_argument_addresses): Use simplify_gen_binary
10478 rather than choosing between plus_constant and gen_rtx_<CODE>.
10479 * expr.c (emit_push_insn): Likewise.
10480 (expand_expr_real_2): Likewise.
10482 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10483 Alan Hayward <alan.hayward@arm.com>
10484 David Sherwood <david.sherwood@arm.com>
10486 * loop-unroll.c (split_iv): Call copy_rtx on the step.
10488 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10489 Alan Hayward <alan.hayward@arm.com>
10490 David Sherwood <david.sherwood@arm.com>
10492 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
10493 calling tree_to_uhwi.
10495 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10496 Alan Hayward <alan.hayward@arm.com>
10497 David Sherwood <david.sherwood@arm.com>
10499 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
10500 INTEGER_CST rather than a negative test for ADDR_EXPR.
10502 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10503 Alan Hayward <alan.hayward@arm.com>
10504 David Sherwood <david.sherwood@arm.com>
10506 * tree-vrp.c (extract_range_from_binary_expr_1): Check
10507 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
10509 2017-09-21 Richard Biener <rguenther@suse.de>
10511 PR tree-optimization/71351
10512 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
10513 graphite_create_new_loop_guard): Remove, fold remaining parts
10515 (translate_isl_ast_node_for): ... here and simplify.
10517 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10520 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
10521 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
10522 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
10523 alternative always use QI mode, for -Os imov (=R,R) alternative
10524 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
10527 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10528 Jeff Law <law@redhat.com>
10530 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
10531 (allocate_stack_space): New function, partially extracted from
10532 s390_emit_prologue.
10533 (s390_emit_prologue): Track offset to most recent stack probe.
10534 Code to allocate space moved into allocate_stack_space.
10535 Dump actions when no stack is allocated.
10536 (s390_prologue_plus_offset): New function.
10537 (s390_emit_stack_probe): Likewise.
10539 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
10541 * common.opt (Wa, Wl, Wp, g, gz=): Add
10543 (gno-column-info): Remove.
10544 (gcolumn-info): Drop RejectNegative.
10545 (gno-): New prefix.
10546 (gno-record-gcc-switches): Remove.
10547 (grecord-gcc-switches): Drop RejectNegative.
10548 (gno-split-dwarf): Remove.
10549 (gsplit-dwarf): Drop RejectNegative.
10550 (gno-strict-dwarf): Remove.
10551 (gstrict-dwarf): Drop RejectNegative.
10552 * config/darwin.opt (gfull, gused): Add RejectNegative.
10553 * dwarf2out.c (gen_producer_string): Drop
10554 gno-record-gcc-switches handler.
10555 * optc-gen.awk: Add g to prefixes with negative forms.
10556 * opts-common.c (remapping_prefix_p): New.
10557 (find_opt): Check it.
10558 (generate_canonical_option): Test g prefix.
10559 (option_map): Add -gno- mapping.
10560 (add_misspelling_candidates): Check remapping_prefix_p.
10562 2017-09-20 Jeff Law <law@redhat.com>
10564 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
10565 thinko in stack clash protection support.
10567 * explow.c (compute_stack_clash_protection_loop_data): Use
10568 CONST_INT_P instead of explicit test. Verify object is a
10569 CONST_INT_P before looking at INTVAL.
10570 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
10571 instead of explicit test.
10573 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
10576 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
10577 address instead of to r1 and r11.
10579 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
10581 * config.gcc: Support "knm".
10582 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
10583 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10585 * config/i386/i386.c (m_KNM): Define.
10586 (processor_target_table): Add "knm".
10588 (ix86_option_override_internal): Add "knm".
10589 (ix86_issue_rate): Add PROCESSOR_KNM.
10590 (ix86_adjust_cost): Ditto.
10591 (ia32_multipass_dfa_lookahead): Ditto.
10592 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
10593 (fold_builtin_cpu): Add M_INTEL_KNM.
10594 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
10595 (processor_type): Add PROCESSOR_KNM.
10596 * config/i386/x86-tune.def: Add m_KNM.
10597 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
10599 2017-09-20 Richard Biener <rguenther@suse.de>
10601 PR tree-optimization/80213
10602 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
10603 are allowed in empty BBs as well.
10604 (canonicalize_loop_closed_ssa): Also look for other complex
10606 (scop_detection::get_sese): Include the loop-closed PHI block
10608 (scop_detection::merge_sese): Remove code adding extra blocks.
10609 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
10610 (build_scops): Assert the final returned scop is invalid.
10612 2017-09-20 Richard Biener <rguenther@suse.de>
10614 PR tree-optimization/82264
10615 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
10617 (vn_phi_lookup): Likewise.
10618 (vn_phi_insert): Likewise.
10620 2017-09-20 Jakub Jelinek <jakub@redhat.com>
10622 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
10623 that fits into uhwi or shwi, add DW_AT_const_value regardless
10624 of early_dwarf without going through RTL, using add_AT_unsigned
10627 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
10628 (DEBUG_LTO_ABBREV_SECTION): Likewise.
10629 (DEBUG_LTO_MACINFO_SECTION): Likewise.
10630 (DEBUG_MACRO_SECTION): Likewise.
10631 (DEBUG_LTO_MACRO_SECTION): Likewise.
10632 (DEBUG_STR_DWO_SECTION): Likewise.
10633 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
10634 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
10635 (DEBUG_LTO_DWO_LINE_SECTION): Define.
10636 (DEBUG_LTO_LINE_STR_SECTION): Define.
10637 (init_sections_and_labels): Initialize debug_line_str_section
10638 variable. Initialize debug_loc_section for -gdwarf-5 to
10639 DEBUG_LOCLISTS_SECTION. Formatting fixes.
10641 2017-09-20 Richard Biener <rguenther@suse.de>
10643 * graphite-sese-to-poly.c (extract_affine): Properly handle
10644 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
10646 2017-09-20 Richard Biener <rguenther@suse.de>
10648 PR tree-optimization/81373
10649 * graphite-scop-detection.c (build_cross_bb_scalars_def):
10650 Force SESE live-out defs to be handled even if they are
10653 2017-09-19 Jeff Law <law@redhat.com>
10655 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
10656 nothing for stack adjustments with REG_STACK_CHECK.
10657 * sched-deps.c (parse_add_or_inc): Reject insns with
10658 REG_STACK_CHECK from dependency breaking.
10659 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
10660 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
10661 * reg-notes.def (STACK_CHECK): New note.
10663 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
10664 (ix86_expand_prologue): Dump stack clash info as needed.
10665 Call ix86_adjust_stack_and_probe_stack_clash as needed.
10667 * function.c (dump_stack_clash_frame_info): New function.
10668 * function.h (dump_stack_clash_frame_info): Prototype.
10669 (enum stack_clash_probes): New enum.
10671 * config/alpha/alpha.c (alpha_expand_prologue): Also check
10672 flag_stack_clash_protection.
10673 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
10674 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
10675 (arm_frame_pointer_required): Likewise.
10676 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
10677 (ia64_expand_prologue): Likewise.
10678 * config/mips/mips.c (mips_expand_prologue): Likewise.
10679 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
10680 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10681 (sparc_flat_expand_prologue): Likewise.
10682 * config/spu/spu.c (spu_expand_prologue): Likewise.
10684 * explow.c: Include "params.h".
10685 (anti_adjust_stack_and_probe_stack_clash): New function.
10686 (get_stack_check_protect): Likewise.
10687 (compute_stack_clash_protection_loop_data): Likewise.
10688 (emit_stack_clash_protection_loop_start): Likewise.
10689 (emit_stack_clash_protection_loop_end): Likewise.
10690 (allocate_dynamic_stack_space): Use get_stack_check_protect.
10691 Use anti_adjust_stack_and_probe_stack_clash.
10692 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
10693 (emit_stack_clash_protection_loop_start): Likewise.
10694 (emit_stack_clash_protection_loop_end): Likewise.
10695 * rtl.h (get_stack_check_protect): Prototype.
10696 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
10697 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
10698 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
10700 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
10702 * doc/tm.texi: Rebuilt.
10703 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
10704 get_stack_check_protect.
10705 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
10706 * config/arm/arm.c (arm_expand_prologue): Likewise.
10707 (arm_frame_pointer_required): Likewise.
10708 * config/i386/i386.c (ix86_expand_prologue): Likewise.
10709 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
10710 * config/mips/mips.c (mips_expand_prologue): Likewise.
10711 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
10712 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
10713 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10714 (sparc_flat_expand_prologue): Likewise.
10716 * common.opt (-fstack-clash-protection): New option.
10717 * flag-types.h (enum stack_check_type): Note difference between
10718 -fstack-check= and -fstack-clash-protection.
10719 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
10720 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
10721 * toplev.c (process_options): Issue warnings/errors for cases
10722 not handled with -fstack-clash-protection.
10723 * doc/invoke.texi (-fstack-clash-protection): Document new option.
10724 (-fstack-check): Note additional problem with -fstack-check=generic.
10725 Note that -fstack-check is primarily for Ada and refer users
10726 to -fstack-clash-protection for stack-clash-protection.
10727 Document new params for stack clash protection.
10729 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
10731 * config/i386/i386.c (ix86_split_long_move): Do not handle
10732 address used for LEA in a special way.
10734 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
10736 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
10738 2017-09-19 Martin Sebor <msebor@redhat.com>
10741 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
10742 of incompatible types.
10744 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
10746 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10747 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
10748 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10749 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
10751 2017-09-19 Richard Biener <rguenther@suse.de>
10753 PR tree-optimization/82244
10754 * tree-vrp.c (remove_range_assertions): Do not propagate
10755 a constant to abnormals but replace the assert with a copy.
10757 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
10759 PR rtl-optimization/57878
10760 PR rtl-optimization/68988
10761 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
10762 avoidance test involving non_reload_pseudos. Move frequency test
10763 below the general fragmentation avoidance test.
10765 2017-09-19 Richard Biener <rguenther@suse.de>
10767 PR tree-optimization/69728
10768 * graphite-sese-to-poly.c (schedule_error): New global.
10769 (add_loop_schedule): Handle empty domain by failing the
10771 (build_original_schedule): Handle schedule_error.
10773 2017-09-19 Richard Biener <rguenther@suse.de>
10775 * graphite-scop-detection.c (scop_detection::can_represent_loop):
10776 Do not iterate to sibling loops but only to siblings of inner
10779 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
10782 * config/m68k/m68k.md (moveq feeding equality comparison): Check
10783 that the registers are different.
10785 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
10787 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
10788 to processor_model and "amdfam17h" to arch_names_table.
10789 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
10791 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10794 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
10796 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10797 Alan Hayward <alan.hayward@arm.com>
10798 David Sherwood <david.sherwood@arm.com>
10800 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
10802 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
10804 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
10805 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
10806 vect_schedule_slp_instance.
10807 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
10808 Update call to vect_slp_analyze_node_operations. Simplify return
10810 (vect_slp_analyze_bb_1): Update call accordingly.
10811 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
10812 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
10813 (vect_schedule_slp): Update call accordingly.
10815 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10816 Alan Hayward <alan.hayward@arm.com>
10817 David Sherwood <david.sherwood@arm.com>
10819 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
10820 with types that aren't in fact scalar.
10822 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10824 * tree-vect-slp.c (vect_record_max_nunits): New function,
10826 (vect_build_slp_tree_1): ...here.
10827 (vect_build_slp_tree_2): Call it for phis too.
10829 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10831 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
10832 to vect_get_vec_def_for_operand when getting the mask operand.
10834 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10835 Alan Hayward <alan.hayward@arm.com>
10836 David Sherwood <david.sherwood@arm.com>
10838 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
10841 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10842 Alan Hayward <alan.hayward@arm.com>
10843 David Sherwood <david.sherwood@arm.com>
10845 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
10846 calculation for vector booleans.
10848 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10849 Alan Hayward <alan.hayward@arm.com>
10850 David Sherwood <david.sherwood@arm.com>
10852 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
10854 (vect_transform_stmt): ...here.
10855 (vect_analyze_stmt): Use it instead of calling
10856 vectorizable_live_operation directly.
10858 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
10860 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
10861 non-SIMT targets in acc vector loops.
10863 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
10865 * configure.ac: Add arc and check if assembler supports gdwarf2.
10866 * configure: Regenerate.
10868 2017-09-18 Richard Biener <rguenther@suse.de>
10870 PR tree-optimization/82220
10871 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
10872 epilogue niters from the min_profitable_iters compute.
10874 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10877 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
10878 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
10879 (ix86_init_pic_reg): Revert 2017-09-01 changes.
10881 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
10884 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
10885 switching to a new text section.
10887 2017-09-18 Richard Biener <rguenther@suse.de>
10889 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
10891 (build_alias_set): Reject aliases with no access function.
10893 2017-09-18 Richard Biener <rguenther@suse.de>
10895 PR tree-optimization/79622
10896 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
10898 (build_cross_bb_scalars_use): Likewise.
10900 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
10902 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
10904 2017-09-18 Alan Modra <amodra@gmail.com>
10907 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
10908 stack_pointer_rtx for count 0. Update comments. Break up
10909 large rtl expression.
10911 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
10913 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
10914 Increase to 20 bytes.
10915 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
10916 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
10917 or avx version of the stub.
10919 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
10922 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
10923 compute the minimum stack alignment. Also update preferred stack
10924 boundary for leaf functions.
10926 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
10928 PR tree-optimization/82228
10929 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
10932 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10934 * common/config/nds32/nds32-common.c
10935 (nds32_option_optimization_table): Refine formatting.
10936 (nds32_option_optimization_table): Use -fsched-pressure and
10937 -fomit-frame-pointer for specific optimization level.
10939 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10941 * config/nds32/nds32.c: Refine formatting and comments.
10942 * config/nds32/nds32.h: Likewise.
10943 * config/nds32/nds32.md: Likewise.
10944 * config/nds32/nds32-cost.c: Likewise.
10945 * config/nds32/nds32-isr.c: Likewise.
10946 * config/nds32/nds32-md-auxiliary.c: Likewise.
10947 * config/nds32/nds32-multiple.md: Likewise.
10948 * config/nds32/nds32-predicates.c: Likewise.
10950 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
10951 Jakub Jelinek <jakub@redhat.com>
10953 Add support for -std=c++2a.
10954 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
10956 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
10958 2017-09-15 Steve Ellcey <sellcey@cavium.com>
10961 * doc/extend.texi (Common Function Attributes): Add
10962 references to ARM, AArch64, and S/390 specific attributes.
10963 (Function Specific Option Pragmas): Add AArch64 and S/390
10964 to list of back ends that support the target pragma.
10966 2017-09-15 Nathan Sidwell <nathan@acm.org>
10968 * doc/standards.texi: Fix C++17 description. Update URLs for
10971 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
10973 * common.opt (Wcast-align=strict): New warning option.
10974 * doc/invoke.texi: Document -Wcast-align=strict.
10976 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
10978 * cgraph.h (cgraph_thunk_info): Add comments.
10979 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
10980 assert for VIRTUAL_* arguments stricter.
10982 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
10984 PR tree-optimization/71026
10985 * match.pd: Move RDIV patterns from fold-const.c
10986 * fold-const.c (distribute_real_division): Removed.
10987 (fold_binary_loc): Remove calls to distribute_real_divison.
10989 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10991 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
10992 c++1z and gnu++1z as deprecated. Change other references to
10993 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
10994 Change -Wc++1z-compat to -Wc++17-compat.
10995 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
10996 * dwarf2out.c (highest_c_language): Handle C++17.
10997 (gen_compile_unit_die): Likewise.
10999 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11001 PR rtl-optimization/82192
11002 * combine.c (make_extraction): Don't look through non-paradoxical
11003 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11006 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11007 Alan Hayward <alan.hayward@arm.com>
11008 David Sherwood <david.sherwood@arm.com>
11010 * target.def (function_arg_offset): New hook.
11011 * targhooks.h (default_function_arg_offset): Declare.
11012 * targhooks.c (default_function_arg_offset): New function.
11013 * function.c (locate_and_pad_parm): Use
11014 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11015 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11016 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11017 * doc/tm.texi: Regenerate.
11018 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11019 * config/spu/spu.c (spu_function_arg_offset): New function.
11020 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11021 * system.h (FUNCTION_ARG_OFFSET): Poison.
11023 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11024 Alan Hayard <alan.hayward@arm.com>
11025 David Sherwood <david.sherwood@arm.com>
11027 * target.def (truly_noop_truncation): New hook.
11028 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11029 than TRULY_NOOP_TRUNCATION.
11030 * hooks.h (hook_bool_uint_uint_true): Declare.
11031 * hooks.c (hook_bool_uint_uint_true): New function.
11032 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11033 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11034 * doc/tm.texi: Regenerate.
11035 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11036 rather than TRULY_NOOP_TRUNCATION in comments.
11037 (simplify_comparison): Likewise.
11038 (record_truncated_value): Likewise.
11039 * expmed.c (extract_bit_field_1): Likewise.
11040 (extract_split_bit_field): Likewise.
11041 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11042 instead of TRULY_NOOP_TRUNCATION.
11043 * function.c (assign_parm_setup_block): Likewise.
11044 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11045 * rtlhooks.c: Include target.h.
11046 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11047 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11048 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11049 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11050 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11051 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11052 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11053 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11054 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11055 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11056 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11057 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11058 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11059 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11060 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11061 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11062 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11063 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11064 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11065 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11066 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11067 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11068 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11069 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11070 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11071 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11072 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11073 rather than TRULY_NOOP_TRUNCATION in comments.
11074 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11075 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11076 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11077 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11078 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11079 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11080 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11081 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11082 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11083 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11084 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11085 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11086 rather than TRULY_NOOP_TRUNCATION in comments.
11087 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11088 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11089 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11090 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11091 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11092 TRULY_NOOP_TRUNCATION condition.
11093 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11094 (TRULY_NOOP_TRUNCATION): Delete.
11095 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11096 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11097 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11098 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11099 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11100 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11101 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11102 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11103 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11104 rather than TRULY_NOOP_TRUNCATION in comments.
11105 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11106 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11107 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11108 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11109 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11110 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11112 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11115 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11116 (*cmp_ior): Likewise.
11117 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11118 (*ior_scc_scc_cmp): Likewise.
11119 (*and_scc_scc): Likewise.
11120 (*and_scc_scc_cmp): Likewise.
11122 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11123 Alan Hayard <alan.hayward@arm.com>
11124 David Sherwood <david.sherwood@arm.com>
11126 * target.def (can_change_mode_class): New hook.
11127 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11128 (hard_regno_nregs): Likewise.
11129 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11130 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11131 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11132 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11133 (LOAD_EXTEND_OP): Update accordingly.
11134 * doc/tm.texi: Regenerate.
11135 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11136 CANNOT_CHANGE_MODE_CLASS.
11137 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11138 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11139 * combine.c (simplify_set): Update accordingly.
11140 * emit-rtl.c (validate_subreg): Likewise.
11141 * recog.c (general_operand): Likewise.
11142 * regcprop.c (mode_change_ok): Likewise.
11143 * reload1.c (choose_reload_regs): Likewise.
11144 (inherit_piecemeal_p): Likewise.
11145 * rtlanal.c (simplify_subreg_regno): Likewise.
11146 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11147 instead of CANNOT_CHANGE_MODE_CLASS.
11148 (reload_cse_simplify_operands): Likewise.
11149 * reload.c (push_reload): Use targetm.can_change_mode_class
11150 instead of CANNOT_CHANGE_MODE_CLASS.
11151 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11152 REG_CANNOT_CHANGE_MODE_P.
11153 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11154 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11155 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11156 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11157 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11158 (arm_can_change_mode_class): New function.
11159 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11160 than CANNOT_CHANGE_MODE_CLASS in comments.
11161 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11162 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11163 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11164 (ix86_can_change_mode_class): ...this new function, inverting the
11165 sense of the return value.
11166 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11167 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11168 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11169 (ia64_can_change_mode_class): New function.
11170 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11171 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11172 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11173 (m32c_can_change_mode_class): ...this new function, inverting the
11174 sense of the return value.
11175 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11176 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11177 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11178 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11179 (mips_can_change_mode_class): ...this new function, inverting the
11180 sense of the return value.
11181 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11182 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11183 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11184 (msp430_can_change_mode_class): New function.
11185 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11186 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11187 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11188 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11189 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11190 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11191 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11192 (pa_cannot_change_mode_class): Replace with...
11193 (pa_can_change_mode_class): ...this new function, inverting the
11194 sense of the return value.
11195 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11196 than CANNOT_CHANGE_MODE_CLASS in comments.
11197 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11198 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11199 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11200 (pdp11_cannot_change_mode_class): Replace with...
11201 (pdp11_can_change_mode_class): ...this new function, inverting the
11202 sense of the return value.
11203 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11204 * config/powerpcspe/powerpcspe-protos.h
11205 (rs6000_cannot_change_mode_class_ptr): Delete.
11206 * config/powerpcspe/powerpcspe.c
11207 (rs6000_cannot_change_mode_class_ptr): Delete.
11208 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11209 (rs6000_option_override_internal): Assign to
11210 targetm.can_change_mode_class instead of
11211 rs6000_cannot_change_mode_class_ptr.
11212 (rs6000_cannot_change_mode_class): Replace with...
11213 (rs6000_can_change_mode_class): ...this new function, inverting the
11214 sense of the return value.
11215 (rs6000_debug_cannot_change_mode_class): Replace with...
11216 (rs6000_debug_can_change_mode_class): ...this new function.
11217 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11218 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
11219 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11220 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11221 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
11223 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
11224 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11225 (rs6000_option_override_internal): Assign to
11226 targetm.can_change_mode_class instead of
11227 rs6000_cannot_change_mode_class_ptr.
11228 (rs6000_cannot_change_mode_class): Replace with...
11229 (rs6000_can_change_mode_class): ...this new function, inverting the
11230 sense of the return value.
11231 (rs6000_debug_cannot_change_mode_class): Replace with...
11232 (rs6000_debug_can_change_mode_class): ...this new function.
11233 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11234 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
11235 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
11236 (s390_can_change_mode_class): ...this new function, inverting the
11237 sense of the return value.
11238 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11239 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11240 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
11241 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11242 (sh_cannot_change_mode_class): Replace with...
11243 (sh_can_change_mode_class): ...this new function, inverting the
11244 sense of the return value.
11245 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11246 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11247 (sparc_can_change_mode_class): New function.
11248 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11249 * config/spu/spu.c (spu_can_change_mode_class): New function.
11250 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11251 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11252 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11253 (visium_can_change_mode_class): New function.
11254 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
11256 2017-09-15 Richard Biener <rguenther@suse.de>
11258 PR tree-optimization/82217
11259 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
11260 but not undefined case.
11262 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11265 * postreload.c (reload_cse_simplify_operands): Skip
11266 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
11268 2017-09-15 Richard Biener <rguenther@suse.de>
11270 PR tree-optimization/68823
11271 * graphite-scop-detection.c (build_alias_set): If we have a
11272 possible dependence check whether we can handle them by just
11273 looking at the DRs DR_ACCESS_FNs.
11274 (build_scops): If build_alias_set fails, fail the SCOP.
11276 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11278 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
11279 to support float128 built-in functions that require the ISA 3.0
11281 (BU_FLOAT128_3_HW): Likewise.
11282 (SQRTF128): Add support for the IEEE 128-bit square root and fma
11283 built-in functions.
11284 (FMAF128): Likewise.
11286 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
11287 support for built-in functions that need the ISA 3.0 IEEE 128-bit
11288 floating point instructions.
11289 (rs6000_invalid_builtin): Likewise.
11290 (rs6000_builtin_mask_names): Likewise.
11291 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
11292 (RS6000_BTM_FLOAT128_HW): Likewise.
11293 (RS6000_BTM_COMMON): Likewise.
11294 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
11296 * doc/extend.texi (RS/6000 built-in functions): Document the
11297 IEEE 128-bit floating point square root and fused multiply-add
11298 built-in functions.
11300 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
11302 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
11303 reg (r2) isn't in the set of registers defined in the prologue.
11305 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11306 Alan Hayward <alan.hayward@arm.com>
11307 David Sherwood <david.sherwood@arm.com>
11309 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
11310 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
11311 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
11312 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
11313 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
11315 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
11316 max_vectorization_factor.
11317 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
11319 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11320 Alan Hayward <alan.hayward@arm.com>
11321 David Sherwood <david.sherwood@arm.com>
11323 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
11324 (vect_worthwhile_without_simd_p): Declare.
11325 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
11326 (vectorizable_reduction): Use it.
11327 * tree-vect-stmts.c (vectorizable_shift): Likewise.
11328 (vectorizable_operation): Likewise.
11330 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11331 Alan Hayward <alan.hayward@arm.com>
11332 David Sherwood <david.sherwood@arm.com>
11334 * tree-vectorizer.h (vect_get_num_copies): New function.
11335 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
11336 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11337 (vectorizable_induction): Likewise.
11338 (vectorizable_live_operation): Likewise.
11339 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11340 (vectorizable_bswap): Likewise.
11341 (vectorizable_call): Likewise.
11342 (vectorizable_conversion): Likewise.
11343 (vectorizable_assignment): Likewise.
11344 (vectorizable_shift): Likewise.
11345 (vectorizable_operation): Likewise.
11346 (vectorizable_store): Likewise.
11347 (vectorizable_load): Likewise.
11348 (vectorizable_condition): Likewise.
11349 (vectorizable_comparison): Likewise.
11350 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
11352 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11353 Alan Hayward <alan.hayward@arm.com>
11354 David Sherwood <david.sherwood@arm.com>
11356 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
11357 of vect_init_vector.
11359 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11360 Alan Hayward <alan.hayward@arm.com>
11361 David Sherwood <david.sherwood@arm.com>
11363 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
11364 an inline wrapper that provides a location.
11365 (gimple_build_vector): Likewise.
11366 * gimple-fold.c (gimple_build_vector_from_val): New function.
11367 (gimple_build_vector): Likewise.
11368 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
11369 functions to build the initial value. Always return a gimple value.
11370 (get_initial_defs_for_reduction): Likewise. Only compute
11372 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
11373 vect_init_vector on the results from get_initial_def(s)_for_reduction.
11374 (vectorizable_induction): Use gimple_build_vector rather than
11377 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11378 Alan Hayward <alan.hayward@arm.com>
11379 David Sherwood <david.sherwood@arm.com>
11381 * target.h (vec_perm_indices): New typedef.
11382 (auto_vec_perm_indices): Likewise.
11383 * optabs-query.h: Include target.h
11384 (can_vec_perm_p): Take a vec_perm_indices *.
11385 * optabs-query.c (can_vec_perm_p): Likewise.
11386 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
11387 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11388 * tree-vect-generic.c (lower_vec_perm): Likewise.
11389 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
11390 (vect_grouped_load_supported): Likewise.
11391 (vect_shift_permute_load_chain): Likewise.
11392 (vect_permute_store_chain): Use auto_vec_perm_indices.
11393 (vect_permute_load_chain): Likewise.
11394 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
11395 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
11396 Update uses of can_vec_perm_p.
11397 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
11398 mode with a number of elements. Take a vec_perm_indices *.
11399 (vect_create_epilog_for_reduction): Update accordingly.
11400 Use auto_vec_perm_indices.
11401 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
11402 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
11403 (vect_transform_slp_perm_load): Likewise.
11404 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
11405 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
11406 (vect_gen_perm_mask_checked): Likewise.
11407 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
11408 (vect_gen_perm_mask_checked): Likewise.
11409 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
11410 (vectorizable_store): Likewise.
11411 (vectorizable_load): Likewise.
11412 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
11413 (vectorizable_bswap): Likewise.
11415 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11416 Alan Hayward <alan.hayward@arm.com>
11417 David Sherwood <david.sherwood@arm.com>
11419 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
11420 * tree.c (build_vector): Likewise.
11421 (build_vector_from_ctor): Update accordingly.
11422 (build_vector_from_val): Likewise.
11423 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
11424 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11425 * tree-vect-generic.c (add_rshift): Likewise.
11426 (expand_vector_divmod): Likewise.
11427 (optimize_vector_constructor): Likewise.
11428 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
11429 (vect_transform_slp_perm_load): Likewise.
11430 (vect_schedule_slp_instance): Likewise.
11431 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
11432 (vectorizable_call): Likewise.
11433 (vect_gen_perm_mask_any): Likewise. Add elements in order.
11434 * expmed.c (make_tree): Likewise.
11435 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
11436 a vector passed to build_vector.
11437 (fold_convert_const): Likewise.
11438 (exact_inverse): Likewise.
11439 (fold_ternary_loc): Likewise.
11440 (fold_relational_const): Likewise.
11441 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
11442 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
11443 (const_unop): Likewise. Store the reduction accumulator in a
11444 variable rather than an array.
11445 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
11446 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
11447 the new vector, rather than constructing it after the input arrays.
11448 (native_interpret_vector): Use auto_vec<tree> when building
11449 a vector passed to build_vector. Add elements in order.
11450 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
11451 auto_vec<tree> when building a vector passed to build_vector.
11452 (vect_create_epilog_for_reduction): Likewise.
11453 (vectorizable_induction): Likewise.
11454 (get_initial_def_for_reduction): Likewise. Fix indentation of
11456 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
11458 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
11459 passed to build_vector.
11461 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11462 Alan Hayward <alan.hayward@arm.com>
11463 David Sherwood <david.sherwood@arm.com>
11465 * tree-core.h (tree_base::u): Add an "nelts" field.
11466 (tree_vector): Use VECTOR_CST_NELTS as the length.
11467 * tree.c (tree_size): Likewise.
11468 (make_vector): Initialize VECTOR_CST_NELTS.
11469 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
11470 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
11471 TYPE_VECTOR_SUBPARTS.
11472 * expr.c (const_vector_mask_from_tree): Consistently use "units"
11473 as the number of units, setting it from VECTOR_CST_NELTS.
11474 (const_vector_from_tree): Likewise.
11475 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
11476 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
11477 (fold_negate_expr_1): Likewise.
11478 (fold_convert_const): Likewise.
11479 (const_binop): Likewise. Differentiate the number of output and
11481 (const_unop): Likewise.
11482 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
11483 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
11484 in cases that did the opposite.
11486 2017-09-14 Richard Biener <rguenther@suse.de>
11488 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
11491 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
11493 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
11495 2017-09-14 Jakub Jelinek <jakub@redhat.com>
11498 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
11499 if and where to split a bb, except for splitting before debug insn
11500 sequences followed by non-label real insn. Delete debug insns
11501 in between basic blocks.
11503 * combine.c (make_compound_operation_int): Formatting fixes.
11505 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
11506 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11507 * config/netbsd.h (LINK_EH_SPEC): Likewise.
11508 * config/sol2.h (LINK_EH_SPEC): Likewise.
11509 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11510 * config/s390/linux.h (LINK_SPEC): Likewise.
11511 * config/freebsd.h (LINK_EH_SPEC): Likewise.
11512 * config/openbsd.h (LINK_EH_SPEC): Likewise.
11513 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11514 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
11515 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
11516 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11517 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
11518 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
11520 2017-09-13 Jakub Jelinek <jakub@redhat.com>
11522 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
11524 (ENDFILE_LINUX_SPEC): Likewise.
11525 (LINK_EH_SPEC): Likewise.
11526 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
11527 (LINK_OS_LINUX_SPEC32): Likewise.
11528 (LINK_OS_LINUX_SPEC64): Likewise.
11529 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
11530 (LINK_OS_LINUX_SPEC): Likewise.
11532 2017-09-13 Martin Liska <mliska@suse.cz>
11534 PR middle-end/82154
11535 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
11536 CASE_HIGH is NULL_TREE.
11538 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11539 Alan Hayward <alan.hayward@arm.com>
11540 David Sherwood <david.sherwood@arm.com>
11542 * target.def (secondary_memory_needed): New hook.
11543 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
11544 instead of SECONDARY_MEMORY_NEEDED.
11545 (secondary_memory_needed_mode): Likewise.
11546 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
11547 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
11548 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
11549 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
11550 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
11551 * doc/tm.texi: Regenerate.
11552 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
11553 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
11554 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11555 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
11556 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
11557 * config/i386/i386.c (inline_secondary_memory_needed): Put the
11558 mode argument first and change the reg_class arguments to reg_class_t.
11559 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
11560 Make static. Update the call to inline_secondary_memory_needed.
11561 (ix86_register_move_cost): Update the call to
11562 inline_secondary_memory_needed.
11563 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11564 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
11566 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
11567 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11569 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
11570 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
11571 * config/mips/mips.c (mips_secondary_memory_needed): Make static
11572 and match hook interface. Add comment from mips.h.
11573 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11574 * config/mmix/mmix.md (truncdfsf2): Refer to
11575 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11577 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
11578 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
11579 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11580 (pa_secondary_memory_needed): New function.
11581 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
11582 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
11583 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11584 (pdp11_secondary_memory_needed): Make static and match hook interface.
11585 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
11586 * config/powerpcspe/powerpcspe-protos.h
11587 (rs6000_secondary_memory_needed_ptr): Delete.
11588 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
11590 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11591 (rs6000_option_override_internal): Assign to
11592 targetm.secondary_memory_needed rather than
11593 rs6000_secondary_memory_needed_ptr.
11594 (rs6000_secondary_memory_needed): Match hook interface.
11595 (rs6000_debug_secondary_memory_needed): Likewise.
11596 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
11597 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
11598 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
11599 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11600 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
11601 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
11603 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
11604 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11605 (rs6000_option_override_internal): Assign to
11606 targetm.secondary_memory_needed rather than
11607 rs6000_secondary_memory_needed_ptr.
11608 (rs6000_secondary_memory_needed): Match hook interface.
11609 (rs6000_debug_secondary_memory_needed): Likewise.
11610 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
11611 * config/s390/s390.c (s390_secondary_memory_needed): New function.
11612 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11613 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
11614 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11615 (sparc_secondary_memory_needed): New function.
11616 * lra-constraints.c (check_and_process_move): Refer to
11617 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11619 (curr_insn_transform): Likewise.
11620 (process_alt_operands): Use targetm.secondary_memory_needed
11621 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11622 (check_secondary_memory_needed_p): Likewise.
11623 (choose_split_class): Likewise.
11624 * reload.c: Unconditionally include code that was previously
11625 conditional on SECONDARY_MEMORY_NEEDED.
11626 (push_secondary_reload): Use targetm.secondary_memory_needed
11627 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11628 (push_reload): Likewise.
11629 * reload1.c: Unconditionally include code that was previously
11630 conditional on SECONDARY_MEMORY_NEEDED.
11631 (choose_reload_regs): Use targetm.secondary_memory_needed
11632 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11633 (gen_reload): Likewise.
11634 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
11636 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11637 Alan Hayward <alan.hayward@arm.com>
11638 David Sherwood <david.sherwood@arm.com>
11640 * target.def (secondary_memory_needed_mode): New hook:
11641 * targhooks.c (default_secondary_memory_needed_mode): Declare.
11642 * targhooks.h (default_secondary_memory_needed_mode): New function.
11643 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
11644 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
11645 * doc/tm.texi: Regenerate.
11646 * lra-constraints.c (check_and_process_move): Use
11647 targetm.secondary_memory_needed_mode instead of
11648 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
11649 (curr_insn_transform): Likewise.
11650 * reload.c (get_secondary_mem): Likewise.
11651 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11652 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
11654 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11655 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11656 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
11657 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11658 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
11660 * config/powerpcspe/powerpcspe-protos.h
11661 (rs6000_secondary_memory_needed_mode): Delete.
11662 * config/powerpcspe/powerpcspe.c
11663 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11664 (rs6000_secondary_memory_needed_mode): Make static.
11665 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11666 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
11668 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11670 (rs6000_secondary_memory_needed_mode): Make static.
11671 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11672 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
11673 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11674 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11675 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11677 (sparc_secondary_memory_needed_mode): New function.
11678 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
11680 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
11682 * config/aarch64/constraints.md (Umq): New constraint.
11683 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
11685 (mov<mode>): Update condition.
11687 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11689 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
11690 when bitposition is the same.
11692 2017-09-13 Richard Biener <rguenther@suse.de>
11694 * dwarf2out.c (output_die_symbol): Remove.
11695 (output_die): Do not output a DIEs symbol.
11697 2017-09-13 Richard Biener <rguenther@suse.de>
11699 PR middle-end/82128
11700 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
11701 default-def to avoid breaking iterator update with the weird
11702 interaction with cgraph_update_edges_for_call_stmt_node.
11704 2017-09-13 Richard Biener <rguenther@suse.de>
11706 * tree-cfg.c (verify_gimple_assign_binary): Add verification
11707 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
11708 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
11709 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
11711 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
11713 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
11714 Disable pc relative literal load irrespective of
11715 TARGET_FIX_ERR_A53_84341 for default.
11717 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
11719 * config/sparc/sparc.c (output_return): Output the source location of
11720 the insn in the delay slot, if any.
11721 (output_sibcall): Likewise.
11723 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
11726 * common.opt (-static-pie): New alias.
11727 (shared): Negate static-pie.
11728 (-no-pie): Update help text.
11730 (static-pie): New option.
11731 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
11732 -static-pie support.
11733 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
11734 (LINK_EH_SPEC): Likewise.
11735 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11736 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11737 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11738 * gcc.c (LINK_COMMAND_SPEC): Likewise.
11739 (init_gcc_specs): Likewise.
11740 (init_spec): Likewise.
11741 (display_help): Update help message for -pie.
11742 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
11745 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
11747 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
11748 (movdi_aarch64): Likewise.
11749 (movti_aarch64): Likewise.
11751 2017-09-12 Simon Wright <simon@pushface.org>
11754 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
11755 calculation of the minor version, always output as 0.
11757 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11760 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11761 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
11762 on it early, rather than manual conversion late. For
11763 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
11764 instead of performing manual conversion.
11766 2017-09-12 Carl Love <cel@us.ibm.com>
11768 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
11769 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
11771 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11772 VMULOSW): Add definitions.
11773 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11774 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11775 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11776 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11777 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11779 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11781 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
11783 (movti_aarch64): Likewise.
11784 (movdf_aarch64): Likewise.
11785 (movtf_aarch64): Likewise.
11786 (load_pairdi): Likewise.
11787 (store_pairdi): Likewise.
11788 (load_pairdf): Likewise.
11789 (store_pairdf): Likewise.
11790 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
11791 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
11792 (ldr_got_small_<mode>): Likewise.
11793 (ldr_got_small_28k_<mode>): Likewise.
11794 (ldr_got_tiny): Likewise.
11795 * config/aarch64/iterators.md (ldst_sz): New.
11796 (ldpstp_sz): Likewise.
11797 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
11799 (thunderx_load): Split load_8 to load_16.
11800 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
11802 (thunderx2t99_storepair_basic): Split store_8 to store_16.
11803 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
11804 (xgene1_store_pair): Split store_8 to store_16.
11805 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
11806 (falkor_st_0_st_sd): Split store_8 to store_16.
11808 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11810 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
11811 and store1/2/3/4 to store_4/8/12/16.
11812 * config/aarch64/aarch64.md: Update for rename.
11813 * config/arm/arm.md: Likewise.: Likewise.
11814 * config/arm/arm.c: Likewise.
11815 * config/arm/thumb1.md: Likewise.
11816 * config/arm/thumb2.md: Likewise.
11817 * config/arm/vfp.md: Likewise.
11818 * config/arm/arm-generic.md: Likewise.
11819 * config/arm/arm1020e.md: Likewise.
11820 * config/arm/arm1026ejs.md: Likewise.
11821 * config/arm/arm1136jfs.md: Likewise.
11822 * config/arm/arm926ejs.md: Likewise.
11823 * config/arm/cortex-a15.md: Likewise.
11824 * config/arm/cortex-a17.md: Likewise.
11825 * config/arm/cortex-a5.md: Likewise.
11826 * config/arm/cortex-a53.md: Likewise.
11827 * config/arm/cortex-a57.md: Likewise.
11828 * config/arm/cortex-a7.md: Likewise.
11829 * config/arm/cortex-a8.md: Likewise.
11830 * config/arm/cortex-a9.md: Likewise.
11831 * config/arm/cortex-m4.md: Likewise.
11832 * config/arm/cortex-m7.md: Likewise.
11833 * config/arm/cortex-r4.md: Likewise.
11834 * config/arm/exynos-m1.md: Likewise.
11835 * config/arm/fa526.md: Likewise.
11836 * config/arm/fa606te.md: Likewise.
11837 * config/arm/fa626te.md: Likewise.
11838 * config/arm/fa726te.md: Likewise.
11839 * config/arm/fmp626.md: Likewise.
11840 * config/arm/iwmmxt.md: Likewise.
11841 * config/arm/ldmstm.md: Likewise.
11842 * config/arm/marvell-pj4.md: Likewise.
11843 * config/arm/xgene1.md: Likewise.
11844 * config/aarch64/thunderx.md: Likewise.
11845 * config/aarch64/thunderx2t99.md: Likewise.
11846 * config/aarch64/falkor.md: Likewise.
11848 2017-09-12 Martin Liska <mliska@suse.cz>
11850 * attribs.c (private_lookup_attribute): New function.
11851 * attribs.h (private_lookup_attribute): Declared here.
11852 (lookup_attribute): Called from this place.
11854 2017-09-12 Richard Biener <rguenther@suse.de>
11856 PR tree-optimization/82157
11857 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
11858 stmts with side-effects.
11860 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11861 Alan Hayward <alan.hayward@arm.com>
11862 David Sherwood <david.sherwood@arm.com>
11864 * target.def (hard_regno_nregs): New hook.
11865 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
11866 * targhooks.h (default_hard_regno_nregs): Declare.
11867 * targhooks.c (default_hard_regno_nregs): New function.
11868 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
11869 (TARGET_HARD_REGNO_NREGS): ...this hook.
11870 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
11871 (CLASS_MAX_NREGS): Likewise.
11872 * doc/tm.texi: Regenerate.
11873 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
11874 instead of HARD_REGNO_NREGS.
11875 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
11876 HARD_REGNO_NREGS in the comment.
11877 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
11878 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
11879 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
11880 Return an unsigned int.
11881 (TARGET_HARD_REGNO_NREGS): Redefine.
11882 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
11883 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
11884 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11885 (arc_hard_regno_nregs): New function.
11886 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
11887 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
11888 (arm_hard_regno_nregs): New function.
11889 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
11890 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
11891 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
11892 (TARGET_HARD_REGNO_NREGS): Redefine.
11893 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
11894 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
11895 (HARD_REGNO_NREGS): Delete.
11896 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
11897 (cr16_hard_regno_nregs): New function.
11898 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
11899 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
11900 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
11901 (cris_hard_regno_nregs): New function.
11902 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
11903 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
11904 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11905 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
11906 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
11907 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
11908 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
11909 (frv_hard_regno_nregs): Make static. Take and return an
11911 (frv_class_max_nregs): Remove outdated copy of documentation.
11912 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
11913 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
11914 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
11915 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
11916 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
11917 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
11918 (TARGET_HARD_REGNO_NREGS): Redefine.
11919 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
11920 (CLASS_MAX_NREGS): Update comment.
11921 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
11922 (ia64_hard_regno_nregs): New function.
11923 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
11924 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
11925 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
11926 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
11927 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
11929 (m32c_hard_regno_nregs): Likewise. Make static.
11930 (TARGET_HARD_REGNO_NREGS): Redefine.
11931 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
11932 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
11933 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
11934 (m68k_hard_regno_nregs): New function.
11935 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
11936 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
11937 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
11938 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
11939 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
11940 Take and return an unsigned int.
11941 (TARGET_HARD_REGNO_NREGS): Redefine.
11942 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
11943 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11944 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
11945 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
11946 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
11947 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
11948 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
11949 (msp430_hard_regno_nregs): Make static. Take and return an
11951 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
11952 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
11953 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
11954 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
11955 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
11956 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
11957 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
11958 (TARGET_HARD_REGNO_NREGS): Redefine.
11959 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
11960 (PA_HARD_REGNO_NREGS): ...this.
11961 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
11962 (PA_HARD_REGNO_NREGS): ...this.
11963 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11964 (pa_hard_regno_nregs): New function.
11965 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
11966 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
11967 (pdp11_hard_regno_nregs): New function.
11968 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
11969 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
11970 (rs6000_hard_regno_nregs_hook): New function.
11971 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
11972 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
11973 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
11974 Take and return an unsigned int. Move earlier in file.
11975 (TARGET_HARD_REGNO_NREGS): Redefine.
11976 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
11977 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
11978 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
11979 (rl78_hard_regno_nregs): Make static. Take and return an
11981 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
11982 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
11983 (rs6000_hard_regno_nregs_hook): New function.
11984 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
11985 * config/rx/rx.c (rx_hard_regno_nregs): New function.
11986 (TARGET_HARD_REGNO_NREGS): Redefine.
11987 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
11988 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
11989 instead of HARD_REGNO_NREGS.
11990 (s390_hard_regno_nregs): New function.
11991 (s390_hard_regno_mode_ok): Add comment from s390.h.
11992 (TARGET_HARD_REGNO_NREGS): Redefine.
11993 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
11994 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
11995 (sh_hard_regno_nregs): New function.
11996 (sh_pass_in_reg_p): Use it.
11997 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
11998 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11999 (sparc_hard_regno_nregs): New function.
12000 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12001 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12002 (spu_function_arg_advance): Use it, supplying a valid register number.
12003 (TARGET_HARD_REGNO_NREGS): Redefine.
12004 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12005 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12006 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12007 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12008 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12009 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12010 (CLASS_MAX_NREGS): Remove copy of old documentation.
12011 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12012 (visium_hard_regno_nregs): New function.
12013 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12014 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12015 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12016 xtensa_hard_regno_nregs): New function.
12017 * system.h (HARD_REGNO_NREGS): Poison.
12019 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12021 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12022 hard_regno_nregs instead of HARD_REGNO_NREGS.
12023 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12024 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12025 (c6x_expand_epilogue): Likewise.
12026 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12027 (frv_read_iacc_argument): Likewise.
12028 * config/sh/sh.c: Include regs.h.
12029 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12030 (regs_used): Likewise.
12031 (output_stack_adjust): Likewise.
12032 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12033 * expmed.c: Include regs.h.
12034 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12035 * ree.c: Include regs.h.
12036 (combine_reaching_defs): Use hard_regno_nregs instead of
12038 (add_removable_extension): Likewise.
12040 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12042 * regs.h (hard_regno_nregs): Turn into a function.
12043 (end_hard_regno): Update accordingly.
12044 * caller-save.c (setup_save_areas): Likewise.
12045 (save_call_clobbered_regs): Likewise.
12046 (replace_reg_with_saved_mem): Likewise.
12047 (insert_restore): Likewise.
12048 (insert_save): Likewise.
12049 * combine.c (can_change_dest_mode): Likewise.
12050 (move_deaths): Likewise.
12051 (distribute_notes): Likewise.
12052 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12053 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12054 (rs6000_split_multireg_move): Likewise.
12055 (rs6000_register_move_cost): Likewise.
12056 (rs6000_memory_move_cost): Likewise.
12057 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12058 (rs6000_split_multireg_move): Likewise.
12059 (rs6000_register_move_cost): Likewise.
12060 (rs6000_memory_move_cost): Likewise.
12061 * cselib.c (cselib_reset_table): Likewise.
12062 (cselib_lookup_1): Likewise.
12063 * emit-rtl.c (set_mode_and_regno): Likewise.
12064 * function.c (aggregate_value_p): Likewise.
12065 * ira-color.c (setup_profitable_hard_regs): Likewise.
12066 (check_hard_reg_p): Likewise.
12067 (calculate_saved_nregs): Likewise.
12068 (assign_hard_reg): Likewise.
12069 (improve_allocation): Likewise.
12070 (calculate_spill_cost): Likewise.
12071 * ira-emit.c (modify_move_list): Likewise.
12072 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12073 (ira_hard_reg_in_set_p): Likewise.
12074 * ira.c (setup_reg_mode_hard_regset): Likewise.
12075 (clarify_prohibited_class_mode_regs): Likewise.
12076 (check_allocation): Likewise.
12077 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12078 (lra_setup_reg_renumber): Likewise.
12079 (setup_try_hard_regno_pseudos): Likewise.
12080 (spill_for): Likewise.
12081 (assign_hard_regno): Likewise.
12082 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12083 * lra-constraints.c (in_class_p): Likewise.
12084 (lra_constraint_offset): Likewise.
12085 (simplify_operand_subreg): Likewise.
12086 (lra_constraints): Likewise.
12087 (split_reg): Likewise.
12088 (split_if_necessary): Likewise.
12089 (invariant_p): Likewise.
12090 (inherit_in_ebb): Likewise.
12091 * lra-lives.c (process_bb_lives): Likewise.
12092 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12093 (get_hard_regs): Likewise.
12094 (do_remat): Likewise.
12095 * lra-spills.c (assign_spill_hard_regs): Likewise.
12096 * mode-switching.c (create_pre_exit): Likewise.
12097 * postreload.c (reload_combine_recognize_pattern): Likewise.
12098 * recog.c (peep2_find_free_register): Likewise.
12099 * regcprop.c (kill_value_regno): Likewise.
12100 (set_value_regno): Likewise.
12101 (copy_value): Likewise.
12102 (maybe_mode_change): Likewise.
12103 (find_oldest_value_reg): Likewise.
12104 (copyprop_hardreg_forward_1): Likewise.
12105 * regrename.c (check_new_reg_p): Likewise.
12106 (regrename_do_replace): Likewise.
12107 * reload.c (push_reload): Likewise.
12108 (combine_reloads): Likewise.
12109 (find_dummy_reload): Likewise.
12110 (operands_match_p): Likewise.
12111 (find_reloads): Likewise.
12112 (find_equiv_reg): Likewise.
12113 (reload_adjust_reg_for_mode): Likewise.
12114 * reload1.c (count_pseudo): Likewise.
12115 (count_spilled_pseudo): Likewise.
12116 (find_reg): Likewise.
12117 (clear_reload_reg_in_use): Likewise.
12118 (free_for_value_p): Likewise.
12119 (allocate_reload_reg): Likewise.
12120 (choose_reload_regs): Likewise.
12121 (reload_adjust_reg_for_temp): Likewise.
12122 (emit_reload_insns): Likewise.
12123 (delete_output_reload): Likewise.
12124 * rtlanal.c (subreg_get_info): Likewise.
12125 * sched-deps.c (sched_analyze_reg): Likewise.
12126 * sel-sched.c (init_regs_for_mode): Likewise.
12127 (mark_unavailable_hard_regs): Likewise.
12128 (choose_best_reg_1): Likewise.
12129 (verify_target_availability): Likewise.
12130 * valtrack.c (dead_debug_insert_temp): Likewise.
12131 * var-tracking.c (track_loc_p): Likewise.
12132 (emit_note_insn_var_location): Likewise.
12133 * varasm.c (make_decl_rtl): Likewise.
12134 * reginfo.c (choose_hard_reg_mode): Likewise.
12135 (init_reg_modes_target): Refer directly to
12136 this_target_regs->x_hard_regno_nregs.
12138 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12140 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12141 instead of hard_regno_nregs.
12143 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12145 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12146 end_hard_regno instead of hard_regno_nregs.
12147 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12148 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12149 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12150 * ira-color.c (improve_allocation): Likewise.
12151 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12152 * lra-lives.c (mark_regno_live): Likewise.
12153 (mark_regno_dead): Likewise.
12154 * lra-remat.c (operand_to_remat): Likewise.
12155 * lra.c (collect_non_operand_hard_regs): Likewise.
12156 * postreload.c (reload_combine_note_store): Likewise.
12157 (move2add_valid_value_p): Likewise.
12158 * reload.c (regno_clobbered_p): Likewise.
12160 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12162 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12164 * config/v850/v850.c (v850_reorg): Likewise.
12165 * reload.c (refers_to_regno_for_reload_p): Likewise.
12166 (find_equiv_reg): Likewise.
12167 * reload1.c (reload_reg_reaches_end_p): Likewise.
12169 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12171 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12173 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12174 * config/arm/arm.c (output_move_neon): Likewise.
12175 (arm_attr_length_move_neon): Likewise.
12176 (neon_split_vcombine): Likewise.
12177 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12178 (c6x_mark_reg_written): Likewise.
12179 (c6x_dwarf_register_span): Likewise.
12180 * config/i386/i386.c (ix86_save_reg): Likewise.
12181 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12182 (rws_access_reg): Likewise.
12183 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12184 * mode-switching.c (create_pre_exit): Likewise.
12185 * ree.c (combine_reaching_defs): Likewise.
12186 (add_removable_extension): Likewise.
12187 * regcprop.c (find_oldest_value_reg): Likewise.
12188 (copyprop_hardreg_forward_1): Likewise.
12189 * reload.c (reload_inner_reg_of_subreg): Likewise.
12190 (push_reload): Likewise.
12191 (combine_reloads): Likewise.
12192 (find_dummy_reload): Likewise.
12193 (reload_adjust_reg_for_mode): Likewise.
12194 * reload1.c (find_reload_regs): Likewise.
12195 (forget_old_reloads_1): Likewise.
12196 (reload_reg_free_for_value_p): Likewise.
12197 (reload_adjust_reg_for_temp): Likewise.
12198 (emit_reload_insns): Likewise.
12199 (delete_output_reload): Likewise.
12200 * sel-sched.c (choose_best_reg_1): Likewise.
12201 (choose_best_pseudo_reg): Likewise.
12203 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12204 Alan Hayward <alan.hayward@arm.com>
12205 David Sherwood <david.sherwood@arm.com>
12207 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12208 * target.def (slow_unaligned_access): New hook.
12209 * targhooks.h (default_slow_unaligned_access): Declare.
12210 * targhooks.c (default_slow_unaligned_access): New function.
12211 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
12212 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
12213 * doc/tm.texi: Regenerate.
12214 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
12215 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
12216 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
12218 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
12219 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
12221 (rs6000_slow_unaligned_access): New function.
12222 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12223 (expand_block_compare): Likewise.
12224 (expand_strn_compare): Likewise.
12225 (rs6000_rtx_costs): Likewise.
12226 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
12227 (riscv_slow_unaligned_access): Likewise.
12228 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
12229 (riscv_slow_unaligned_access_p): ...this and make static.
12230 (riscv_option_override): Update accordingly.
12231 (riscv_slow_unaligned_access): New function.
12232 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12233 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
12234 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12235 (rs6000_slow_unaligned_access): New function.
12236 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12237 (rs6000_rtx_costs): Likewise.
12238 * config/rs6000/rs6000-string.c (expand_block_compare)
12239 (expand_strn_compare): Use targetm.slow_unaligned_access instead
12240 of SLOW_UNALIGNED_ACCESS.
12241 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
12242 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
12243 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
12244 of SLOW_UNALIGNED_ACCESS.
12245 * expmed.c (simple_mem_bitfield_p): Likewise.
12246 * expr.c (alignment_for_piecewise_move): Likewise.
12247 (emit_group_load_1): Likewise.
12248 (emit_group_store): Likewise.
12249 (copy_blkmode_from_reg): Likewise.
12250 (emit_push_insn): Likewise.
12251 (expand_assignment): Likewise.
12252 (store_field): Likewise.
12253 (expand_expr_real_1): Likewise.
12254 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12255 * lra-constraints.c (simplify_operand_subreg): Likewise.
12256 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12257 * gimple-ssa-store-merging.c: Likewise in block comment at start
12259 * tree-ssa-strlen.c: Include target.h.
12260 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
12261 of SLOW_UNALIGNED_ACCESS.
12262 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
12264 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12266 PR rtl-optimization/82185
12267 * expmed.c (emit_store_flag_int): Only test tem if it has been
12270 2017-09-12 Richard Biener <rguenther@suse.de>
12272 PR middle-end/82149
12273 * match.pd ((FTYPE) N CMP CST): Fix typo.
12275 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
12277 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
12279 (mips_near_type_p): Add 'short_call' attribute as a synonym
12281 * doc/extend.texi (short_call): Document new function attribute.
12283 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12286 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
12287 assertion check that in the condition.
12288 (get_atomic_generic_size): Likewise. Before testing if parameter
12289 has pointer type, if it has array type, call for C++
12290 default_conversion to perform array-to-pointer conversion.
12292 2017-09-12 Richard Biener <rguenther@suse.de>
12294 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
12295 for operations we cannot scalarize.
12297 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
12299 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
12300 vectors heap vectors. Clean up comments.
12301 Make visited_bbs a reference.
12302 (profitable_jump_thread_path): Make GC
12303 vectors heap vectors. Clean up comments.
12305 (convert_and_register_jump_thread_path): Make GC vectors heap
12307 (check_subpath_and_update_thread_path): Same. Clean up comments.
12308 Make visited_bbs a reference.
12309 (handle_phi): Abstract common code to to
12310 register_jump_thread_path_if_profitable.
12311 Rename VAR_BB to DEF_BB.
12313 Make GC vectors heap vectors.
12314 Make visited_bbs a reference.
12315 (handle_assignment): Same.
12316 (register_jump_thread_path_if_profitable): New.
12317 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
12319 Make GC vectors heap vectors. Clean up comments.
12320 Make visited_bbs a reference.
12321 (find_jump_threads_backwards): Make visited_bbs live in the stack.
12322 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
12325 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
12328 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
12329 words of E_DImode object are reachable by xtensa_uimm8x4 access.
12331 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
12333 Revert r251800 and r251799.
12335 2017-09-11 Martin Jambor <mjambor@suse.cz>
12338 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
12339 arguments in advance.
12340 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
12341 use it to find predecessor edges.
12342 (naive_outof_ssa): Collect vector of predecessors.
12344 2017-09-08 Jason Merrill <jason@redhat.com>
12346 PR c++/70029 - ICE with ref-qualifier and -flto
12347 * langhooks.h (struct lang_hooks_for_types): Add
12348 copy_lang_qualifiers.
12349 * attribs.c (build_type_attribute_qual_variant): Use it.
12350 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
12352 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
12353 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
12355 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
12358 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
12359 (*mulsi3_sp64): New instruction.
12360 (mulsi3): New expander.
12362 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
12364 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
12366 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12368 * sancov.c: Include memmodel.h.
12370 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
12373 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
12376 2017-09-07 Carl Love <cel@us.ibm.com>
12378 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
12379 the sldi instruction.
12381 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
12383 * sancov.c: Include tm_p.h.
12385 2017-09-07 Jakub Jelinek <jakub@redhat.com>
12388 * output.h (switch_to_other_text_partition): New declaration.
12389 * varasm.c (switch_to_other_text_partition): New function.
12390 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
12391 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
12392 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
12393 to the other text partition before emitting LCL label and switch back
12394 after emitting the word after it.
12396 2017-09-07 Richard Biener <rguenther@suse.de>
12398 * passes.def (pass_split_crit_edges): Remove instance before PRE.
12399 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
12400 critical edges here, after loop init.
12401 (pass_data_pre): Remove PROP_no_crit_edges flags.
12402 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
12403 for valueization of call args to avoid leaking VN_TOP.
12404 (visit_use): Assert we do not visit default defs.
12405 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
12406 Use error_mark_node to more easily detect leaking VN_TOP.
12407 All default-defs are varying, not VN_TOP. Mark them visited.
12408 (run_scc_vn): Make code match comment.
12410 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12412 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
12413 OPTION_MASK_FLOAT128_KEYWORD.
12414 (POWERPC_MASKS): Likewise.
12415 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
12416 support for the -mfloat128-type option, and make -mfloat128
12417 default on PowerPC Linux systems. Define or undefine
12418 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
12419 Define __float128 to be __ieee128 if IEEE 128-bit support is
12420 enabled, or undefine it.
12421 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
12422 Delete defining __FLOAT128_TYPE__.
12423 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
12424 -mfloat128-type option and make -mfloat128 default on PowerPC
12426 (TARGET_FLOAT128_TYPE): Likewise.
12427 (-mfloat128-type): Likewise.
12428 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12429 Delete the -mfloat128-type option and make -mfloat128 default on
12430 PowerPC Linux systems. Always use __ieee128 to be the keyword for
12431 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
12432 128-bit floating point is enabled. Change tests from using
12433 -mfloat128-type to -mfloat128.
12434 (rs6000_mangle_type): Use the correct mangling for the __float128
12435 type even if normal long double is restricted to 64-bits.
12436 (floatn_mode): Enable the _Float128 type by default on VSX Linux
12438 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
12439 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
12440 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
12442 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
12443 documentation for -mfloat128.
12445 2017-09-06 Olivier Hainque <hainque@adacore.com>
12447 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
12449 2017-09-06 Wish Wu <wishwu007@gmail.com>
12450 Jakub Jelinek <jakub@redhat.com>
12452 * asan.c (initialize_sanitizer_builtins): Add
12453 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
12454 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
12455 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
12456 BT_FN_VOID_UINT64_PTR variables.
12457 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
12458 (BT_FN_VOID_UINT16_UINT16): Likewise.
12459 (BT_FN_VOID_UINT32_UINT32): Likewise.
12460 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
12461 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
12462 (BT_FN_VOID_UINT64_PTR): Likewise.
12463 * common.opt (flag_sanitize_coverage): New variable.
12464 (fsanitize-coverage=trace-pc): Remove.
12465 (fsanitize-coverage=): Add.
12466 * flag-types.h (enum sanitize_coverage_code): New enum.
12467 * fold-const.c (fold_range_test): Disable non-short-circuit
12468 optimization if flag_sanitize_coverage.
12469 (fold_truth_andor): Likewise.
12470 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12471 * opts.c (COVERAGE_SANITIZER_OPT): Define.
12472 (coverage_sanitizer_opts): New array.
12473 (get_closest_sanitizer_option): Add OPTS argument, handle also
12474 OPT_fsanitize_coverage_.
12475 (parse_sanitizer_options): Adjusted to also handle
12476 OPT_fsanitize_coverage_.
12477 (common_handle_option): Add OPT_fsanitize_coverage_.
12478 * sancov.c (instrument_comparison, instrument_switch): New function.
12479 (sancov_pass): Add trace-cmp support.
12480 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
12481 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
12482 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
12483 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
12484 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
12485 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
12486 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
12487 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
12488 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
12489 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
12491 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12493 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12494 error. Only quit immediately if parsing is complete.
12495 (BEGIN): Initialize fatal_err and parse_done.
12496 (begin fpu, end fpu): Check number of arguments.
12497 (begin arch, end arch): Likewise.
12498 (begin cpu, end cpu): Likewise.
12499 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12500 (optalias): Likewise.
12502 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12504 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12505 * config/arm/arm-isa.h: Delete. Move definitions to ...
12506 * arm-cpus.in: ... here. Use new feature and fgroup values.
12507 * config/arm/arm.c (arm_option_override): Use lower case for feature
12509 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12510 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12511 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12513 (print_isa_bits_for): New function.
12514 (gen_isa): New function.
12515 (gen_comm_data): Use print_isa_bits_for.
12516 (define feature): New keyword.
12517 (define fgroup): New keyword.
12518 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
12519 (arm-isa.h): Add rule to generate file.
12520 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12521 case for feature bit names.
12523 2017-09-06 Richard Biener <rguenther@suse.de>
12525 * tree-ssa-pre.c (NECESSARY): Remove.
12526 (create_expression_by_pieces): Do not touch pass-local flags.
12527 (insert_into_preds_of_block): Likewise.
12528 (do_pre_regular_insertion): Likewise.
12529 (eliminate_insert): Likewise.
12530 (eliminate_dom_walker::before_dom_children): Likewise.
12531 (fini_eliminate): Do not look at inserted_exprs.
12532 (mark_operand_necessary): Remove.
12533 (remove_dead_inserted_code): Replace with simple work-list
12534 algorithm based on inserted_exprs and SSA uses.
12535 (pass_pre::execute): Re-order fini_eliminate and
12536 remove_dead_inserted_code.
12538 2017-09-06 Olivier Hainque <hainque@adacore.com>
12540 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
12541 for VxWorks 7. Adjust surrounding comments.
12543 2017-09-06 Richard Biener <rguenther@suse.de>
12545 * gimple-ssa-strength-reduction.c
12546 (find_candidates_dom_walker::before_dom_children): Also allow
12549 2017-09-06 Richard Biener <rguenther@suse.de>
12551 PR tree-optimization/82108
12552 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
12553 for gap in the non-permutation SLP case.
12555 2017-09-06 Martin Jambor <mjambor@suse.cz>
12557 PR tree-optimization/82078
12558 * tree-sra.c (sort_and_splice_var_accesses): Move call to
12559 add_access_to_work_queue...
12560 (build_accesses_from_assign): ...here.
12561 (propagate_all_subaccesses): Make sure racc is the group
12562 representative, if there is one.
12564 2017-09-06 Jakub Jelinek <jakub@redhat.com>
12566 PR middle-end/82095
12567 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
12570 2017-09-06 Richard Biener <rguenther@suse.de>
12572 * gimple-ssa-strength-reduction.c
12573 (find_candidates_dom_walker::before_doom_children): Use a
12574 type and not a mode check.
12576 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
12579 * config/arm/predicates.md (arm_general_adddi_operand): Create new
12581 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
12583 2017-09-05 Jeff Law <law@redhat.com>
12585 PR tree-optimization/64910
12586 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
12587 cases where we have 3 or more operands.
12589 2017-09-05 Jakub Jelinek <jakub@redhat.com>
12591 PR middle-end/81768
12592 * omp-low.c (lower_omp_for): Recompute tree invariant if
12593 gimple_omp_for_initial/final is ADDR_EXPR.
12595 PR middle-end/81768
12596 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
12597 into gimple val before gimplification fo the COND_EXPR.
12599 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
12601 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
12602 REGION_COPY argument.
12603 (thread_through_all_blocks): Remove unused argument to
12604 duplicate_thread_path.
12606 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12607 Alan Hayward <alan.hayward@arm.com>
12608 David Sherwood <david.sherwood@arm.com>
12610 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
12611 Take a scalar_mode rather than a machine_mode.
12612 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12613 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
12614 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12615 (aarch64_gen_adjusted_ldpstp): Likewise.
12616 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
12618 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12619 Alan Hayward <alan.hayward@arm.com>
12620 David Sherwood <david.sherwood@arm.com>
12622 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
12623 Take a scalar_int_mode instead of a machine_mode.
12624 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12625 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12626 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12627 (aarch64_simd_attr_length_rglist): Delete.
12628 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
12629 a scalar_int_mode instead of a machine_mode.
12630 (aarch64_add_offset): Likewise.
12631 (aarch64_internal_mov_immediate): Likewise
12632 (aarch64_add_constant_internal): Likewise.
12633 (aarch64_add_constant): Likewise.
12634 (aarch64_movw_imm): Likewise.
12635 (aarch64_rtx_arith_op_extract_p): Likewise.
12636 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12637 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12638 Remove assert that the mode isn't a vector.
12639 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12640 (aarch64_expand_mov_immediate): Update calls after above changes.
12641 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
12642 (aarch64_and_bitmask_imm): Check for scalar integer modes.
12643 (aarch64_move_imm): Likewise.
12644 (aarch64_can_const_movi_rtx_p): Likewise.
12645 (aarch64_strip_extend): Likewise.
12646 (aarch64_extr_rtx_p): Likewise.
12647 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
12648 a CONST_INT when the mode parameter is VOIDmode.
12649 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
12651 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12653 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
12654 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12655 (bitwise_type_for_mode): Update accordingly.
12657 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12659 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
12660 * stor-layout.c (mode_for_size_tree): Likewise.
12661 (mode_for_array): Update accordingly.
12662 (layout_decl): Likewise.
12663 (compute_record_mode): Likewise. Only set the mode once.
12665 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12667 * target.def (get_mask_mode): Change return type to opt_mode.
12669 * doc/tm.texi: Regenerate.
12670 * targhooks.h (default_get_mask_mode): Return an opt_mode.
12671 * targhooks.c (default_get_mask_mode): Likewise.
12672 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12673 * optabs-query.c (can_vec_mask_load_store_p): Update use of
12674 targetm.get_mask_mode.
12675 * tree.c (build_truth_vector_type): Likewise.
12677 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12679 * machmode.h (mode_for_vector): Return an opt_mode.
12680 * stor-layout.c (mode_for_vector): Likewise.
12681 (mode_for_int_vector): Update accordingly.
12682 (layout_type): Likewise.
12683 * config/i386/i386.c (emit_memmov): Likewise.
12684 (ix86_expand_set_or_movmem): Likewise.
12685 (ix86_expand_vector_init): Likewise.
12686 (ix86_get_mask_mode): Likewise.
12687 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
12689 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
12690 * expmed.c (extract_bit_field_1): Likewise.
12691 * expr.c (expand_expr_real_2): Likewise.
12692 * optabs-query.c (can_vec_perm_p): Likewise.
12693 (can_vec_mask_load_store_p): Likewise.
12694 * optabs.c (expand_vec_perm): Likewise.
12695 * targhooks.c (default_get_mask_mode): Likewise.
12696 * tree-vect-stmts.c (vectorizable_store): Likewise.
12697 (vectorizable_load): Likewise.
12698 (get_vectype_for_scalar_type_and_size): Likewise.
12700 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12702 * machmode.h (mode_for_int_vector): New function.
12703 * stor-layout.c (mode_for_int_vector): Likewise.
12704 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
12705 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
12706 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12707 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12708 (s390_expand_vcond): Likewise.
12710 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12712 * machmode.h (opt_machine_mode): New type.
12713 (opt_mode<T>): Allow construction from anything that can be
12715 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
12716 (mode_for_size): Return an opt_machine_mode.
12717 * stor-layout.c (mode_for_size): Likewise.
12718 (mode_for_size_tree): Update call accordingly.
12719 (bitwise_mode_for_mode): Likewise.
12720 (make_fract_type): Likewise.
12721 (make_accum_type): Likewise.
12722 * caller-save.c (replace_reg_with_saved_mem): Update call
12724 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12725 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12726 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12727 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12728 * expmed.c (extract_bit_field_1): Likewise.
12729 * reload.c (get_secondary_mem): Likewise.
12730 * varasm.c (assemble_integer): Likewise.
12731 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
12734 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12736 * machmode.h (decimal_float_mode_for_size): New function.
12737 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
12738 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
12739 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
12740 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
12741 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
12742 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
12744 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12746 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
12747 (get_builtin_sync_mode): Likewise.
12748 (expand_ifn_atomic_compare_exchange): Likewise.
12749 (expand_builtin_atomic_clear): Likewise.
12750 (expand_builtin_atomic_test_and_set): Likewise.
12751 (fold_builtin_atomic_always_lock_free): Likewise.
12752 * calls.c (compute_argument_addresses): Likewise.
12753 (emit_library_call_value_1): Likewise.
12754 (store_one_arg): Likewise.
12755 * combine.c (combine_instructions): Likewise.
12756 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
12757 * config/arm/arm.c (arm_function_value): Likewise.
12758 (aapcs_allocate_return_reg): Likewise.
12759 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
12760 * config/i386/i386.c (construct_container): Likewise.
12761 (ix86_gimplify_va_arg): Likewise.
12762 (ix86_expand_sse_cmp): Likewise.
12763 (emit_memmov): Likewise.
12764 (emit_memset): Likewise.
12765 (expand_small_movmem_or_setmem): Likewise.
12766 (ix86_expand_pextr): Likewise.
12767 (ix86_expand_pinsr): Likewise.
12768 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
12769 * config/microblaze/microblaze.c (microblaze_block_move_straight):
12771 * config/mips/mips.c (mips_function_value_1) Likewise.
12772 (mips_block_move_straight): Likewise.
12773 (mips_expand_ins_as_unaligned_store): Likewise.
12774 * config/powerpcspe/powerpcspe.c
12775 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12776 (rs6000_darwin64_record_arg_flush): Likewise.
12777 * config/rs6000/rs6000.c
12778 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12779 (rs6000_darwin64_record_arg_flush): Likewise.
12780 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
12781 (sparc_function_value_1): Likewise.
12782 * config/spu/spu.c (adjust_operand): Likewise.
12783 (spu_emit_branch_or_set): Likewise.
12784 (arith_immediate_p): Likewise.
12785 * emit-rtl.c (gen_lowpart_common): Likewise.
12786 * expr.c (expand_expr_real_1): Likewise.
12787 * function.c (assign_parm_setup_block): Likewise.
12788 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
12789 * reload1.c (alter_reg): Likewise.
12790 * stor-layout.c (mode_for_vector): Likewise.
12791 (layout_type): Likewise.
12793 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12795 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
12796 (spu_convert_move): Likewise.
12797 * lower-subreg.c (resolve_simple_move): Likewise.
12799 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12802 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
12803 define_insn to a define_expand.
12804 (altivec_vsum2sws_direct): New define_insn.
12805 (altivec_vsumsws): Convert from a define_insn to a define_expand.
12807 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12809 * config/arm/arm.c (arm_option_params_internal): Improve setting of
12812 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
12817 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
12818 realign stack if stack alignment needed is less than incoming
12821 2017-09-05 Marek Polacek <polacek@redhat.com>
12824 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
12827 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12829 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
12831 2017-09-05 Richard Biener <rguenther@suse.de>
12833 PR tree-optimization/82084
12834 * fold-const.c (can_native_encode_string_p): Handle wide characters.
12836 2017-09-05 Richard Biener <rguenther@suse.de>
12838 PR tree-optimization/82102
12839 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
12841 2017-09-05 Martin Liska <mliska@suse.cz>
12843 PR tree-optimization/82032
12844 * tree-cfg.c (generate_range_test): New function.
12845 * tree-cfg.h (generate_range_test): Declared here.
12846 * tree-cfgcleanup.c (convert_single_case_switch): New function.
12847 (cleanup_control_expr_graph): Use it.
12848 * tree-switch-conversion.c (try_switch_expansion): Remove
12850 (emit_case_nodes): Use generate_range_test.
12852 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12855 * config/i386/i386.md (*<btsc><mode>_mask): Add
12856 TARGET_USE_BT to insn constraint.
12857 (*btr<mode>_mask): Ditto.
12859 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
12861 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
12862 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
12864 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
12867 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
12868 TARGET_NEON and TARGET_IWMMXT.
12869 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
12870 TARGET_NEON and TARGET_IWMMXT.
12871 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
12873 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12875 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
12876 (ix86_rewrite_tls_address): Ditto.
12877 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
12878 (ix86_rewrite_tls_address_1): Ditto.
12879 (ix86_rewrite_tls_address): Ditto.
12880 * config/i386/predicates.md (tls_address_pattern): New predicate.
12881 * config/i386/i386.md (TLS address splitter): New splitter.
12883 2017-09-04 Richard Biener <rguenther@suse.de>
12885 PR tree-optimization/82084
12886 * fold-const.h (can_native_encode_string_p): Declare.
12887 * fold-const.c (can_native_encode_string_p): Factor out from ...
12888 (native_encode_string): ... here.
12889 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
12890 vectorizing stores from constants we later cannot handle.
12892 2017-09-04 Marek Polacek <polacek@redhat.com>
12895 * doc/invoke.texi: Update -Wtautological-compare documentation.
12897 2017-09-04 Jeff Law <law@redhat.com>
12899 PR tree-optimization/64910
12900 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
12901 swap the first and last operand if the last is a constant.
12903 2017-09-04 Marek Polacek <polacek@redhat.com>
12906 * convert.c (do_narrow): When sanitizing signed integer overflows,
12907 bail out for signed types.
12908 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
12910 2017-09-04 Richard Biener <rguenther@suse.de>
12912 PR tree-optimization/82060
12913 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12914 Move devirtualization after stmt folding and before EH/AB/noreturn
12915 cleanup to get the stmt refs canonicalized. Use a bool instead
12916 of gimple_modified_p since that doesn't work for NOPs. Schedule
12917 NOPs generated by folding for removal.
12919 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12920 Alan Hayward <alan.hayward@arm.com>
12921 David Sherwood <david.sherwood@arm.com>
12923 * coretypes.h (pad_direction): New enum.
12924 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
12925 (FUNCTION_ARG_PADDING): Likewise.
12926 * target.def (function_arg_padding): New hook.
12927 * targhooks.h (default_function_arg_padding): Declare.
12928 * targhooks.c (default_function_arg_padding): New function.
12929 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
12930 (TARGET_FUNCTION_ARG_PADDING): ...this.
12931 * doc/tm.texi: Regenerate.
12932 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
12933 instead of direction.
12934 (compute_argument_addresses): Likewise.
12935 (load_register_parameters): Likewise.
12936 (emit_library_call_value_1): Likewise.
12937 (store_one_arg): Use targetm.calls.function_arg_padding instead
12938 of FUNCTION_ARG_PADDING.
12939 (must_pass_in_stack_var_size_or_pad): Likewise.
12940 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
12941 (emit_group_store): Likewise.
12942 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
12943 instead of FUNCTION_ARG_PADDING.
12944 (emit_push_insn): Likewise, and propagate enum change throughout
12946 * function.h (direction): Delete.
12947 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
12949 * function.c (assign_parm_find_stack_rtl): Likewise.
12950 (assign_parm_setup_block_p): Likewise.
12951 (assign_parm_setup_block): Likewise.
12952 (gimplify_parameters): Likewise.
12953 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
12954 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
12956 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
12957 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12958 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
12959 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
12960 (aarch64_function_arg_padding): ...this new function.
12961 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
12962 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12963 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
12964 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12965 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
12966 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12967 (arm_pad_arg_upward): Replace with...
12968 (arm_function_arg_padding): ...this new function.
12969 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
12971 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
12972 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
12973 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12974 (ia64_hpux_function_arg_padding): Replace with...
12975 (ia64_function_arg_padding): ...this new function. Use pad_direction
12976 instead of direction. Check for TARGET_HPUX.
12977 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
12978 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12979 (iq2000_function_arg_padding): New function.
12980 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
12981 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
12982 (mips_function_arg_padding): ...this new function.
12983 (mips_pad_reg_upward): Update accordingly.
12984 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12985 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
12986 targetm.calls.function_arg_padding.
12987 (FUNCTION_ARG_PADDING): Delete.
12988 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12989 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
12990 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12991 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
12992 (nios2_block_reg_padding): Return pad_direction instead of direction.
12993 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
12994 instead of direction.
12995 (nios2_function_arg_padding): Likewise. Make static.
12996 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12997 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
12998 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
12999 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13000 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13001 (pa_function_arg_padding): Make static. Return pad_direction instead
13003 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13004 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13005 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13006 instead of direction. Use targetm.calls.function_arg_padding.
13007 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13008 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13009 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13010 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13011 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13013 (function_arg_padding): Rename to...
13014 (rs6000_function_arg_padding): ...this. Make static. Return
13015 pad_direction instead of direction.
13016 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13017 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13018 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13019 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13020 instead of direction. Use targetm.calls.function_arg_padding.
13021 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13022 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13023 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13024 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13025 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13026 (function_arg_padding): Rename to...
13027 (rs6000_function_arg_padding): ...this. Make static. Return
13028 pad_direction instead of direction.
13029 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13030 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13031 * config/s390/s390.c (s390_function_arg_padding): New function.
13032 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13033 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13034 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13035 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13036 (function_arg_padding): Rename to...
13037 (sparc_function_arg_padding): ...this. Make static. Return
13038 pad_direction instead of direction.
13039 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13040 * config/spu/spu.c (spu_function_arg_padding): New function.
13041 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13042 * system.h (FUNCTION_ARG_PADDING): Poison.
13044 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13045 Alan Hayward <alan.hayward@arm.com>
13046 David Sherwood <david.sherwood@arm.com>
13048 * target.def (modes_tieable_p): New hook.
13049 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13050 (TARGET_MODES_TIEABLE_P): ...this.
13051 * doc/tm.texi.in: Regenerate.
13052 * hooks.h (hook_bool_mode_mode_true): Declare.
13053 * hooks.c (hook_bool_mode_mode_true): New function.
13054 * combine.c (subst): Use targetm.modes_tieable_p instead of
13056 * dse.c (find_shift_sequence): Likewise.
13057 * expmed.c (extract_low_bits): Likewise.
13058 * lower-subreg.c: Include target.h.
13059 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13061 * rtlanal.c (rtx_cost): Likewise.
13062 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13063 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13064 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13065 (TARGET_MODES_TIEABLE_P): Redefine.
13066 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13067 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13068 (TARGET_MODES_TIEABLE_P): Redefine.
13069 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13070 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13071 (arc_modes_tieable_p): New function.
13072 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13073 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13074 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13075 (arm_modes_tieable_p): Make static.
13076 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13077 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13078 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13079 (TARGET_MODES_TIEABLE_P): Redefine.
13080 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13081 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13082 (TARGET_MODES_TIEABLE_P): Redefine.
13083 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13084 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13085 (cr16_modes_tieable_p): New function.
13086 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13087 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13088 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13089 (TRULY_NOOP_TRUNCATION): Update comment.
13090 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13091 (TRULY_NOOP_TRUNCATION): Update comment.
13092 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13093 (frv_modes_tieable_p): New function.
13094 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13095 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13096 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13097 (TARGET_MODES_TIEABLE_P): Redefine.
13098 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13099 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13100 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13101 (TARGET_MODES_TIEABLE_P): Redefine.
13102 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13103 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13104 (ia64_modes_tieable_p): New function.
13105 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13106 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13107 (iq2000_modes_tieable_p): New function.
13108 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13109 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13110 (lm32_modes_tieable_p): New function.
13111 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13112 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13113 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13114 (TARGET_MODES_TIEABLE_P): Redefine.
13115 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13116 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13117 (m32r_modes_tieable_p): New function.
13118 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13119 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13120 (m68k_modes_tieable_p): New function.
13121 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13122 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13123 (mcore_modes_tieable_p): New function.
13124 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13125 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13127 (TARGET_MODES_TIEABLE_P): Redefine.
13128 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13129 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13130 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13131 (TARGET_MODES_TIEABLE_P): Redefine.
13132 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13133 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13134 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13135 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13136 (mn10300_modes_tieable_p): ...this and make static.
13137 (TARGET_MODES_TIEABLE_P): Redefine.
13138 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13139 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13140 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13141 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13142 (msp430_modes_tieable_p): Make static.
13143 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13144 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13145 (TARGET_MODES_TIEABLE_P): Redefine.
13146 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13147 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13148 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13149 (TARGET_MODES_TIEABLE_P): Redefine.
13150 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13151 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13152 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13153 (TARGET_MODES_TIEABLE_P): Redefine.
13154 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13155 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13156 (pdp11_modes_tieable_p): New function.
13157 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13158 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13159 (rs6000_modes_tieable_p): New function.
13160 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13161 * config/powerpcspe/powerpcspe.md: Update comment.
13162 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13163 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13164 (TARGET_MODES_TIEABLE_P): Redefine.
13165 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13166 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13167 (rl78_modes_tieable_p): New function.
13168 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13169 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13170 (rs6000_modes_tieable_p): New function.
13171 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13172 * config/rs6000/rs6000.md: Update comment.
13173 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13174 * config/rx/rx.c (rx_modes_tieable_p): New function.
13175 (TARGET_MODES_TIEABLE_P): Redefine.
13176 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13177 * config/s390/s390.c (s390_modes_tieable_p): New function.
13178 (TARGET_MODES_TIEABLE_P): Redefine.
13179 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13180 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13181 (sh_modes_tieable_p): New function.
13182 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13183 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13184 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13185 (sparc_modes_tieable_p): Make static.
13186 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13187 * config/spu/spu.c (spu_modes_tieable_p): New function.
13188 (TARGET_MODES_TIEABLE_P): Redefine.
13189 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13190 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13191 (TARGET_MODES_TIEABLE_P): Redefine.
13192 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13193 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13194 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13195 * config/v850/v850.c (v850_modes_tieable_p): New function.
13196 (TARGET_MODES_TIEABLE_P): Redefine.
13197 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13198 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13199 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13200 (visium_modes_tieable_p): New function.
13201 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13202 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13203 (xtensa_modes_tieable_p): New function.
13204 * system.h (MODES_TIEABLE_P): Poison.
13206 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13207 Alan Hayward <alan.hayward@arm.com>
13208 David Sherwood <david.sherwood@arm.com>
13210 * target.def (hard_regno_mode_ok): New hook.
13211 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
13212 (TARGET_HARD_REGNO_MODE_OK): ...this.
13213 * doc/tm.texi.in: Regenerate.
13214 * hooks.h (hook_bool_uint_mode_true): Declare.
13215 * hooks.c (hook_bool_uint_mode_true): New function.
13216 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
13217 HARD_REGNO_MODE_OK.
13218 * genpreds.c (write_insn_preds_c): Add an include of target.h.
13219 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
13220 instead of HARD_REGNO_MODE_OK.
13221 * caller-save.c: Include target.h.
13222 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
13223 HARD_REGNO_MODE_OK.
13224 * combine.c (can_combine_p): Likewise.
13225 (combinable_i3pat): Likewise.
13226 (can_change_dest_mode): Likewise.
13227 * expr.c (init_expr_target): Likewise.
13228 (convert_move): Likewise.
13229 (convert_modes): Likewise.
13230 * ira.c (setup_prohibited_class_mode_regs): Likewise.
13231 (setup_prohibited_mode_move_regs): Likewise.
13232 * ira.h (target_ira): Likewise.
13233 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13234 * lra-constraints.c (process_alt_operands): Likewise.
13235 (split_reg): Likewise.
13236 * recog.c (peep2_find_free_register): Likewise.
13237 * ree.c (combine_reaching_defs): Likewise.
13238 * regcprop.c (maybe_mode_change): Likewise.
13239 * reginfo.c (init_reg_sets_1): Likewise.
13240 (choose_hard_reg_mode): Likewise.
13241 (simplifiable_subregs): Likewise.
13242 * regrename.c (check_new_reg_p): Likewise.
13243 * reload.c (find_valid_class): Likewise.
13244 (find_valid_class_1): Likewise.
13245 (reload_inner_reg_of_subreg): Likewise.
13246 (push_reload): Likewise.
13247 (combine_reloads): Likewise.
13248 (find_dummy_reload): Likewise.
13249 (find_reloads): Likewise.
13250 * reload1.c (find_reg): Likewise.
13251 (set_reload_reg): Likewise.
13252 (allocate_reload_reg): Likewise.
13253 (choose_reload_regs): Likewise.
13254 (reload_adjust_reg_for_temp): Likewise.
13255 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
13256 (simplify_subreg_regno): Likewise.
13257 * sel-sched.c (init_regs_for_mode): Likewise.
13258 * varasm.c (make_decl_rtl): Likewise.
13259 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
13260 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
13261 HARD_REGNO_MODE_OK.
13262 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
13263 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
13264 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13265 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
13266 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
13267 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13268 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
13269 (arc_mode_class): Delete.
13270 (HARD_REGNO_MODE_OK): Delete.
13271 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13272 (arc_hard_regno_mode_ok): Rename old array to...
13273 (arc_hard_regno_mode_ok_modes): ...this.
13274 (arc_conditional_register_usage): Update accordingly.
13275 (arc_mode_class): Make static.
13276 (arc_hard_regno_mode_ok): New function.
13277 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
13278 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
13279 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13280 (arm_hard_regno_mode_ok): Make static.
13281 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
13282 HARD_REGNO_MODE_OK.
13283 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
13284 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
13285 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
13287 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13288 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
13289 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
13290 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
13291 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
13292 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13293 * config/bfin/predicates.md (valid_reg_operand): Use
13294 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
13295 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
13296 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
13297 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13298 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
13299 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
13300 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13301 (cr16_hard_regno_mode_ok): Make static and return a bool.
13302 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
13303 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13304 (cris_hard_regno_mode_ok): New function.
13305 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
13306 (epiphany_mode_class): Delete.
13307 (HARD_REGNO_MODE_OK): Delete.
13308 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
13309 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13310 (hard_regno_mode_ok): Rename to...
13311 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
13312 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
13313 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
13314 HARD_REGNO_MODE_OK.
13315 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
13316 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
13317 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13318 (frv_hard_regno_mode_ok): Make static and return a bool.
13319 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
13320 HARD_REGNO_MODE_OK.
13321 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
13322 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
13323 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
13324 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
13326 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13327 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
13328 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
13329 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
13331 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13332 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
13333 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13334 (ia64_hard_regno_mode_ok): New function.
13335 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
13336 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13337 (iq2000_hard_regno_mode_ok): New function.
13338 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
13339 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13340 (lm32_hard_regno_mode_ok): New function.
13341 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
13342 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
13343 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
13344 instead of HARD_REGNO_MODE_OK.
13345 (m32c_hard_regno_ok): Rename to...
13346 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
13347 (m32c_cannot_change_mode_class): Update accordingly.
13348 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13349 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
13350 (m32r_mode_class): Delete.
13351 (HARD_REGNO_MODE_OK): Delete.
13352 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13353 (m32r_hard_regno_mode_ok): Rename to...
13354 (m32r_hard_regno_modes): ...this.
13355 (m32r_mode_class): Make static.
13356 (m32r_hard_regno_mode_ok): New function.
13357 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
13358 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
13359 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13360 (m68k_hard_regno_mode_ok): Make static.
13361 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
13362 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13363 (mcore_hard_regno_mode_ok): New function.
13364 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
13365 (HARD_REGNO_MODE_OK): Delete.
13366 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
13368 (microblaze_hard_regno_mode_ok_p): ...this and make static.
13369 (microblaze_hard_regno_mode_ok): New function.
13370 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13371 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
13372 (mips_hard_regno_mode_ok): Delete.
13373 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
13374 (mips_hard_regno_mode_ok_p): ...this and make static.
13375 (mips_hard_regno_mode_ok_p): Rename to...
13376 (mips_hard_regno_mode_ok_uncached): ...this.
13377 (mips_hard_regno_mode_ok): New function.
13378 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
13379 of HARD_REGNO_MODE_OK.
13380 (mips_option_override): Update after above name changes.
13381 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13382 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
13383 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
13384 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
13385 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
13386 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13387 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
13388 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
13389 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
13390 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13391 (msp430_hard_regno_mode_ok): Make static and return a bool.
13392 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
13393 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
13394 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
13396 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13397 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
13398 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
13399 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
13400 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
13401 (PA_HARD_REGNO_MODE_OK): ...this
13402 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
13403 (PA_HARD_REGNO_MODE_OK): ...this.
13404 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13405 (pa_hard_regno_mode_ok): New function.
13406 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
13407 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13408 (pdp11_hard_regno_mode_ok): New function.
13409 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
13410 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
13412 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
13414 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13415 (rs6000_hard_regno_mode_ok): Rename to...
13416 (rs6000_hard_regno_mode_ok_uncached): ...this.
13417 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13418 (rs6000_hard_regno_mode_ok): New function.
13419 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
13420 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
13421 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
13422 (riscv_hard_regno_mode_ok): ...this and make static.
13423 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13424 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
13425 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
13426 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13427 (rl78_hard_regno_mode_ok): Make static and return bool.
13428 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
13429 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
13431 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
13432 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13433 (rs6000_hard_regno_mode_ok): Rename to...
13434 (rs6000_hard_regno_mode_ok_uncached): ...this.
13435 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13436 (rs6000_hard_regno_mode_ok): New function.
13437 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
13438 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
13439 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13440 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
13441 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
13442 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
13443 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13444 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
13445 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
13446 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13447 (sh_hard_regno_mode_ok): Make static.
13448 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
13449 instead of HARD_REGNO_MODE_OK.
13450 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
13451 (sparc_mode_class): Delete.
13452 (HARD_REGNO_MODE_OK): Delete.
13453 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13454 (hard_regno_mode_classes): Make static.
13455 (sparc_mode_class): Likewise.
13456 (sparc_hard_regno_mode_ok): New function.
13457 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
13458 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
13459 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
13461 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13462 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
13463 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
13464 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
13465 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
13466 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13467 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
13468 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
13469 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13470 (visium_hard_regno_mode_ok): New function.
13471 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
13472 instead of HARD_REGNO_MODE_OK.
13473 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
13474 (HARD_REGNO_MODE_OK): Delete.
13475 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
13476 (xtensa_hard_regno_mode_ok_p): ...this and make static.
13477 (xtensa_option_override): Update accordingly.
13478 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13479 (xtensa_hard_regno_mode_ok): New function.
13480 * system.h (HARD_REGNO_MODE_OK): Poison.
13482 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13483 Alan Hayward <alan.hayward@arm.com>
13484 David Sherwood <david.sherwood@arm.com>
13486 * target.def (hard_regno_call_part_clobbered): New hook.
13487 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
13488 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
13489 * doc/tm.texi: Regenerate.
13490 * hooks.h (hook_bool_uint_mode_false): Declare.
13491 * hooks.c (hook_bool_uint_mode_false): New function.
13492 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13493 * cselib.c (cselib_process_insn): Use
13494 targetm.hard_regno_call_part_clobbered instead of
13495 HARD_REGNO_CALL_PART_CLOBBERED.
13496 * ira-conflicts.c (ira_build_conflicts): Likewise.
13497 * ira-costs.c (ira_tune_allocno_costs): Likewise.
13498 * lra-constraints.c (need_for_call_save_p): Likewise.
13499 * lra-lives.c: Include target.h.
13500 (check_pseudos_live_through_calls): Use
13501 targetm.hard_regno_call_part_clobbered instead of
13502 HARD_REGNO_CALL_PART_CLOBBERED.
13503 * regcprop.c: Include target.h.
13504 (copyprop_hardreg_forward_1): Use
13505 targetm.hard_regno_call_part_clobbered instead of
13506 HARD_REGNO_CALL_PART_CLOBBERED.
13507 * reginfo.c (choose_hard_reg_mode): Likewise.
13508 * regrename.c (check_new_reg_p): Likewise.
13509 * reload.c (find_equiv_reg): Likewise.
13510 * reload1.c (emit_reload_insns): Likewise.
13511 * sched-deps.c (deps_analyze_insn): Likewise.
13512 * sel-sched.c (init_regs_for_mode): Likewise.
13513 (mark_unavailable_hard_regs): Likewise.
13514 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
13515 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13516 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
13518 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13519 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13520 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
13522 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
13524 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13525 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13526 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
13528 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13529 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13530 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
13532 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13533 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
13535 * config/powerpcspe/powerpcspe.c
13536 (rs6000_hard_regno_call_part_clobbered): New function.
13537 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13538 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13539 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
13541 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13542 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13543 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
13545 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13546 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13547 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
13549 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13550 Alan Hayward <alan.hayward@arm.com>
13551 David Sherwood <david.sherwood@arm.com>
13553 * rtl.h (subreg_memory_offset): Declare.
13554 * emit-rtl.c (subreg_memory_offset): New function.
13555 * expmed.c (store_bit_field_1): Use it.
13556 * expr.c (undefined_operand_subword_p): Likewise.
13557 * simplify-rtx.c (simplify_subreg): Likewise.
13559 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
13561 PR rtl-optimization/57448
13564 * optabs.c (expand_atomic_load): Place compiler memory barriers if
13565 using atomic_load pattern.
13566 (expand_atomic_store): Likewise.
13568 2017-09-04 Jakub Jelinek <jakub@redhat.com>
13571 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
13572 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
13573 handling. Use replace_call_with_value with NULL instead of
13574 gsi_replace, unlink_stmt_vdef and release_defs.
13576 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
13579 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
13581 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13584 * rtl.h (emit_library_call_value_1): Declare.
13585 (emit_library_call): Replace declaration with a series of overloads.
13586 Remove the parameter count argument.
13587 (emit_library_call_value): Likewise.
13588 * calls.c (emit_library_call_value_1): Make global. Replace varargs
13589 with an "rtx_mode_t *".
13590 (emit_library_call_value): Delete.
13591 (emit_library_call): Likewise.
13592 * asan.c (asan_emit_stack_protection): Update calls accordingly.
13593 (asan_emit_allocas_unpoison): Likewise.
13594 * builtins.c (expand_builtin_powi): Likewise.
13595 (expand_asan_emit_allocas_unpoison): Likewise.
13596 * cfgexpand.c (expand_main_function): Likewise.
13597 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
13598 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
13599 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
13600 * config/arm/arm.c (arm_trampoline_init): Likewise.
13601 (arm_call_tls_get_addr): Likewise.
13602 (arm_expand_divmod_libfunc): Likewise.
13603 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
13604 (smulsi3_highpart): Likewise.
13605 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
13606 (c6x_expand_compare): Likewise.
13607 (c6x_expand_movmem): Likewise.
13608 * config/frv/frv.c (frv_trampoline_init): Likewise.
13609 * config/i386/i386.c (ix86_trampoline_init): Likewise.
13610 (ix86_expand_divmod_libfunc): Likewise.
13611 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
13612 (ia64_expand_compare): Likewise.
13613 (ia64_profile_hook): Likewise.
13614 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
13615 (nonlocal_goto): Likewise.
13616 (restore_stack_nonlocal): Likewise.
13617 * config/m32r/m32r.c (block_move_call): Likewise.
13618 (m32r_trampoline_init): Likewise.
13619 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
13620 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
13621 (m68k_call_m68k_read_tp): Likewise.
13622 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
13623 (microblaze_expand_divide): Likewise.
13624 * config/mips/mips.h (mips_args): Likewise.
13625 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
13626 (MIPS_ICACHE_SYNC): Likewise.
13627 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
13628 (nios2_trampoline_init): Likewise.
13629 * config/pa/pa.c (hppa_tls_call): Likewise.
13630 (pa_trampoline_init): Likewise.
13631 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
13632 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
13633 (expand_strn_compare): Likewise.
13634 (rs6000_generate_compare): Likewise.
13635 (rs6000_expand_float128_convert): Likewise.
13636 (output_profile_hook): Likewise.
13637 (rs6000_trampoline_init): Likewise.
13638 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
13639 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
13640 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
13641 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
13642 (rs6000_generate_compare): Likewise.
13643 (rs6000_expand_float128_convert): Likewise.
13644 (output_profile_hook): Likewise.
13645 (rs6000_trampoline_init): Likewise.
13646 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
13647 * config/sh/sh.c (sh_trampoline_init): Likewise.
13648 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
13649 (sparc_emit_float_lib_cmp): Likewise.
13650 (sparc32_initialize_trampoline): Likewise.
13651 (sparc64_initialize_trampoline): Likewise.
13652 (sparc_profile_hook): Likewise.
13653 * config/spu/spu.c (ea_load_store): Likewise.
13654 * config/spu/spu.md (floatunssidf2): Likewise.
13655 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
13656 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
13657 * config/visium/visium.c (expand_block_move_4): Likewise.
13658 (expand_block_move_2): Likewise.
13659 (expand_block_move_1): Likewise.
13660 (expand_block_set_4): Likewise.
13661 (expand_block_set_2): Likewise.
13662 (expand_block_set_1): Likewise.
13663 (visium_trampoline_init): Likewise.
13664 (visium_profile_hook): Likewise.
13665 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
13666 (xtensa_setup_frame_addresses): Likewise.
13667 (xtensa_trampoline_init): Likewise.
13668 * except.c (sjlj_emit_function_enter): Likewise.
13669 (sjlj_emit_function_exit): Likewise.
13670 * explow.c (allocate_dynamic_stack_space): Likewise.
13671 (probe_stack_range): Likewise.
13672 * expr.c (convert_mode_scalar): Likewise.
13673 * optabs.c (expand_binop): Likewise.
13674 (expand_twoval_binop_libfunc): Likewise.
13675 (expand_unop): Likewise.
13676 (prepare_cmp_insn): Likewise.
13677 (prepare_float_lib_cmp): Likewise.
13678 (expand_float): Likewise.
13679 (expand_fix): Likewise.
13680 (expand_fixed_convert): Likewise.
13681 (maybe_emit_sync_lock_test_and_set): Likewise.
13682 (expand_atomic_compare_and_swap): Likewise.
13683 (expand_mem_thread_fence): Likewise.
13684 (expand_atomic_fetch_op): Likewise.
13686 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
13688 * doc/generic.texi (OpenACC): Adjust URL.
13689 * doc/invoke.texi (C Dialect Options): Ditto.
13691 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
13693 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
13694 predicate for operand 1. Add (m,<S>) constraint.
13695 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
13696 Prevent memory operand 1 with register operand 2.
13698 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
13700 PR rtl-optimization/82024
13701 * combine.c (try_combine): If the combination result is a PARALLEL,
13702 and we only need to retain the SET in there that would be placed
13703 at I2, check that we can place that at I3 instead, before doing so.
13705 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13708 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
13710 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
13711 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
13714 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
13715 Jeff Law <law@redhat.com>
13717 * varasm.c (bss_initializer_p): Do not put constants into .bss
13718 (categorize_decl_for_section): Handle bss_initializer_p returning
13719 false when DECL_INITIAL is NULL.
13721 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13724 * config/s390/s390.c (s390_can_inline_p): New function.
13726 2017-09-01 Jeff Law <law@redhat.com>
13728 PR tree-optimization/82052
13729 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13730 Always initialize the returned slot after a hash table miss
13731 when INSERT is true.
13733 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
13735 * config/s390/s390.md (mem_signal_fence): Remove.
13736 * doc/md.texi (mem_signal_fence): Remove.
13737 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
13739 * target-insns.def (mem_signal_fence): Remove.
13741 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13744 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
13747 * asan.c (create_odr_indicator): Strip name encoding from assembler
13748 name before appending it after __odr_asan_.
13750 2017-09-01 Martin Liska <mliska@suse.cz>
13752 PR tree-optimization/82059
13753 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
13754 frequency only when an edge is redirected.
13756 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13758 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
13759 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
13760 (arc_conditional_register_usage): Remove ARC600 lp_count
13762 (arc_file_start): Emit Tag_ARC_CPU_variation.
13763 (arc_can_use_doloop_p): New conditions to use ZOLs.
13764 (hwloop_fail): New function.
13765 (hwloop_optimize): Likewise.
13766 (hwloop_pattern_reg): Likewise.
13767 (arc_doloop_hooks): New struct, to be used with reorg_loops.
13768 (arc_reorg_loops): New function, calls reorg_loops.
13769 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
13770 (arc600_corereg_hazard): Remove ZOL checking, case handled by
13772 (arc_loop_hazard): Remove function, functionality moved into
13774 (arc_hazard): Remove arc_loop_hazard call.
13775 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
13776 into hwloop_optimize.
13777 (arc_label_align): Remove ZOL handling.
13778 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
13779 * config/arc/arc.md (doloop_begin): Remove pattern.
13780 (doloop_begin_i): Likewise.
13781 (doloop_end_i): Likewise.
13782 (doloop_fallback): Likewise.
13783 (doloop_fallback_m): Likewise.
13784 (doloop_end): Reimplement expand.
13785 (arc_lp): New pattern for LP instruction.
13786 (loop_end): New pattern.
13787 (loop_fail): Likewise.
13788 (decrement_and_branch_until_zero): Likewise.
13789 * config/arc/arc.opt (mlpc-width): New option.
13790 * doc/invoke.texi (mlpc-width): Document option.
13792 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13794 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
13795 (arc_ccfsm_advance): Fix checking for delay slots.
13796 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
13798 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13800 * config/arc/arc.md (movqi_insn): Add stores to save constant long
13802 (movhi_insn): Update store instruction constraint which are saving
13803 6-bit short immediates.
13804 (movsi_insn): Consider also short scaled load operations.
13805 (zero_extendhisi2_i): Use Usd constraint instead of T.
13806 (extendhisi2_i): Add q constraint.
13807 (arc_clzsi2): Add type and length attributes.
13808 (arc_ctzsi2): Likewise.
13809 * config/arc/constraints.md (Usc): Update constraint, the
13810 assembler can parse two relocations for a single instruction.
13812 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13814 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
13815 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
13817 2017-08-31 Olivier Hainque <hainque@adacore.com>
13819 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
13820 match as powerpc-wrs-vxworks*.
13822 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
13824 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
13825 register constraint for by-element operand.
13826 (aarch64_mls_elt_merge<mode>): Likewise.
13828 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13830 * config/arc/arc.c (arc_can_follow_jump): Check for short
13833 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13835 * config.gcc: Use g.opt for arc.
13836 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
13837 functionality moved to ...
13838 (legitimate_scaled_address_p): New function, ...here.
13839 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
13840 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
13841 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
13843 (arc_override_options): Handle G option.
13844 (arc_output_pic_addr_const): Correct function definition.
13845 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
13846 (arc_decl_anon_ns_mem_p): Delete.
13847 (arc_in_small_data_p): Overhaul this function to take into
13848 consideration the value given via G option.
13849 (arc_rewrite_small_data_1): Renamed and corrected old
13850 arc_rewrite_small_data function.
13851 (arc_rewrite_small_data): New function.
13852 (small_data_pattern): Don't use pic_offset_table_rtx.
13853 * config/arc/arc.h (CC1_SPEC): Recognize G option.
13854 * config/arc/simdext.md (movmisalignv2hi): Use
13855 prepare_move_operands function.
13857 (movmisalign*): Likewise.
13858 * doc/invoke.texi (ARC options): Document -G option.
13860 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13862 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
13864 * config/arc/arc.c (arc_print_operand): Output scalled address for
13865 sdata whenever is possible.
13866 (arc_in_small_data_p): Allow sdata for 64bit datum when double
13867 load/stores are available.
13868 (compact_sda_memory_operand): Check for the alignment required by
13869 code density instructions.
13870 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
13872 * config/arc/constraints.md (Usd): Update constraint.
13873 (Us0): New constraint.
13874 (Usc): Update constraint.
13876 2017-08-31 Richard Biener <rguenther@suse.de>
13878 PR middle-end/82054
13879 * dwarf2out.c (dwarf2out_early_global_decl): Process each
13880 function only once.
13882 2017-08-31 Tamar Christina <tamar.christina@arm.com>
13884 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
13885 Resize type_signature.
13887 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13888 Alan Hayward <alan.hayward@arm.com>
13889 David Sherwood <david.sherwood@arm.com>
13891 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
13892 subregs whose inner modes can be stored in GPRs.
13893 (aarch64_classify_index): Likewise.
13895 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13896 Alan Hayward <alan.hayward@arm.com>
13897 David Sherwood <david.sherwood@arm.com>
13899 * config/aarch64/iterators.md (V_cmp_result): Rename to...
13900 (V_INT_EQUIV): ...this.
13901 (v_cmp_result): Rename to...
13902 (v_int_equiv): ...this.
13903 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
13904 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
13905 (copysign<mode>3): Likewise.
13906 (aarch64_simd_bsl<mode>_internal): Likewise.
13907 (aarch64_simd_bsl<mode>): Likewise.
13908 (vec_cmp<mode><mode>): Likewise.
13909 (vcond<mode><mode>): Likewise.
13910 (vcond<v_cmp_mixed><mode>): Likewise.
13911 (vcondu<mode><v_cmp_mixed>): Likewise.
13912 (aarch64_cm<optab><mode>): Likewise.
13913 (aarch64_cmtst<mode>): Likewise.
13914 (aarch64_fac<optab><mode>): Likewise.
13915 (vec_perm_const<mode>): Likewise.
13916 (vcond_mask_<mode><v_cmp_result>): Rename to...
13917 (vcond_mask_<mode><v_int_equiv>): ...this.
13918 (vec_cmp<mode><v_cmp_result>): Rename to...
13919 (vec_cmp<mode><v_int_equiv>): ...this.
13921 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13922 Alan Hayward <alan.hayward@arm.com>
13923 David Sherwood <david.sherwood@arm.com>
13925 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
13927 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
13928 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
13929 (UNSPEC_LD4_DREG): New unspecs.
13930 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
13931 (aarch64_ld2<mode>_dreg_be): Replace with...
13932 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
13934 (aarch64_ld3<mode>_dreg_le)
13935 (aarch64_ld3<mode>_dreg_be): Replace with...
13936 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
13938 (aarch64_ld4<mode>_dreg_le)
13939 (aarch64_ld4<mode>_dreg_be): Replace with...
13940 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
13943 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13945 PR tree-optimization/81987
13946 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
13947 insert an initializer in a location not dominated by the stride
13950 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13952 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
13953 on the entire header of the finally block in the fallthru case.
13955 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13957 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
13959 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
13961 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
13962 rs6000_emit_move_from_cr and call renamed function.
13963 (rs6000_emit_prologue): Call renamed functions.
13964 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
13965 movesi_from_cr, remove volatile CRs.
13967 2017-08-30 Jon Beniston <jon@beniston.com>
13968 Richard Biener <rguenther@suse.de>
13970 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
13971 of VECTOR_MODE_P check.
13972 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
13973 element vector types.
13975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13977 * df.h (df_read_modify_subreg_p): Remove in favor of...
13978 * rtl.h (read_modify_subreg_p): ...this new function. Take a
13979 const_rtx instead of an rtx.
13980 * cprop.c (local_cprop_find_used_regs): Update accordingly.
13981 * df-problems.c (df_word_lr_mark_ref): Likewise.
13982 * ira-lives.c (mark_pseudo_reg_live): Likewise.
13983 (mark_pseudo_reg_dead): Likewise.
13984 (mark_ref_dead): Likewise.
13985 * reginfo.c (init_subregs_of_mode): Likewise.
13986 * sched-deps.c (sched_analyze_1): Likewise.
13987 * df-scan.c (df_def_record_1): Likewise.
13988 (df_uses_record): Likewise.
13989 (df_read_modify_subreg_p): Remove in favor of...
13990 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
13991 const_rtx instead of an rtx.
13993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13994 Alan Hayward <alan.hayward@arm.com>
13995 David Sherwood <david.sherwood@arm.com>
13997 * rtl.h (partial_subreg_p): New function.
13998 * caller-save.c (save_call_clobbered_regs): Use it.
13999 * calls.c (expand_call): Likewise.
14000 * combine.c (combinable_i3pat): Likewise.
14001 (simplify_set): Likewise.
14002 (make_extraction): Likewise.
14003 (make_compound_operation_int): Likewise.
14004 (gen_lowpart_or_truncate): Likewise.
14005 (force_to_mode): Likewise.
14006 (make_field_assignment): Likewise.
14007 (reg_truncated_to_mode): Likewise.
14008 (record_truncated_value): Likewise.
14009 (move_deaths): Likewise.
14010 * cse.c (record_jump_cond): Likewise.
14011 (cse_insn): Likewise.
14012 * cselib.c (cselib_lookup_1): Likewise.
14013 * expmed.c (extract_bit_field_using_extv): Likewise.
14014 * function.c (assign_parm_setup_reg): Likewise.
14015 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14016 * ira-build.c (create_insn_allocnos): Likewise.
14017 * lra-coalesce.c (merge_pseudos): Likewise.
14018 * lra-constraints.c (match_reload): Likewise.
14019 (simplify_operand_subreg): Likewise.
14020 (curr_insn_transform): Likewise.
14021 * lra-lives.c (process_bb_lives): Likewise.
14022 * lra.c (new_insn_reg): Likewise.
14023 (lra_substitute_pseudo): Likewise.
14024 * regcprop.c (mode_change_ok): Likewise.
14025 (maybe_mode_change): Likewise.
14026 (copyprop_hardreg_forward_1): Likewise.
14027 * reload.c (push_reload): Likewise.
14028 (find_reloads): Likewise.
14029 (find_reloads_subreg_address): Likewise.
14030 * reload1.c (alter_reg): Likewise.
14031 (eliminate_regs_1): Likewise.
14032 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14034 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14036 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14037 back to if statements, including unpack.
14039 2017-08-30 Martin Liska <mliska@suse.cz>
14041 PR inline-asm/82001
14042 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14044 (func_checker::compare_asm_inputs_outputs): ... this function.
14045 (func_checker::compare_gimple_asm): Use the function to compare
14046 also ASM constrains.
14047 * ipa-icf-gimple.h: Rename the function.
14049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14050 Alan Hayward <alan.hayward@arm.com>
14051 David Sherwood <david.sherwood@arm.com>
14053 * coretypes.h (complex_mode): New type.
14054 * gdbhooks.py (build_pretty_printer): Handle it.
14055 * machmode.h (complex_mode): New class.
14056 (complex_mode::includes_p): New function.
14057 (is_complex_int_mode): Likewise.
14058 (is_complex_float_mode): Likewise.
14059 * genmodes.c (get_mode_class): Handle complex mode classes.
14060 * function.c (expand_function_end): Use is_complex_int_mode.
14062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14063 Alan Hayward <alan.hayward@arm.com>
14064 David Sherwood <david.sherwood@arm.com>
14066 * coretypes.h (scalar_mode_pod): New typedef.
14067 * gdbhooks.py (build_pretty_printer): Handle it.
14068 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14069 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14070 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14071 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14072 as_a <scalar_mode>.
14074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14075 Alan Hayward <alan.hayward@arm.com>
14076 David Sherwood <david.sherwood@arm.com>
14078 * machmode.h (mode_for_vector): Take a scalar_mode instead
14080 * stor-layout.c (mode_for_vector): Likewise.
14081 * explow.c (promote_mode): Use as_a <scalar_mode>.
14082 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14085 Alan Hayward <alan.hayward@arm.com>
14086 David Sherwood <david.sherwood@arm.com>
14088 * target.def (preferred_simd_mode): Take a scalar_mode
14089 instead of a machine_mode.
14090 * targhooks.h (default_preferred_simd_mode): Likewise.
14091 * targhooks.c (default_preferred_simd_mode): Likewise.
14092 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14093 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14094 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14095 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14096 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14097 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14098 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14099 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14101 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14102 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14103 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14104 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14105 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14106 * doc/tm.texi: Regenerate.
14107 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14111 Alan Hayward <alan.hayward@arm.com>
14112 David Sherwood <david.sherwood@arm.com>
14114 * target.def (scalar_mode_supported_p): Take a scalar_mode
14115 instead of a machine_mode.
14116 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14117 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14118 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14119 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14120 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14121 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14122 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14123 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14124 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14125 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14126 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14127 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14128 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14129 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14131 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14132 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14133 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14134 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14135 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14137 * doc/tm.texi: Regenerate.
14139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14140 Alan Hayward <alan.hayward@arm.com>
14141 David Sherwood <david.sherwood@arm.com>
14143 * coretypes.h (opt_scalar_mode): New typedef.
14144 * gdbhooks.py (build_pretty_printers): Handle it.
14145 * machmode.h (mode_iterator::get_2xwider): Add overload for
14147 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14149 * expr.c (convert_mode_scalar): Likewise.
14150 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14151 * optabs.c (expand_float): Likewise.
14152 (expand_fix): Likewise.
14153 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14156 Alan Hayward <alan.hayward@arm.com>
14157 David Sherwood <david.sherwood@arm.com>
14159 * optabs.c (expand_float): Explicitly check for scalars before
14160 using a branching expansion.
14161 (expand_fix): Likewise.
14163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14164 Alan Hayward <alan.hayward@arm.com>
14165 David Sherwood <david.sherwood@arm.com>
14167 * expr.c (convert_mode): Split scalar handling out into...
14168 (convert_mode_scalar): ...this new function. Treat the modes
14171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14172 Alan Hayward <alan.hayward@arm.com>
14173 David Sherwood <david.sherwood@arm.com>
14175 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14177 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14180 Alan Hayward <alan.hayward@arm.com>
14181 David Sherwood <david.sherwood@arm.com>
14183 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14184 rather than a machine_mode.
14185 (fixed_from_string): Likewise.
14186 (fixed_convert): Likewise.
14187 (fixed_convert_from_int): Likewise.
14188 (fixed_convert_from_real): Likewise.
14189 (real_convert_from_fixed): Likewise.
14190 * fixed-value.c (fixed_from_double_int): Likewise.
14191 (fixed_from_string): Likewise.
14192 (fixed_convert): Likewise.
14193 (fixed_convert_from_int): Likewise.
14194 (fixed_convert_from_real): Likewise.
14195 (real_convert_from_fixed): Likewise.
14196 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14199 Alan Hayward <alan.hayward@arm.com>
14200 David Sherwood <david.sherwood@arm.com>
14202 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14203 of separate mode class checks. Do not allow vector modes here.
14204 (immed_wide_int_const): Use as_a <scalar_mode>.
14205 * explow.c (trunc_int_for_mode): Likewise.
14206 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14207 (wi::shwi): Likewise.
14208 (wi::min_value): Likewise.
14209 (wi::max_value): Likewise.
14210 * dwarf2out.c (loc_descriptor): Likewise.
14211 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
14212 for CONST_WIDE_INT.
14214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14215 Alan Hayward <alan.hayward@arm.com>
14216 David Sherwood <david.sherwood@arm.com>
14218 * tree.h (SCALAR_TYPE_MODE): New macro.
14219 * expr.c (expand_expr_addr_expr_1): Use it.
14220 (expand_expr_real_2): Likewise.
14221 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
14222 (fold_convert_const_fixed_from_int): Likewise.
14223 (fold_convert_const_fixed_from_real): Likewise.
14224 (native_encode_fixed): Likewise
14225 (native_encode_complex): Likewise
14226 (native_encode_vector): Likewise.
14227 (native_interpret_fixed): Likewise.
14228 (native_interpret_real): Likewise.
14229 (native_interpret_complex): Likewise.
14230 (native_interpret_vector): Likewise.
14231 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
14232 (simd_clone_adjust_argument_types): Likewise.
14233 (simd_clone_init_simd_arrays): Likewise.
14234 (simd_clone_adjust): Likewise.
14235 * stor-layout.c (layout_type): Likewise.
14236 * tree.c (build_minus_one_cst): Likewise.
14237 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14238 * tree-inline.c (estimate_move_cost): Likewise.
14239 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
14240 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
14241 (vectorizable_reduction): Likewise.
14242 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
14243 (vect_recog_mixed_size_cond_pattern): Likewise.
14244 (check_bool_pattern): Likewise.
14245 (adjust_bool_pattern): Likewise.
14246 (search_type_for_mask_1): Likewise.
14247 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
14248 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14249 (vectorizable_load): Likewise.
14250 (vectorizable_store): Likewise.
14251 * ubsan.c (ubsan_encode_value): Likewise.
14252 * varasm.c (output_constant): Likewise.
14254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14255 Alan Hayward <alan.hayward@arm.com>
14256 David Sherwood <david.sherwood@arm.com>
14258 * coretypes.h (scalar_mode): New class.
14259 * machmode.h (scalar_mode): Likewise.
14260 (scalar_mode::includes_p): New function.
14261 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
14262 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
14263 * genmodes.c (get_mode_class): Handle remaining scalar modes.
14264 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
14265 * expmed.c (store_bit_field_1): Likewise.
14266 (extract_bit_field_1): Likewise.
14267 * expr.c (write_complex_part): Likewise.
14268 (read_complex_part): Likewise.
14269 (emit_move_complex_push): Likewise.
14270 (expand_expr_real_2): Likewise.
14271 * function.c (assign_parm_setup_reg): Likewise.
14272 (assign_parms_unsplit_complex): Likewise.
14273 * optabs.c (expand_binop): Likewise.
14274 * rtlanal.c (subreg_get_info): Likewise.
14275 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14276 * varasm.c (output_constant_pool_2): Likewise.
14278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14279 Alan Hayward <alan.hayward@arm.com>
14280 David Sherwood <david.sherwood@arm.com>
14282 * expmed.c (extract_high_half): Use scalar_int_mode and remove
14284 (expmed_mult_highpart_optab): Likewise.
14285 (expmed_mult_highpart): Likewise.
14287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14288 Alan Hayward <alan.hayward@arm.com>
14289 David Sherwood <david.sherwood@arm.com>
14291 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
14292 instead of a machine_mode.
14293 (builtin_memset_read_str): Likewise.
14294 * builtins.c (c_readstr): Likewise.
14295 (builtin_memcpy_read_str): Likewise.
14296 (builtin_strncpy_read_str): Likewise.
14297 (builtin_memset_read_str): Likewise.
14298 (builtin_memset_gen_str): Likewise.
14299 (expand_builtin_signbit): Use scalar_int_mode for local variables.
14300 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
14301 instead of a machine_mode.
14302 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
14304 (make_extraction): Likewise.
14305 (try_widen_shift_mode): Take and return scalar_int_modes instead
14307 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
14308 a scalar_int_mode instead of a machine_mode.
14309 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
14310 (avr_addr_space_pointer_mode): Likewise.
14311 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
14312 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
14313 (msp430_unwind_word_mode): Likewise.
14314 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
14315 (spu_addr_space_pointer_mode): Likewise.
14316 (spu_addr_space_address_mode): Likewise.
14317 (spu_libgcc_cmp_return_mode): Likewise.
14318 (spu_libgcc_shift_count_mode): Likewise.
14319 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
14320 (rl78_addr_space_pointer_mode): Likewise.
14321 (fl78_unwind_word_mode): Likewise.
14322 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
14324 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
14325 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
14326 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
14327 (mips_valid_pointer_mode): Likewise.
14328 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
14329 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
14330 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
14332 (ft32_addr_space_address_mode): Likewise.
14333 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
14334 scalar_int_mode instead of a machine_mode.
14335 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
14337 (m32c_addr_space_address_mode): Likewise.
14338 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
14339 (rs6000_eh_return_filter_mode): Likewise.
14340 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
14341 (rs6000_eh_return_filter_mode): Likewise.
14342 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
14343 (s390_libgcc_shift_count_mode): Likewise.
14344 (s390_unwind_word_mode): Likewise.
14345 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
14347 * target.def (mode_rep_extended): Likewise.
14348 (valid_pointer_mode): Likewise.
14349 (addr_space.valid_pointer_mode): Likewise.
14350 (eh_return_filter_mode): Return a scalar_int_mode rather than
14352 (libgcc_cmp_return_mode): Likewise.
14353 (libgcc_shift_count_mode): Likewise.
14354 (unwind_word_mode): Likewise.
14355 (addr_space.pointer_mode): Likewise.
14356 (addr_space.address_mode): Likewise.
14357 * doc/tm.texi: Regenerate.
14358 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
14360 (do_jump): Use scalar_int_mode for local variables.
14361 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
14362 rather than a machine_mode.
14363 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
14364 (scompare_loc_descriptor_wide): Likewise.
14365 (scompare_loc_descriptor_narrow): Likewise.
14366 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
14368 * except.c (sjlj_emit_dispatch_table): Likewise.
14369 (expand_builtin_eh_copy_values): Likewise.
14370 * explow.c (convert_memory_address_addr_space_1): Likewise.
14371 Take a scalar_int_mode rather than a machine_mode.
14372 (convert_memory_address_addr_space): Take a scalar_int_mode rather
14373 than a machine_mode.
14374 (memory_address_addr_space): Use scalar_int_mode for local variables.
14375 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
14376 rather than a machine_mode.
14377 * expmed.c (mask_rtx): Likewise.
14378 (init_expmed_one_conv): Likewise.
14379 (expand_mult_highpart_adjust): Likewise.
14380 (extract_high_half): Likewise.
14381 (expmed_mult_highpart_optab): Likewise.
14382 (expmed_mult_highpart): Likewise.
14383 (expand_smod_pow2): Likewise.
14384 (expand_sdiv_pow2): Likewise.
14385 (emit_store_flag_int): Likewise.
14386 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
14388 (extract_low_bits): Likewise.
14389 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
14391 * expr.c (pieces_addr::adjust): Likewise.
14392 (can_store_by_pieces): Likewise.
14393 (store_by_pieces): Likewise.
14394 (clear_by_pieces_1): Likewise.
14395 (expand_expr_addr_expr_1): Likewise.
14396 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
14397 (expand_expr_real_1): Likewise.
14398 (try_casesi): Likewise.
14399 * final.c (shorten_branches): Likewise.
14400 * fold-const.c (fold_convert_const_int_from_fixed): Change the
14401 type of "mode" to machine_mode.
14402 * internal-fn.c (expand_arith_overflow_result_store): Take a
14403 scalar_int_mode rather than a machine_mode.
14404 (expand_mul_overflow): Use scalar_int_mode for local variables.
14405 * loop-doloop.c (doloop_modify): Likewise.
14406 (doloop_optimize): Likewise.
14407 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
14408 than a machine_mode.
14409 (expand_doubleword_shift_condmove): Likewise.
14410 (expand_doubleword_shift): Likewise.
14411 (expand_doubleword_clz): Likewise.
14412 (expand_doubleword_popcount): Likewise.
14413 (expand_doubleword_parity): Likewise.
14414 (expand_absneg_bit): Use scalar_int_mode for local variables.
14415 (prepare_float_lib_cmp): Likewise.
14416 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
14417 rather than a machine_mode.
14418 (convert_memory_address_addr_space): Likewise.
14419 (get_mode_bounds): Likewise.
14420 (get_address_mode): Return a scalar_int_mode rather than a
14422 * rtlanal.c (get_address_mode): Likewise.
14423 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
14424 than a machine_mode.
14425 * targhooks.c (default_mode_rep_extended): Likewise.
14426 (default_valid_pointer_mode): Likewise.
14427 (default_addr_space_valid_pointer_mode): Likewise.
14428 (default_eh_return_filter_mode): Return a scalar_int_mode rather
14429 than a machine_mode.
14430 (default_libgcc_cmp_return_mode): Likewise.
14431 (default_libgcc_shift_count_mode): Likewise.
14432 (default_unwind_word_mode): Likewise.
14433 (default_addr_space_pointer_mode): Likewise.
14434 (default_addr_space_address_mode): Likewise.
14435 * targhooks.h (default_eh_return_filter_mode): Likewise.
14436 (default_libgcc_cmp_return_mode): Likewise.
14437 (default_libgcc_shift_count_mode): Likewise.
14438 (default_unwind_word_mode): Likewise.
14439 (default_addr_space_pointer_mode): Likewise.
14440 (default_addr_space_address_mode): Likewise.
14441 (default_mode_rep_extended): Take a scalar_int_mode rather than
14443 (default_valid_pointer_mode): Likewise.
14444 (default_addr_space_valid_pointer_mode): Likewise.
14445 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
14447 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
14448 rather than a machine_mode.
14449 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
14450 for local variables.
14451 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14452 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
14453 than a machine_mode.
14455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14456 Alan Hayward <alan.hayward@arm.com>
14457 David Sherwood <david.sherwood@arm.com>
14459 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
14460 the mode argument to scalar_int_mode.
14461 (do_jump_by_parts_zero_rtx): Likewise.
14462 (do_jump_by_parts_equality_rtx): Likewise.
14463 (do_jump_by_parts_greater): Take a mode argument.
14464 (do_jump_by_parts_equality): Likewise.
14465 (do_jump_1): Update calls accordingly.
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 * is-a.h (safe_dyn_cast): New function.
14472 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
14473 (jump_table_for_label): Likewise.
14474 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
14475 instead of an rtx_insn *.
14476 (shorten_branches): Use dyn_cast instead of LABEL_P and
14477 JUMP_TABLE_DATA_P. Use jump_table_for_label and
14478 rtx_jump_table_data::get_data_mode.
14479 (final_scan_insn): Likewise.
14481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14482 Alan Hayward <alan.hayward@arm.com>
14483 David Sherwood <david.sherwood@arm.com>
14485 * combine.c (try_combine): Use is_a <scalar_int_mode> when
14486 trying to combine a full-register integer set with a subreg
14489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14490 Alan Hayward <alan.hayward@arm.com>
14491 David Sherwood <david.sherwood@arm.com>
14493 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
14494 that is always either address_mode or pointer_mode.
14496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14497 Alan Hayward <alan.hayward@arm.com>
14498 David Sherwood <david.sherwood@arm.com>
14500 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
14501 when the two are known to be equal.
14503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14504 Alan Hayward <alan.hayward@arm.com>
14505 David Sherwood <david.sherwood@arm.com>
14507 * simplify-rtx.c (simplify_const_unary_operation): Use
14508 is_a <scalar_int_mode> instead of checking for a nonzero
14509 precision. Forcibly convert op_mode to a scalar_int_mode
14510 in that case. More clearly differentiate the operand and
14511 result modes and use the former when deciding what the value
14512 of a count-bits operation should be. Use is_int_mode instead
14513 of checking for a MODE_INT. Remove redundant check for whether
14514 this mode has a zero precision.
14516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14517 Alan Hayward <alan.hayward@arm.com>
14518 David Sherwood <david.sherwood@arm.com>
14520 * optabs.c (widen_leading): Change the type of the mode argument
14521 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
14522 (widen_bswap): Likewise.
14523 (expand_parity): Likewise.
14524 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
14525 (expand_ffs): Likewise.
14526 (epand_unop): Check for scalar integer modes before calling the
14529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14530 Alan Hayward <alan.hayward@arm.com>
14531 David Sherwood <david.sherwood@arm.com>
14533 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
14535 (expand_expr_real_1): Update call accordingly.
14537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14538 Alan Hayward <alan.hayward@arm.com>
14539 David Sherwood <david.sherwood@arm.com>
14541 * expmed.c (store_bit_field_using_insv): Add op0_mode and
14542 value_mode arguments. Use scalar_int_mode internally.
14543 (store_bit_field_1): Rename the new integer mode from imode
14544 to op0_mode and use it instead of GET_MODE (op0). Update calls
14545 to store_split_bit_field, store_bit_field_using_insv and
14546 store_fixed_bit_field.
14547 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
14548 Use scalar_int_mode internally. Use a bit count rather than a mode
14549 when calculating the largest bit size for get_best_mode.
14550 Update calls to store_split_bit_field and store_fixed_bit_field_1.
14551 (store_fixed_bit_field_1): Add mode and value_mode arguments.
14552 Remove assertion that OP0 has a scalar integer mode.
14553 (store_split_bit_field): Add op0_mode and value_mode arguments.
14554 Update calls to extract_fixed_bit_field.
14555 (extract_bit_field_using_extv): Add an op0_mode argument.
14556 Use scalar_int_mode internally.
14557 (extract_bit_field_1): Rename the new integer mode from imode to
14558 op0_mode and use it instead of GET_MODE (op0). Update calls to
14559 extract_split_bit_field, extract_bit_field_using_extv and
14560 extract_fixed_bit_field.
14561 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
14562 to extract_split_bit_field and extract_fixed_bit_field_1.
14563 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
14564 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
14565 on the target mode.
14566 (extract_split_bit_field): Add an op0_mode argument. Update call
14567 to extract_fixed_bit_field.
14569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14570 Alan Hayward <alan.hayward@arm.com>
14571 David Sherwood <david.sherwood@arm.com>
14573 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
14574 * explow.c (hard_function_value): Likewise.
14575 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
14576 convert_to_mode call outside the loop.
14577 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
14578 for the mode iterator. Require the mode specified by max_pieces
14580 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
14582 (copy_blkmode_to_reg): Likewise.
14583 (set_storage_via_setmem): Likewise.
14584 * optabs.c (prepare_cmp_insn): Likewise.
14585 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14586 * stor-layout.c (finish_bitfield_representative): Likewise.
14588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14589 Alan Hayward <alan.hayward@arm.com>
14590 David Sherwood <david.sherwood@arm.com>
14592 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
14593 * expr.c (convert_move): Use them.
14594 (convert_modes): Likewise.
14595 (store_expr_with_bounds): Likewise.
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 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
14602 parameter for the mode of "x". Remove the "known_x", "known_mode"
14603 and "known_ret" arguments. Change the type of the mode argument
14604 to scalar_int_mode.
14605 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
14606 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
14607 (reg_num_sign_bit_copies_for_combine): Likewise.
14608 * rtlanal.c (nonzero_bits1): Likewise.
14609 (num_sign_bit_copies1): Likewise.
14610 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
14611 (reg_num_sign_bit_copies_general): Likewise.
14612 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
14613 (reg_nonzero_bits_general): Likewise.
14615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14616 Alan Hayward <alan.hayward@arm.com>
14617 David Sherwood <david.sherwood@arm.com>
14619 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
14620 than in subroutines. Return 1 for non-integer modes.
14621 (cached_num_sign_bit_copies): Change the type of the mode parameter
14622 to scalar_int_mode.
14623 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
14624 classes. Handle CONST_INT_P first and then check whether X also
14625 has a scalar integer mode. Check the same thing for inner registers
14626 of a SUBREG and for values that are being extended or truncated.
14628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14629 Alan Hayward <alan.hayward@arm.com>
14630 David Sherwood <david.sherwood@arm.com>
14632 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
14633 in subroutines. Return the mode mask for non-integer modes.
14634 (cached_nonzero_bits): Change the type of the mode parameter
14635 to scalar_int_mode.
14636 (nonzero_bits1): Likewise. Remove early exit for other mode
14637 classes. Handle CONST_INT_P first and then check whether X
14638 also has a scalar integer mode.
14640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14641 Alan Hayward <alan.hayward@arm.com>
14642 David Sherwood <david.sherwood@arm.com>
14644 * expr.c (widest_int_mode_for_size): Make the comment match the code.
14645 Return a scalar_int_mode and assert that the size is greater than
14647 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
14648 (op_by_pieces_d::op_by_pieces_d): Likewise.
14649 (op_by_pieces_d::run): Likewise.
14650 (can_store_by_pieces): Likewise.
14652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14653 Alan Hayward <alan.hayward@arm.com>
14654 David Sherwood <david.sherwood@arm.com>
14656 * combine.c (extract_left_shift): Add a mode argument and update
14658 (make_compound_operation_int): Change the type of the mode parameter
14659 to scalar_int_mode and update the call to extract_left_shift.
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 * combine.c (simplify_and_const_int): Change the type of the mode
14666 parameter to scalar_int_mode.
14667 (simplify_and_const_int_1): Likewise. Update recursive call.
14669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14670 Alan Hayward <alan.hayward@arm.com>
14671 David Sherwood <david.sherwood@arm.com>
14673 * combine.c (simplify_compare_const): Check that the mode is a
14674 scalar_int_mode (rather than VOIDmode) before testing its
14676 (simplify_comparison): Move COMPARISON_P handling out of the
14677 loop and restrict the latter part of the loop to scalar_int_modes.
14678 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
14679 and when considering SUBREG_REGs. Use is_int_mode instead of
14680 checking GET_MODE_CLASS against MODE_INT.
14682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14683 Alan Hayward <alan.hayward@arm.com>
14684 David Sherwood <david.sherwood@arm.com>
14686 * combine.c (try_widen_shift_mode): Move check for equal modes to...
14687 (simplify_shift_const_1): ...here. Use scalar_int_mode for
14688 shift_unit_mode and for modes involved in scalar shifts.
14690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14691 Alan Hayward <alan.hayward@arm.com>
14692 David Sherwood <david.sherwood@arm.com>
14694 * combine.c (force_int_to_mode): New function, split out from...
14695 (force_to_mode): ...here. Keep xmode up-to-date and use it
14696 instead of GET_MODE (x).
14698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14699 Alan Hayward <alan.hayward@arm.com>
14700 David Sherwood <david.sherwood@arm.com>
14702 * optabs-query.h (extraction_insn::struct_mode): Change type to
14703 opt_scalar_int_mode and update comment.
14704 (extraction_insn::field_mode): Change type to scalar_int_mode.
14705 (extraction_insn::pos_mode): Likewise.
14706 * combine.c (make_extraction): Update accordingly.
14707 * optabs-query.c (get_traditional_extraction_insn): Likewise.
14708 (get_optab_extraction_insn): Likewise.
14709 * recog.c (simplify_while_replacing): Likewise.
14710 * expmed.c (narrow_bit_field_mem): Change the type of the mode
14711 parameter to opt_scalar_int_mode.
14713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14714 Alan Hayward <alan.hayward@arm.com>
14715 David Sherwood <david.sherwood@arm.com>
14717 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
14718 to a scalar_int_mode instead of a machine_mode.
14719 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
14720 (get_best_mode): Return a boolean and use a pointer argument to store
14721 the selected mode. Replace the limit mode parameter with a bit limit.
14722 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
14723 for the values returned by bit_field_mode_iterator::next_mode.
14724 (store_bit_field): Update call to get_best_mode.
14725 (store_fixed_bit_field): Likewise.
14726 (extract_fixed_bit_field): Likewise.
14727 * expr.c (optimize_bitfield_assignment_op): Likewise.
14728 * fold-const.c (optimize_bit_field_compare): Likewise.
14729 (fold_truth_andor_1): Likewise.
14730 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
14731 Update for new type of m_mode.
14732 (get_best_mode): As above.
14734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14735 Alan Hayward <alan.hayward@arm.com>
14736 David Sherwood <david.sherwood@arm.com>
14738 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
14739 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
14740 (store_bit_field): Check is_a <scalar_int_mode> before calling
14741 strict_volatile_bitfield_p.
14742 (extract_bit_field): Likewise.
14744 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14745 Alan Hayward <alan.hayward@arm.com>
14746 David Sherwood <david.sherwood@arm.com>
14748 * target.def (cstore_mode): Return a scalar_int_mode.
14749 * doc/tm.texi: Regenerate.
14750 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
14751 * targhooks.h (default_cstore_mode): Likewise.
14752 * targhooks.c (default_cstore_mode): Likewise, using a forced
14754 * expmed.c (emit_cstore): Expect the target of the cstore to be
14757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14758 Alan Hayward <alan.hayward@arm.com>
14759 David Sherwood <david.sherwood@arm.com>
14761 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
14763 (niter_desc): Likewise mode.
14764 (iv_analyze): Add a mode parameter.
14766 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
14767 and change its type to scalar_int_mode.
14768 * loop-iv.c: Update commentary at head of file.
14769 (iv_constant): Pass the mode paraeter before the rtx it describes
14770 and change its type to scalar_int_mode. Remove VOIDmode handling.
14771 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
14772 (iv_extend): Likewise.
14773 (shorten_into_mode): Likewise.
14774 (iv_add): Use scalar_int_mode.
14775 (iv_mult): Likewise.
14776 (iv_shift): Likewise.
14777 (canonicalize_iv_subregs): Likewise.
14778 (get_biv_step_1): Pass the outer_mode parameter before the rtx
14779 it describes and change its mode to scalar_int_mode. Also change
14780 the type of the returned inner_mode to scalar_int_mode.
14781 (get_biv_step): Likewise, turning outer_mode from a pointer
14782 into a direct parameter. Update call to get_biv_step_1.
14783 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
14784 iv_constant and get_biv_step.
14785 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
14786 and change its type to scalar_int_mode. Don't initialise iv->mode
14787 to VOIDmode and remove later checks for its still being VOIDmode.
14788 Update calls to iv_analyze_op and iv_analyze_expr. Check
14789 is_a <scalar_int_mode> when changing the mode under consideration.
14790 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
14791 Update call to iv_analyze_expr.
14792 (iv_analyze_op): Add a mode parameter. Reject subregs whose
14793 inner register is not also a scalar_int_mode. Update call to
14795 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
14796 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
14797 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
14798 separate mode class checks. Update calls to iv_analyze. Remove
14799 fix-up of VOIDmodes after iv_analyze_biv.
14800 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
14801 don't have a scalar_int_mode. Update call to biv_p.
14803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14804 Alan Hayward <alan.hayward@arm.com>
14805 David Sherwood <david.sherwood@arm.com>
14807 * cfgexpand.c (convert_debug_memory_address): Use
14808 as_a <scalar_int_mode>.
14809 * combine.c (expand_compound_operation): Likewise.
14810 (make_extraction): Likewise.
14811 (change_zero_ext): Likewise.
14812 (simplify_comparison): Likewise.
14813 * cse.c (cse_insn): Likewise.
14814 * dwarf2out.c (minmax_loc_descriptor): Likewise.
14815 (mem_loc_descriptor): Likewise.
14816 (loc_descriptor): Likewise.
14817 * expmed.c (init_expmed_one_mode): Likewise.
14818 (synth_mult): Likewise.
14819 (emit_store_flag_1): Likewise.
14820 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
14821 of a comparison with size.
14822 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
14823 (reduce_to_bit_field_precision): Likewise.
14824 * function.c (expand_function_end): Likewise.
14825 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
14826 * loop-doloop.c (doloop_modify): Likewise.
14827 * optabs.c (expand_binop): Likewise.
14828 (expand_unop): Likewise.
14829 (expand_copysign_absneg): Likewise.
14830 (prepare_cmp_insn): Likewise.
14831 (maybe_legitimize_operand): Likewise.
14832 * recog.c (const_scalar_int_operand): Likewise.
14833 * rtlanal.c (get_address_mode): Likewise.
14834 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14835 (simplify_cond_clz_ctz): Likewise.
14836 * tree-nested.c (get_nl_goto_field): Likewise.
14837 * tree.c (build_vector_type_for_mode): Likewise.
14838 * var-tracking.c (use_narrower_mode): Likewise.
14840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14841 Alan Hayward <alan.hayward@arm.com>
14842 David Sherwood <david.sherwood@arm.com>
14844 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
14845 * builtins.c (expand_builtin_signbit): Use it.
14846 * cfgexpand.c (expand_debug_expr): Likewise.
14847 * dojump.c (do_jump): Likewise.
14848 (do_compare_and_jump): Likewise.
14849 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
14850 * expmed.c (make_tree): Likewise.
14851 * expr.c (expand_expr_real_2): Likewise.
14852 (expand_expr_real_1): Likewise.
14853 (try_casesi): Likewise.
14854 * fold-const-call.c (fold_const_call_ss): Likewise.
14855 * fold-const.c (unextend): Likewise.
14856 (extract_muldiv_1): Likewise.
14857 (fold_single_bit_test): Likewise.
14858 (native_encode_int): Likewise.
14859 (native_encode_string): Likewise.
14860 (native_interpret_int): Likewise.
14861 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
14862 * internal-fn.c (expand_addsub_overflow): Likewise.
14863 (expand_neg_overflow): Likewise.
14864 (expand_mul_overflow): Likewise.
14865 (expand_arith_overflow): Likewise.
14866 * match.pd: Likewise.
14867 * stor-layout.c (layout_type): Likewise.
14868 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14869 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
14870 * tree-ssanames.c (get_range_info): Likewise.
14871 * tree-switch-conversion.c (array_value_type) Likewise.
14872 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
14873 (vect_recog_divmod_pattern): Likewise.
14874 (vect_recog_mixed_size_cond_pattern): Likewise.
14875 * tree-vrp.c (extract_range_basic): Likewise.
14876 (simplify_float_conversion_using_ranges): Likewise.
14877 * tree.c (int_fits_type_p): Likewise.
14878 * ubsan.c (instrument_bool_enum_load): Likewise.
14879 * varasm.c (mergeable_string_section): Likewise.
14880 (narrowing_initializer_constant_valid_p): Likewise.
14881 (output_constant): Likewise.
14883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14884 Alan Hayward <alan.hayward@arm.com>
14885 David Sherwood <david.sherwood@arm.com>
14887 * machmode.h (NARROWEST_INT_MODE): New macro.
14888 * expr.c (alignment_for_piecewise_move): Use it instead of
14889 GET_CLASS_NARROWEST_MODE (MODE_INT).
14890 (push_block): Likewise.
14891 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
14893 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14896 Alan Hayward <alan.hayward@arm.com>
14897 David Sherwood <david.sherwood@arm.com>
14899 * postreload.c (move2add_valid_value_p): Change the type of the
14900 mode parameter to scalar_int_mode.
14901 (move2add_use_add2_insn): Add a mode parameter and use it instead
14903 (move2add_use_add3_insn): Likewise.
14904 (reload_cse_move2add): Update accordingly.
14906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14907 Alan Hayward <alan.hayward@arm.com>
14908 David Sherwood <david.sherwood@arm.com>
14910 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
14912 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
14913 * optabs.c (expand_unop): Likewise.
14915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14916 Alan Hayward <alan.hayward@arm.com>
14917 David Sherwood <david.sherwood@arm.com>
14919 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
14920 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
14921 (popcount_loc_descriptor): Likewise.
14922 (bswap_loc_descriptor): Likewise.
14923 (rotate_loc_descriptor): Likewise.
14924 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
14925 calling the functions above.
14927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14928 Alan Hayward <alan.hayward@arm.com>
14929 David Sherwood <david.sherwood@arm.com>
14931 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
14933 (try_combine): Likewise.
14934 (simplify_if_then_else): Likewise.
14935 * cse.c (cse_insn): Likewise.
14936 * dwarf2out.c (mem_loc_descriptor): Likewise.
14937 * emit-rtl.c (gen_lowpart_common): Likewise.
14938 * simplify-rtx.c (simplify_truncation): Likewise.
14939 (simplify_binary_operation_1): Likewise.
14940 (simplify_const_relational_operation): Likewise.
14941 (simplify_ternary_operation): Likewise.
14942 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
14944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14945 Alan Hayward <alan.hayward@arm.com>
14946 David Sherwood <david.sherwood@arm.com>
14948 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
14949 * reload.c (push_reload): Likewise.
14950 (find_reloads): Likewise.
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 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
14957 (make_compound_operation_int): Likewise.
14958 (change_zero_ext): Likewise.
14959 * expr.c (convert_move): Likewise.
14960 (convert_modes): Likewise.
14961 * fwprop.c (forward_propagate_subreg): Likewise.
14962 * loop-iv.c (get_biv_step_1): Likewise.
14963 * optabs.c (widen_operand): Likewise.
14964 * postreload.c (move2add_valid_value_p): Likewise.
14965 * recog.c (simplify_while_replacing): Likewise.
14966 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14967 (simplify_binary_operation_1): Likewise. Remove redundant
14968 mode equality check.
14970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14971 Alan Hayward <alan.hayward@arm.com>
14972 David Sherwood <david.sherwood@arm.com>
14974 * combine.c (combine_simplify_rtx): Add checks for
14975 is_a <scalar_int_mode>.
14976 (simplify_if_then_else): Likewise.
14977 (make_field_assignment): Likewise.
14978 (simplify_comparison): Likewise.
14979 * ifcvt.c (noce_try_bitop): Likewise.
14980 * loop-invariant.c (canonicalize_address_mult): Likewise.
14981 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14984 Alan Hayward <alan.hayward@arm.com>
14985 David Sherwood <david.sherwood@arm.com>
14987 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
14988 is_a <scalar_int_mode> instead of != BLKmode.
14990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14991 Alan Hayward <alan.hayward@arm.com>
14992 David Sherwood <david.sherwood@arm.com>
14994 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
14995 instead of != VOIDmode.
14996 * combine.c (if_then_else_cond): Likewise.
14997 (change_zero_ext): Likewise.
14998 * dwarf2out.c (mem_loc_descriptor): Likewise.
14999 (loc_descriptor): Likewise.
15000 * rtlanal.c (canonicalize_condition): Likewise.
15001 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15004 Alan Hayward <alan.hayward@arm.com>
15005 David Sherwood <david.sherwood@arm.com>
15007 * simplify-rtx.c (simplify_binary_operation_1): Use
15008 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15011 Alan Hayward <alan.hayward@arm.com>
15012 David Sherwood <david.sherwood@arm.com>
15014 * wide-int.h (int_traits<unsigned char>) New class.
15015 (int_traits<unsigned short>) Likewise.
15016 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15017 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15019 * combine.c (set_nonzero_bits_and_sign_copies): Use
15020 is_a <scalar_int_mode>.
15021 (find_split_point): Likewise.
15022 (combine_simplify_rtx): Likewise.
15023 (simplify_logical): Likewise.
15024 (expand_compound_operation): Likewise.
15025 (expand_field_assignment): Likewise.
15026 (make_compound_operation): Likewise.
15027 (extended_count): Likewise.
15028 (change_zero_ext): Likewise.
15029 (simplify_comparison): Likewise.
15030 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15031 (ucompare_loc_descriptor): Likewise.
15032 (minmax_loc_descriptor): Likewise.
15033 (mem_loc_descriptor): Likewise.
15034 (loc_descriptor): Likewise.
15035 * expmed.c (init_expmed_one_mode): Likewise.
15036 * lra-constraints.c (lra_constraint_offset): Likewise.
15037 * optabs.c (prepare_libcall_arg): Likewise.
15038 * postreload.c (move2add_note_store): Likewise.
15039 * reload.c (operands_match_p): Likewise.
15040 * rtl.h (load_extend_op): Likewise.
15041 * rtlhooks.c (gen_lowpart_general): Likewise.
15042 * simplify-rtx.c (simplify_truncation): Likewise.
15043 (simplify_unary_operation_1): Likewise.
15044 (simplify_binary_operation_1): Likewise.
15045 (simplify_const_binary_operation): Likewise.
15046 (simplify_const_relational_operation): Likewise.
15047 (simplify_subreg): Likewise.
15048 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15049 * var-tracking.c (adjust_mems): Likewise.
15050 (prepare_call_arguments): Likewise.
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 * machmode.h (is_int_mode): New fuction.
15057 * combine.c (find_split_point): Use it.
15058 (combine_simplify_rtx): Likewise.
15059 (simplify_if_then_else): Likewise.
15060 (simplify_set): Likewise.
15061 (simplify_shift_const_1): Likewise.
15062 (simplify_comparison): Likewise.
15063 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15064 * cse.c (notreg_cost): Likewise.
15065 (cse_insn): Likewise.
15066 * cselib.c (cselib_lookup_1): Likewise.
15067 * dojump.c (do_jump_1): Likewise.
15068 (do_compare_rtx_and_jump): Likewise.
15069 * dse.c (get_call_args): Likewise.
15070 * dwarf2out.c (rtl_for_decl_init): Likewise.
15071 (native_encode_initializer): Likewise.
15072 * expmed.c (emit_store_flag_1): Likewise.
15073 (emit_store_flag): Likewise.
15074 * expr.c (convert_modes): Likewise.
15075 (store_field): Likewise.
15076 (expand_expr_real_1): Likewise.
15077 * fold-const.c (fold_read_from_constant_string): Likewise.
15078 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15079 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15080 * optabs.c (expand_binop): Likewise.
15081 (expand_unop): Likewise.
15082 (expand_abs_nojump): Likewise.
15083 (expand_one_cmpl_abs_nojump): Likewise.
15084 * simplify-rtx.c (mode_signbit_p): Likewise.
15085 (val_signbit_p): Likewise.
15086 (val_signbit_known_set_p): Likewise.
15087 (val_signbit_known_clear_p): Likewise.
15088 (simplify_relational_operation_1): Likewise.
15089 * tree.c (vector_type_mode): Likewise.
15091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15092 Alan Hayward <alan.hayward@arm.com>
15093 David Sherwood <david.sherwood@arm.com>
15095 * machmode.h (smallest_mode_for_size): Fix formatting.
15096 (smallest_int_mode_for_size): New function.
15097 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15098 instead of smallest_mode_for_size.
15099 * combine.c (make_extraction): Likewise.
15100 * config/arc/arc.c (arc_expand_movmem): Likewise.
15101 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15102 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15103 * config/s390/s390.c (s390_expand_insv): Likewise.
15104 * config/sparc/sparc.c (assign_int_registers): Likewise.
15105 * config/spu/spu.c (spu_function_value): Likewise.
15106 (spu_function_arg): Likewise.
15107 * coverage.c (get_gcov_type): Likewise.
15108 (get_gcov_unsigned_t): Likewise.
15109 * dse.c (find_shift_sequence): Likewise.
15110 * expmed.c (store_bit_field_1): Likewise.
15111 * expr.c (convert_move): Likewise.
15112 (store_field): Likewise.
15113 * internal-fn.c (expand_arith_overflow): Likewise.
15114 * optabs-query.c (get_best_extraction_insn): Likewise.
15115 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15116 * stor-layout.c (layout_type): Likewise.
15117 (initialize_sizetypes): Likewise.
15118 * targhooks.c (default_get_mask_mode): Likewise.
15119 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15122 Alan Hayward <alan.hayward@arm.com>
15123 David Sherwood <david.sherwood@arm.com>
15125 * machmode.h (opt_mode::else_blk): New function.
15126 (int_mode_for_mode): Declare.
15127 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15128 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15130 * cfgexpand.c (expand_debug_expr): Likewise.
15131 * combine.c (gen_lowpart_or_truncate): Likewise.
15132 (gen_lowpart_for_combine): Likewise.
15133 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15134 * config/avr/avr.c (avr_to_int_mode): Likewise.
15135 (avr_out_plus_1): Likewise.
15136 (avr_out_plus): Likewise.
15137 (avr_out_round): Likewise.
15138 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15139 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15140 (s390_expand_vcond): Likewise.
15141 * config/spu/spu.c (spu_split_immediate): Likewise.
15142 (spu_expand_mov): Likewise.
15143 * dse.c (get_stored_val): Likewise.
15144 * expmed.c (store_bit_field_1): Likewise.
15145 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15147 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15148 (extract_low_bits): Likewise.
15149 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15150 handling rather than repeating the check.
15151 (emit_group_store): Likewise.
15152 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15153 * optabs.c (expand_absneg_bit): Likewise.
15154 (expand_copysign_absneg): Likewise.
15155 (expand_copysign_bit): Likewise.
15156 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15157 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15158 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15159 * var-tracking.c (prepare_call_arguments): Likewise.
15160 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15161 int_mode_for_mode instead of mode_for_size.
15162 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15165 Alan Hayward <alan.hayward@arm.com>
15166 David Sherwood <david.sherwood@arm.com>
15168 * machmode.h (int_mode_for_size): New function.
15169 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15170 instead of mode_for_size.
15171 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15173 * combine.c (expand_field_assignment): Use int_mode_for_size
15174 instead of mode_for_size.
15175 (make_extraction): Likewise.
15176 (simplify_shift_const_1): Likewise.
15177 (simplify_comparison): Likewise.
15178 * dojump.c (do_jump): Likewise.
15179 * dwarf2out.c (mem_loc_descriptor): Likewise.
15180 * emit-rtl.c (init_derived_machine_modes): Likewise.
15181 * expmed.c (flip_storage_order): Likewise.
15182 (convert_extracted_bit_field): Likewise.
15183 * expr.c (copy_blkmode_from_reg): Likewise.
15184 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15185 * internal-fn.c (expand_mul_overflow): Likewise.
15186 * lower-subreg.c (simple_move): Likewise.
15187 * optabs-libfuncs.c (init_optabs): Likewise.
15188 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15189 * tree.c (vector_type_mode): Likewise.
15190 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15191 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15192 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15193 * tree-vect-stmts.c (vectorizable_load): Likewise.
15194 (vectorizable_store): Likewise.
15196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15197 Alan Hayward <alan.hayward@arm.com>
15198 David Sherwood <david.sherwood@arm.com>
15200 * coretypes.h (pod_mode): New type.
15201 (scalar_int_mode_pod): New typedef.
15202 * machmode.h (pod_mode): New class.
15203 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15204 * genmodes.c (emit_mode_int_n): Update accordingly.
15205 * lower-subreg.h (target_lower_subreg): Change type to
15206 scalar_int_mode_pod.
15207 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15208 scalar_int_mode_pod.
15210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15211 Alan Hayward <alan.hayward@arm.com>
15212 David Sherwood <david.sherwood@arm.com>
15214 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
15215 machine_mode to scalar_int_mode.
15216 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
15217 (rs6000_option_override_internal): Remove cast to int.
15218 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
15219 machine_mode to scalar_int_mode.
15220 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
15221 (rs6000_option_override_internal): Remove cast to int.
15222 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
15223 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
15225 * config/s390/s390.c (s390_expand_builtin): Likewise.
15226 * coretypes.h (scalar_int_mode): New type.
15227 (opt_scalar_int_mode): New typedef.
15228 * machmode.h (scalar_int_mode): New class.
15229 (scalar_int_mode::includes_p): New function.
15230 (byte_mode): Change type to scalar_int_mode.
15231 (word_mode): Likewise.
15232 (ptr_mode): Likewise.
15233 * emit-rtl.c (byte_mode): Likewise.
15234 (word_mode): Likewise.
15235 (ptr_mode): Likewise.
15236 (init_derived_machine_modes): Update accordingly.
15237 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
15238 and MODE_PARTIAL_INT.
15239 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
15240 opt_scalar_int_mode.
15242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15243 Alan Hayward <alan.hayward@arm.com>
15244 David Sherwood <david.sherwood@arm.com>
15246 * target.def (libgcc_floating_mode_supported_p): Take a
15248 * doc/tm.texi: Regenerate.
15249 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
15251 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15252 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
15255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15256 Alan Hayward <alan.hayward@arm.com>
15257 David Sherwood <david.sherwood@arm.com>
15259 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
15260 * doc/tm.texi: Regenerate.
15261 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
15262 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
15263 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
15264 * targhooks.h (default_floatn_mode): Likewise.
15265 * targhooks.c (default_floatn_mode): Likewise.
15266 * tree.c (build_common_tree_nodes): Update accordingly.
15268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15269 Alan Hayward <alan.hayward@arm.com>
15270 David Sherwood <david.sherwood@arm.com>
15272 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
15273 (mode_iterator::iterate_p): Likewise.
15274 (mode_iterator::get_wider): Likewise.
15275 * expr.c (init_expr_target): Use opt_scalar_float_mode.
15277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15278 Alan Hayward <alan.hayward@arm.com>
15279 David Sherwood <david.sherwood@arm.com>
15281 * coretypes.h (opt_scalar_float_mode): New typedef.
15282 * machmode.h (float_mode_for_size): New function.
15283 * emit-rtl.c (double_mode): Delete.
15284 (init_emit_once): Use float_mode_for_size.
15285 * stor-layout.c (layout_type): Likewise.
15286 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
15288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15289 Alan Hayward <alan.hayward@arm.com>
15290 David Sherwood <david.sherwood@arm.com>
15292 * output.h (assemble_real): Take a scalar_float_mode.
15293 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
15294 * config/arm/arm.md (consttable_4): Likewise.
15295 (consttable_8): Likewise.
15296 (consttable_16): Likewise.
15297 * config/mips/mips.md (consttable_float): Likewise.
15298 * config/s390/s390.c (s390_output_pool_entry): Likewise.
15299 * varasm.c (assemble_real): Take a scalar_float_mode.
15300 (output_constant_pool_2): Update accordingly.
15301 (output_constant): Likewise.
15303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15304 Alan Hayward <alan.hayward@arm.com>
15305 David Sherwood <david.sherwood@arm.com>
15307 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
15308 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
15309 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
15310 (native_encode_real): Likewise.
15311 (native_interpret_real): Likewise.
15312 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
15313 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15316 Alan Hayward <alan.hayward@arm.com>
15317 David Sherwood <david.sherwood@arm.com>
15319 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
15320 <scalar_float_mode>. Simplify.
15321 (gen_extend_conv_libfunc): Likewise.
15323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15324 Alan Hayward <alan.hayward@arm.com>
15325 David Sherwood <david.sherwood@arm.com>
15327 * coretypes.h (scalar_float_mode): New type.
15328 * machmode.h (mode_traits::from_int): Use machine_mode if
15329 USE_ENUM_MODES is defined.
15330 (is_a): New function.
15332 (dyn_cast): Likewise.
15333 (scalar_float_mode): New class.
15334 (scalar_float_mode::includes_p): New function.
15335 (is_float_mode): Likewise.
15336 * gdbhooks.py (MachineModePrinter): New class.
15337 (build_pretty_printer): Use it for scalar_float_mode.
15338 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
15339 (format_helper::format_helper): Turn into a template.
15340 * genmodes.c (get_mode_class): New function.
15341 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
15342 or machine_mode if none.
15343 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
15344 as_a <scalar_float_mode>.
15345 * dwarf2out.c (mem_loc_descriptor): Likewise.
15346 (insert_float): Likewise.
15347 (add_const_value_attribute): Likewise.
15348 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15349 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
15350 (expand_unop): Update accordingly.
15351 (expand_abs_nojump): Likewise.
15352 (expand_copysign_absneg): Take a scalar_float_mode.
15353 (expand_copysign_bit): Likewise.
15354 (expand_copysign): Update accordingly.
15356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15357 Alan Hayward <alan.hayward@arm.com>
15358 David Sherwood <david.sherwood@arm.com>
15360 * coretypes.h (opt_mode): New class.
15361 * machmode.h (opt_mode): Likewise.
15362 (opt_mode::else_void): New function.
15363 (opt_mode::require): Likewise.
15364 (opt_mode::exists): Likewise.
15365 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
15366 (GET_MODE_2XWIDER_MODE): Likewise.
15367 (mode_iterator::get_wider): Update accordingly.
15368 (mode_iterator::get_2xwider): Likewise.
15369 (mode_iterator::get_known_wider): Likewise, turning into a template.
15370 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
15371 forcing a wider mode to exist.
15372 * config/cr16/cr16.h (LONG_REG_P): Likewise.
15373 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15374 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
15375 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
15376 * lower-subreg.c (init_lower_subreg): Likewise.
15377 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
15378 on the final iteration.
15379 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
15380 a wider mode exists before asking for a move pattern.
15381 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
15382 forcing a wider mode to exist.
15383 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
15384 returning false if no such mode exists.
15385 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
15386 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
15387 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
15388 Avoid checking for a MODE_INT if we already know the mode is not a
15390 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
15391 forcing a wider mode to exist.
15392 (expmed_mult_highpart_optab): Likewise.
15393 (expmed_mult_highpart): Likewise.
15394 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
15396 * lto-streamer-in.c (lto_input_mode_table): Likewise.
15397 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
15398 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15399 * internal-fn.c (expand_mul_overflow): Update use of
15400 GET_MODE_2XWIDER_MODE.
15401 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15402 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
15403 GET_MODE_WIDER_MODE.
15404 (convert_plusminus_to_widen): Likewise.
15405 * tree-switch-conversion.c (array_value_type): Likewise.
15406 * var-tracking.c (emit_note_insn_var_location): Likewise.
15407 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15408 Return false inside rather than outside the loop if no wider mode
15410 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
15411 and GET_MODE_2XWIDER_MODE
15412 (can_compare_p): Use else_void.
15413 * gdbhooks.py (OptMachineModePrinter): New class.
15414 (build_pretty_printer): Use it for opt_mode.
15416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15417 Alan Hayward <alan.hayward@arm.com>
15418 David Sherwood <david.sherwood@arm.com>
15420 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
15421 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
15423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15424 Alan Hayward <alan.hayward@arm.com>
15425 David Sherwood <david.sherwood@arm.com>
15427 * machmode.h (mode_traits): New structure.
15428 (get_narrowest_mode): New function.
15429 (mode_iterator::start): Likewise.
15430 (mode_iterator::iterate_p): Likewise.
15431 (mode_iterator::get_wider): Likewise.
15432 (mode_iterator::get_known_wider): Likewise.
15433 (mode_iterator::get_2xwider): Likewise.
15434 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
15435 (FOR_EACH_MODE): Likewise.
15436 (FOR_EACH_MODE_FROM): Likewise.
15437 (FOR_EACH_MODE_UNTIL): Likewise.
15438 (FOR_EACH_WIDER_MODE): Likewise.
15439 (FOR_EACH_2XWIDER_MODE): Likewise.
15440 * builtins.c (expand_builtin_strlen): Use new mode iterators.
15441 * combine.c (simplify_comparison): Likewise
15442 * config/i386/i386.c (type_natural_mode): Likewise.
15443 * cse.c (cse_insn): Likewise.
15444 * dse.c (find_shift_sequence): Likewise.
15445 * emit-rtl.c (init_derived_machine_modes): Likewise.
15446 (init_emit_once): Likewise.
15447 * explow.c (hard_function_value): Likewise.
15448 * expmed.c (extract_fixed_bit_field_1): Likewise.
15449 (extract_bit_field_1): Likewise.
15450 (expand_divmod): Likewise.
15451 (emit_store_flag_1): Likewise.
15452 * expr.c (init_expr_target): Likewise.
15453 (convert_move): Likewise.
15454 (alignment_for_piecewise_move): Likewise.
15455 (widest_int_mode_for_size): Likewise.
15456 (emit_block_move_via_movmem): Likewise.
15457 (copy_blkmode_to_reg): Likewise.
15458 (set_storage_via_setmem): Likewise.
15459 (compress_float_constant): Likewise.
15460 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15461 * optabs-query.c (get_best_extraction_insn): Likewise.
15462 * optabs.c (expand_binop): Likewise.
15463 (expand_twoval_unop): Likewise.
15464 (expand_twoval_binop): Likewise.
15465 (widen_leading): Likewise.
15466 (widen_bswap): Likewise.
15467 (expand_parity): Likewise.
15468 (expand_unop): Likewise.
15469 (prepare_cmp_insn): Likewise.
15470 (prepare_float_lib_cmp): Likewise.
15471 (expand_float): Likewise.
15472 (expand_fix): Likewise.
15473 (expand_sfix_optab): Likewise.
15474 * postreload.c (move2add_use_add2_insn): Likewise.
15475 * reg-stack.c (reg_to_stack): Likewise.
15476 * reginfo.c (choose_hard_reg_mode): Likewise.
15477 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15478 * stor-layout.c (mode_for_size): Likewise.
15479 (smallest_mode_for_size): Likewise.
15480 (mode_for_vector): Likewise.
15481 (finish_bitfield_representative): Likewise.
15482 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
15483 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
15484 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15485 * var-tracking.c (prepare_call_arguments): Likewise.
15487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15488 Alan Hayward <alan.hayward@arm.com>
15489 David Sherwood <david.sherwood@arm.com>
15491 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
15492 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
15493 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
15494 * machmode.h (mode_size): Move earlier in file.
15495 (mode_precision): Likewise.
15496 (mode_inner): Likewise.
15497 (mode_nunits): Likewise.
15498 (mode_unit_size): Likewise.
15499 (unit_unit_precision): Likewise.
15500 (mode_wider): Likewise.
15501 (mode_2xwider): Likewise.
15502 (machine_mode): New class.
15503 (mode_to_bytes): New function.
15504 (mode_to_bits): Likewise.
15505 (mode_to_precision): Likewise.
15506 (mode_to_inner): Likewise.
15507 (mode_to_unit_size): Likewise.
15508 (mode_to_unit_precision): Likewise.
15509 (mode_to_nunits): Likewise.
15510 (GET_MODE_SIZE): Use mode_to_bytes.
15511 (GET_MODE_BITSIZE): Use mode_to_bits.
15512 (GET_MODE_PRECISION): Use mode_to_precision.
15513 (GET_MODE_INNER): Use mode_to_inner.
15514 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
15515 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
15516 (GET_MODE_NUNITS): Use mode_to_nunits.
15517 * system.h (ALWAYS_INLINE): New macro.
15518 * config/powerpcspe/powerpcspe-c.c
15519 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
15520 int for arg1_mode and arg2_mode.
15522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15523 Alan Hayward <alan.hayward@arm.com>
15524 David Sherwood <david.sherwood@arm.com>
15526 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15527 Prefix mode names with E_ in case statements.
15528 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15529 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
15530 (aarch64_split_simd_move): Likewise.
15531 (aarch64_gen_storewb_pair): Likewise.
15532 (aarch64_gen_loadwb_pair): Likewise.
15533 (aarch64_gen_store_pair): Likewise.
15534 (aarch64_gen_load_pair): Likewise.
15535 (aarch64_get_condition_code_1): Likewise.
15536 (aarch64_constant_pool_reload_icode): Likewise.
15537 (get_rsqrte_type): Likewise.
15538 (get_rsqrts_type): Likewise.
15539 (get_recpe_type): Likewise.
15540 (get_recps_type): Likewise.
15541 (aarch64_gimplify_va_arg_expr): Likewise.
15542 (aarch64_simd_container_mode): Likewise.
15543 (aarch64_emit_load_exclusive): Likewise.
15544 (aarch64_emit_store_exclusive): Likewise.
15545 (aarch64_expand_compare_and_swap): Likewise.
15546 (aarch64_gen_atomic_cas): Likewise.
15547 (aarch64_emit_bic): Likewise.
15548 (aarch64_emit_atomic_swap): Likewise.
15549 (aarch64_emit_atomic_load_op): Likewise.
15550 (aarch64_evpc_trn): Likewise.
15551 (aarch64_evpc_uzp): Likewise.
15552 (aarch64_evpc_zip): Likewise.
15553 (aarch64_evpc_ext): Likewise.
15554 (aarch64_evpc_rev): Likewise.
15555 (aarch64_evpc_dup): Likewise.
15556 (aarch64_gen_ccmp_first): Likewise.
15557 (aarch64_gen_ccmp_next): Likewise.
15558 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15559 (alpha_emit_xfloating_libcall): Likewise.
15560 (emit_insxl): Likewise.
15561 (alpha_arg_type): Likewise.
15562 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
15563 (arc_preferred_simd_mode): Likewise.
15564 (arc_secondary_reload): Likewise.
15565 (get_arc_condition_code): Likewise.
15566 (arc_print_operand): Likewise.
15567 (arc_legitimate_constant_p): Likewise.
15568 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15569 * config/arc/arc.md (casesi_load): Likewise.
15570 (casesi_compact_jump): Likewise.
15571 * config/arc/predicates.md (proper_comparison_operator): Likewise.
15572 (cc_use_register): Likewise.
15573 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15574 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15575 (arm_init_iwmmxt_builtins): Likewise.
15576 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
15577 (neon_expand_vector_init): Likewise.
15578 (arm_attr_length_move_neon): Likewise.
15579 (maybe_get_arm_condition_code): Likewise.
15580 (arm_emit_vector_const): Likewise.
15581 (arm_preferred_simd_mode): Likewise.
15582 (arm_output_iwmmxt_tinsr): Likewise.
15583 (thumb1_output_casesi): Likewise.
15584 (thumb2_output_casesi): Likewise.
15585 (arm_emit_load_exclusive): Likewise.
15586 (arm_emit_store_exclusive): Likewise.
15587 (arm_expand_compare_and_swap): Likewise.
15588 (arm_evpc_neon_vuzp): Likewise.
15589 (arm_evpc_neon_vzip): Likewise.
15590 (arm_evpc_neon_vrev): Likewise.
15591 (arm_evpc_neon_vtrn): Likewise.
15592 (arm_evpc_neon_vext): Likewise.
15593 (arm_validize_comparison): Likewise.
15594 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
15595 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
15596 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
15597 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
15598 (c6x_preferred_simd_mode): Likewise.
15599 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
15600 (epiphany_rtx_costs): Likewise.
15601 * config/epiphany/predicates.md (proper_comparison_operator):
15603 * config/frv/frv.c (condexec_memory_operand): Likewise.
15604 (frv_emit_move): Likewise.
15605 (output_move_single): Likewise.
15606 (output_condmove_single): Likewise.
15607 (frv_hard_regno_mode_ok): Likewise.
15608 (frv_matching_accg_mode): Likewise.
15609 * config/h8300/h8300.c (split_adds_subs): Likewise.
15610 (h8300_rtx_costs): Likewise.
15611 (h8300_print_operand): Likewise.
15612 (compute_mov_length): Likewise.
15613 (output_logical_op): Likewise.
15614 (compute_logical_op_length): Likewise.
15615 (compute_logical_op_cc): Likewise.
15616 (h8300_shift_needs_scratch_p): Likewise.
15617 (output_a_shift): Likewise.
15618 (compute_a_shift_length): Likewise.
15619 (compute_a_shift_cc): Likewise.
15620 (expand_a_rotate): Likewise.
15621 (output_a_rotate): Likewise.
15622 * config/i386/i386.c (classify_argument): Likewise.
15623 (function_arg_advance_32): Likewise.
15624 (function_arg_32): Likewise.
15625 (function_arg_64): Likewise.
15626 (function_value_64): Likewise.
15627 (ix86_gimplify_va_arg): Likewise.
15628 (ix86_legitimate_constant_p): Likewise.
15629 (put_condition_code): Likewise.
15630 (split_double_mode): Likewise.
15631 (ix86_avx256_split_vector_move_misalign): Likewise.
15632 (ix86_expand_vector_logical_operator): Likewise.
15633 (ix86_split_idivmod): Likewise.
15634 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
15635 (ix86_build_const_vector): Likewise.
15636 (ix86_build_signbit_mask): Likewise.
15637 (ix86_match_ccmode): Likewise.
15638 (ix86_cc_modes_compatible): Likewise.
15639 (ix86_expand_branch): Likewise.
15640 (ix86_expand_sse_cmp): Likewise.
15641 (ix86_expand_sse_movcc): Likewise.
15642 (ix86_expand_int_sse_cmp): Likewise.
15643 (ix86_expand_vec_perm_vpermi2): Likewise.
15644 (ix86_expand_vec_perm): Likewise.
15645 (ix86_expand_sse_unpack): Likewise.
15646 (ix86_expand_int_addcc): Likewise.
15647 (ix86_split_to_parts): Likewise.
15648 (ix86_vectorize_builtin_gather): Likewise.
15649 (ix86_vectorize_builtin_scatter): Likewise.
15650 (avx_vpermilp_parallel): Likewise.
15651 (inline_memory_move_cost): Likewise.
15652 (ix86_tieable_integer_mode_p): Likewise.
15653 (x86_maybe_negate_const_int): Likewise.
15654 (ix86_expand_vector_init_duplicate): Likewise.
15655 (ix86_expand_vector_init_one_nonzero): Likewise.
15656 (ix86_expand_vector_init_one_var): Likewise.
15657 (ix86_expand_vector_init_concat): Likewise.
15658 (ix86_expand_vector_init_interleave): Likewise.
15659 (ix86_expand_vector_init_general): Likewise.
15660 (ix86_expand_vector_set): Likewise.
15661 (ix86_expand_vector_extract): Likewise.
15662 (emit_reduc_half): Likewise.
15663 (ix86_emit_i387_round): Likewise.
15664 (ix86_mangle_type): Likewise.
15665 (ix86_expand_round_sse4): Likewise.
15666 (expand_vec_perm_blend): Likewise.
15667 (canonicalize_vector_int_perm): Likewise.
15668 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
15669 (expand_vec_perm_1): Likewise.
15670 (expand_vec_perm_interleave3): Likewise.
15671 (expand_vec_perm_even_odd_pack): Likewise.
15672 (expand_vec_perm_even_odd_1): Likewise.
15673 (expand_vec_perm_broadcast_1): Likewise.
15674 (ix86_vectorize_vec_perm_const_ok): Likewise.
15675 (ix86_expand_vecop_qihi): Likewise.
15676 (ix86_expand_mul_widen_hilo): Likewise.
15677 (ix86_expand_sse2_abs): Likewise.
15678 (ix86_expand_pextr): Likewise.
15679 (ix86_expand_pinsr): Likewise.
15680 (ix86_preferred_simd_mode): Likewise.
15681 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
15682 * config/i386/sse.md (*andnot<mode>3): Likewise.
15683 (<mask_codefor><code><mode>3<mask_name>): Likewise.
15684 (*<code><mode>3): Likewise.
15685 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
15686 (ia64_expand_atomic_op): Likewise.
15687 (ia64_arg_type): Likewise.
15688 (ia64_mode_to_int): Likewise.
15689 (ia64_scalar_mode_supported_p): Likewise.
15690 (ia64_vector_mode_supported_p): Likewise.
15691 (expand_vec_perm_broadcast): Likewise.
15692 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
15693 (iq2000_function_arg_advance): Likewise.
15694 (iq2000_function_arg): Likewise.
15695 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
15696 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
15697 (m68k_libcall_value): Likewise.
15698 (m68k_function_value): Likewise.
15699 (sched_attr_op_type): Likewise.
15700 * config/mcore/mcore.c (mcore_output_move): Likewise.
15701 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
15703 (microblaze_function_arg): Likewise.
15704 * config/mips/mips.c (mips16_build_call_stub): Likewise.
15705 (mips_print_operand): Likewise.
15706 (mips_mode_ok_for_mov_fmt_p): Likewise.
15707 (mips_vector_mode_supported_p): Likewise.
15708 (mips_preferred_simd_mode): Likewise.
15709 (mips_expand_vpc_loongson_even_odd): Likewise.
15710 (mips_expand_vec_unpack): Likewise.
15711 (mips_expand_vi_broadcast): Likewise.
15712 (mips_expand_vector_init): Likewise.
15713 (mips_expand_vec_reduc): Likewise.
15714 (mips_expand_msa_cmp): Likewise.
15715 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
15716 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
15717 (cc_flags_for_mode): Likewise.
15718 * config/msp430/msp430.c (msp430_print_operand): Likewise.
15719 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
15720 (nds32_output_casesi_pc_relative): Likewise.
15721 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15722 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
15723 (nvptx_gen_unpack): Likewise.
15724 (nvptx_gen_pack): Likewise.
15725 (nvptx_gen_shuffle): Likewise.
15726 (nvptx_gen_wcast): Likewise.
15727 (nvptx_preferred_simd_mode): Likewise.
15728 * config/pa/pa.c (pa_secondary_reload): Likewise.
15729 * config/pa/predicates.md (base14_operand): Likewise.
15730 * config/powerpcspe/powerpcspe-c.c
15731 (altivec_resolve_overloaded_builtin): Likewise.
15732 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
15734 (rs6000_preferred_simd_mode): Likewise.
15735 (output_vec_const_move): Likewise.
15736 (rs6000_expand_vector_extract): Likewise.
15737 (rs6000_split_vec_extract_var): Likewise.
15738 (reg_offset_addressing_ok_p): Likewise.
15739 (rs6000_legitimate_offset_address_p): Likewise.
15740 (rs6000_legitimize_address): Likewise.
15741 (rs6000_emit_set_const): Likewise.
15742 (rs6000_const_vec): Likewise.
15743 (rs6000_emit_move): Likewise.
15744 (spe_build_register_parallel): Likewise.
15745 (rs6000_darwin64_record_arg_recurse): Likewise.
15746 (swap_selector_for_mode): Likewise.
15747 (spe_init_builtins): Likewise.
15748 (paired_init_builtins): Likewise.
15749 (altivec_init_builtins): Likewise.
15750 (do_load_for_compare): Likewise.
15751 (rs6000_generate_compare): Likewise.
15752 (rs6000_expand_float128_convert): Likewise.
15753 (emit_load_locked): Likewise.
15754 (emit_store_conditional): Likewise.
15755 (rs6000_output_function_epilogue): Likewise.
15756 (rs6000_handle_altivec_attribute): Likewise.
15757 (rs6000_function_value): Likewise.
15758 (emit_fusion_gpr_load): Likewise.
15759 (emit_fusion_p9_load): Likewise.
15760 (emit_fusion_p9_store): Likewise.
15761 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
15762 (fusion_gpr_mem_load): Likewise.
15763 (fusion_addis_mem_combo_load): Likewise.
15764 (fusion_addis_mem_combo_store): Likewise.
15765 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
15766 (fusion_gpr_mem_load): Likewise.
15767 (fusion_addis_mem_combo_load): Likewise.
15768 (fusion_addis_mem_combo_store): Likewise.
15769 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15771 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
15772 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15773 (rs6000_preferred_simd_mode): Likewise.
15774 (output_vec_const_move): Likewise.
15775 (rs6000_expand_vector_extract): Likewise.
15776 (rs6000_split_vec_extract_var): Likewise.
15777 (reg_offset_addressing_ok_p): Likewise.
15778 (rs6000_legitimate_offset_address_p): Likewise.
15779 (rs6000_legitimize_address): Likewise.
15780 (rs6000_emit_set_const): Likewise.
15781 (rs6000_const_vec): Likewise.
15782 (rs6000_emit_move): Likewise.
15783 (rs6000_darwin64_record_arg_recurse): Likewise.
15784 (swap_selector_for_mode): Likewise.
15785 (paired_init_builtins): Likewise.
15786 (altivec_init_builtins): Likewise.
15787 (rs6000_expand_float128_convert): Likewise.
15788 (emit_load_locked): Likewise.
15789 (emit_store_conditional): Likewise.
15790 (rs6000_output_function_epilogue): Likewise.
15791 (rs6000_handle_altivec_attribute): Likewise.
15792 (rs6000_function_value): Likewise.
15793 (emit_fusion_gpr_load): Likewise.
15794 (emit_fusion_p9_load): Likewise.
15795 (emit_fusion_p9_store): Likewise.
15796 * config/rx/rx.c (rx_gen_move_template): Likewise.
15797 (flags_from_mode): Likewise.
15798 * config/s390/predicates.md (s390_alc_comparison): Likewise.
15799 (s390_slb_comparison): Likewise.
15800 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
15801 (s390_vector_mode_supported_p): Likewise.
15802 (s390_cc_modes_compatible): Likewise.
15803 (s390_match_ccmode_set): Likewise.
15804 (s390_canonicalize_comparison): Likewise.
15805 (s390_emit_compare_and_swap): Likewise.
15806 (s390_branch_condition_mask): Likewise.
15807 (s390_rtx_costs): Likewise.
15808 (s390_secondary_reload): Likewise.
15809 (__SECONDARY_RELOAD_CASE): Likewise.
15810 (s390_expand_cs): Likewise.
15811 (s390_preferred_simd_mode): Likewise.
15812 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
15813 * config/sh/sh.c (sh_print_operand): Likewise.
15814 (dump_table): Likewise.
15815 (sh_secondary_reload): Likewise.
15816 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15817 * config/sh/sh.md (casesi_worker_1): Likewise.
15818 (casesi_worker_2): Likewise.
15819 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
15820 (fcc_comparison_operator): Likewise.
15821 * config/sparc/sparc.c (sparc_expand_move): Likewise.
15822 (emit_soft_tfmode_cvt): Likewise.
15823 (sparc_preferred_simd_mode): Likewise.
15824 (output_cbranch): Likewise.
15825 (sparc_print_operand): Likewise.
15826 (sparc_expand_vec_perm_bmask): Likewise.
15827 (vector_init_bshuffle): Likewise.
15828 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15829 (spu_vector_mode_supported_p): Likewise.
15830 (spu_expand_insv): Likewise.
15831 (spu_emit_branch_or_set): Likewise.
15832 (spu_handle_vector_attribute): Likewise.
15833 (spu_builtin_splats): Likewise.
15834 (spu_builtin_extract): Likewise.
15835 (spu_builtin_promote): Likewise.
15836 (spu_expand_sign_extend): Likewise.
15837 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15838 (tilegx_simd_int): Likewise.
15839 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
15840 (tilepro_simd_int): Likewise.
15841 * config/v850/v850.c (const_double_split): Likewise.
15842 (v850_print_operand): Likewise.
15843 (ep_memory_offset): Likewise.
15844 * config/vax/vax.c (vax_rtx_costs): Likewise.
15845 (vax_output_int_move): Likewise.
15846 (vax_output_int_add): Likewise.
15847 (vax_output_int_subtract): Likewise.
15848 * config/visium/predicates.md (visium_branch_operator): Likewise.
15849 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
15850 (visium_print_operand_address): Likewise.
15851 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15852 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
15853 (xtensa_expand_conditional_branch): Likewise.
15854 (xtensa_copy_incoming_a7): Likewise.
15855 (xtensa_output_literal): Likewise.
15856 * dfp.c (decimal_real_maxval): Likewise.
15857 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15860 Alan Hayward <alan.hayward@arm.com>
15861 David Sherwood <david.sherwood@arm.com>
15863 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
15864 (mode_nunits_inline): Likewise.
15865 (mode_inner_inline): Likewise.
15866 (mode_unit_size_inline): Likewise.
15867 (mode_unit_precision_inline): Likewise.
15868 (emit_insn_modes_h): Likewise. Also emit a #define of the
15870 (emit_mode_wider): Add an E_ prefix to mode names.
15871 (emit_mode_complex): Likewise.
15872 (emit_mode_inner): Likewise.
15873 (emit_mode_adjustments): Likewise.
15874 (emit_mode_int_n): Likewise.
15875 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
15876 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
15877 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
15878 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
15879 (CRC32_BUILTIN, ENTRY): Likewise.
15880 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
15881 (aarch64_pop_regs): Likewise.
15882 (aarch64_process_components): Likewise.
15883 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
15884 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
15885 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
15886 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
15887 * config/arm/arm.c (arm_init_libfuncs): Likewise.
15888 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
15890 * config/i386/i386-builtin.def (pcmpestr): Likewise.
15891 (pcmpistr): Likewise.
15892 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
15893 * config/mmix/mmix.c (mmix_output_condition): Likewise.
15894 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15896 * config/rl78/rl78.c (mduc_regs): Likewise.
15897 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15898 (htm_expand_builtin): Likewise.
15899 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
15900 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
15901 * config/xtensa/xtensa.c (print_operand): Likewise.
15902 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
15903 (NUM_MODE_VECTOR_INT): Likewise.
15904 * genoutput.c (null_operand): Likewise.
15905 (output_operand_data): Likewise.
15906 * genrecog.c (print_parameter_value): Likewise.
15907 * lra.c (debug_operand_data): Likewise.
15909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15911 * dwarf2out.c (scompare_loc_descriptor_wide)
15912 (scompare_loc_descriptor_narrow): New functions, split out from...
15913 (scompare_loc_descriptor): ...here.
15914 * expmed.c (emit_store_flag_int): New function, split out from...
15915 (emit_store_flag): ...here.
15917 2017-08-30 Richard Biener <rguenther@suse.de>
15919 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
15920 (dwarf2out_early_finish): Move setting of AT_pubnames from
15921 early debug output to early finish.
15923 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15925 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
15926 and -mdata-region to the assembler.
15928 2017-08-30 Richard Biener <rguenther@suse.de>
15930 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
15932 (gen_subprogram_die): Add DW_AT_object_pointer only early.
15933 (dwarf2out_early_global_decl): Only generate a DIE for the
15934 abstract origin if it doesn't already exist or is a declaration DIE.
15935 (resolve_addr): Do not add the linkage name twice when
15936 generating a stub DIE for the DW_TAG_GNU_call_site target.
15938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15940 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15941 Use machine_mode rather than int for arg1_mode.
15943 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
15946 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
15947 that the second argument of the built-in functions to unpack
15948 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
15949 switch statement instead a lot of if statements.
15950 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
15951 Allow 64-bit values to be in Altivec registers as well as
15952 traditional floating point registers.
15953 (pack<mode>, FMOVE128_VSX iterator): Likewise.
15955 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
15957 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
15958 MAX_REGS_PER_ADDRESS == 1.
15960 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
15962 * config/i386/i386.opt (flag_fentry): Do not init to -1.
15963 * config/i386/i386.c (ix86_option_override_internal): Simplify
15964 setting of opts->x_flag_entry.
15966 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15967 Jakub Jelinek <jakub@redhat.com>
15968 Richard Biener <rguenther@suse.de>
15970 PR tree-optimization/81503
15971 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
15972 folded constant fits in the target type; reorder tests for clarity.
15974 2017-08-29 Martin Liska <mliska@suse.cz>
15976 * passes.def: Include pass_lower_switch.
15977 * stmt.c (dump_case_nodes): Remove and move to
15978 tree-switch-conversion.
15979 (case_values_threshold): Likewise.
15980 (expand_switch_as_decision_tree_p): Likewise.
15981 (emit_case_decision_tree): Likewise.
15982 (expand_case): Likewise.
15983 (balance_case_nodes): Likewise.
15984 (node_has_low_bound): Likewise.
15985 (node_has_high_bound): Likewise.
15986 (node_is_bounded): Likewise.
15987 (emit_case_nodes): Likewise.
15988 (struct simple_case_node): New struct.
15989 (add_case_node): Remove.
15990 (emit_case_dispatch_table): Use vector instead of case_list.
15991 (reset_out_edges_aux): Remove.
15992 (compute_cases_per_edge): Likewise.
15993 (expand_case): Build list of simple_case_node.
15994 (expand_sjlj_dispatch_table): Use it.
15995 * tree-switch-conversion.c (struct case_node): Moved from
15996 stmt.c and adjusted.
15997 (emit_case_nodes): Likewise.
15998 (node_has_low_bound): Likewise.
15999 (node_has_high_bound): Likewise.
16000 (node_is_bounded): Likewise.
16001 (case_values_threshold): Likewise.
16002 (reset_out_edges_aux): Likewise.
16003 (compute_cases_per_edge): Likewise.
16004 (add_case_node): Likewise.
16005 (dump_case_nodes): Likewise.
16006 (balance_case_nodes): Likewise.
16007 (expand_switch_as_decision_tree_p): Likewise.
16008 (emit_jump): Likewise.
16009 (emit_case_decision_tree): Likewise.
16010 (try_switch_expansion): Likewise.
16011 (do_jump_if_equal): Likewise.
16012 (emit_cmp_and_jump_insns): Likewise.
16013 (fix_phi_operands_for_edge): New function.
16014 (record_phi_operand_mapping): Likewise.
16015 (class pass_lower_switch): New pass.
16016 (pass_lower_switch::execute): New function.
16017 (make_pass_lower_switch): Likewise.
16018 (conditional_probability):
16019 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16020 * tree-pass.h: Add make_pass_lower_switch.
16022 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16025 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16028 2017-08-29 Richard Biener <rguenther@suse.de>
16030 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16031 we do not add a DW_AT_inline attribute twice.
16032 (gen_subprogram_die): Remove code setting DW_AT_inline on
16033 DECL_ABSTRACT_P nodes.
16035 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16037 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16038 calls to internal functions.
16039 (gimplify_modify_expr): Likewise.
16040 * tree-call-cdce.c (use_internal_fn): Likewise.
16041 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16042 (convert_to_divmod): Set the nothrow flag.
16043 * tree-if-conv.c (predicate_mem_writes): Likewise.
16044 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16045 (vectorizable_call): Likewise.
16046 (vectorizable_store): Likewise.
16047 (vectorizable_load): Likewise.
16048 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16049 (vect_recog_mask_conversion_pattern): Likewise.
16051 2017-08-29 Martin Liska <mliska@suse.cz>
16054 * gcc.c (driver_handle_option): Add new argument.
16055 * opts-common.c (handle_option): Pass
16056 target_option_override_hook.
16057 * opts-global.c (lang_handle_option): Add new option.
16058 (set_default_handlers): Add new argument.
16059 (decode_options): Likewise.
16060 * opts.c (target_handle_option): Likewise.
16061 (common_handle_option): Call target_option_override_hook.
16062 * opts.h (struct cl_option_handler_func): Add hook for
16063 target option override.
16064 (struct cl_option_handlers): Likewise.
16065 (set_default_handlers): Add new argument.
16066 (decode_options): Likewise.
16067 (common_handle_option): Likewise.
16068 (target_handle_option): Likewise.
16069 * toplev.c (toplev::main): Pass targetm.target_option.override
16072 2017-08-29 Richard Biener <rguenther@suse.de>
16073 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16075 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16076 life to the active subtree.
16078 2017-08-28 Jeff Law <law@redhat.com>
16080 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16081 derive_equivalences.
16082 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16083 Code moved into....
16084 (edge_info::derive_equivalences): New private member function
16086 * tree-ssa-dom.c (class edge_info): Changed from a struct
16087 to a class. Add ctor/dtor, methods and data members.
16088 (edge_info::edge_info): Renamed from allocate_edge_info.
16089 Initialize additional members.
16090 (edge_info::~edge_info): New.
16091 (free_dom_edge_info): Delete the edge info.
16092 (record_edge_info): Use new class & associated member functions.
16093 Tighten forms for testing for edge equivalences.
16094 (record_temporary_equivalences): Iterate over the simple
16095 equivalences rather than assuming there's only one per edge.
16096 (cprop_into_successor_phis): Iterate over the simple
16097 equivalences rather than assuming there's only one per edge.
16098 (optimize_stmt): Use operand_equal_p rather than pointer
16099 equality for mini-DSE code.
16101 2017-08-28 Nathan Sidwell <nathan@acm.org>
16103 * gcc.c (execute): Fold SIGPIPE handling into switch
16104 statement. Adjust internal error message.
16106 2017-08-28 Richard Biener <rguenther@suse.de>
16109 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16110 Do nothing for removed DIEs.
16112 2017-08-28 Richard Biener <rguenther@suse.de>
16114 PR tree-optimization/81977
16115 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16118 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16121 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16122 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16123 using targetm.gen_mem_thread_fence.
16125 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16128 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16129 predicate to register_operand. Reorder operands.
16130 (*btr<mode>): Ditto.
16131 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16132 (*btr<mode>_mask): Ditto.
16134 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16136 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16137 * config/rs6000/xmmintrin.h: New file.
16138 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16140 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16143 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16144 parameter and_insn and return it.
16145 (recombine_lvx_pattern): Insert a copy to ensure availability of
16146 the base register of the copied masking operation at the point of
16147 the instruction replacement.
16148 (recombine_stvx_pattern): Likewise.
16150 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16152 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16153 undocumented switches.
16154 (-mpower9-dform-vector): Likewise.
16155 (-mpower9-dform): Likewise.
16156 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16157 comments to delete references to -mpower9-dform* switches.
16158 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16159 Delete reference to -mpower9-dform* switches, test for
16160 -mpower9-vector instead.
16161 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16162 (OTHER_P9_VECTOR_MASKS): Likewise.
16163 (POWERPC_MASKS): Likewise.
16164 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16165 tests against -mpower9-dform* to -mpower9-vector. Delete code
16166 that checked for -mpower9-dform* consistancy with other options.
16167 Add test for -mpower9-misc to enable other power9 switches.
16168 (rs6000_init_hard_regno_mode_ok): Likewise.
16169 (rs6000_option_override_internal): Likewise.
16170 (rs6000_emit_prologue): Likewise.
16171 (rs6000_emit_epilogue): Likewise.
16172 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16173 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16174 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16176 (emit_fusion_p9_store): Likewise.
16177 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16178 resetting these macros if the assembler does not support ISA 3.0
16180 (TARGET_P9_DFORM_VECTOR): Likewise.
16181 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16182 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16184 2017-08-25 Alan Modra <amodra@gmail.com>
16186 PR rtl-optimization/81747
16187 * cse.c (cse_extended_basic_block): Don't attempt to record
16188 equivalences for degenerate conditional branches that branch
16189 to their fall-through.
16191 2017-08-24 Martin Sebor <msebor@redhat.com>
16193 PR middle-end/81908
16194 * gimple-fold.c (size_must_be_zero_p): New function.
16195 (gimple_fold_builtin_memory_op): Call it.
16197 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16199 * config/rs6000/mm_malloc.h: New file.
16201 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16203 PR tree-optimization/81913
16204 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16205 analysis when either IVs in condition can wrap.
16207 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16209 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16210 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
16212 2017-08-24 Richard Biener <rguenther@suse.de>
16215 * targhooks.c (default_target_can_inline_p): Properly
16216 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
16217 is present and always compare.
16218 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
16219 infer -mfpmath=sse from TARGET_SSE_P.
16220 (ix86_can_inline_p): Properly use target_option_default_node when
16221 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
16223 2017-08-24 Richard Biener <rguenther@suse.de>
16226 * dwarf2out.c (output_die): Handle flag_generate_offload like
16228 (output_comp_unit): Likewise.
16229 (gen_array_type_die): Likewise.
16230 (dwarf2out_early_finish): Likewise.
16231 (note_variable_value_in_expr): Likewise.
16232 (dwarf2out_finish): Likewise. Adjust assert.
16233 * cgraphunit.c (symbol_table::compile): Move setting of
16234 flag_generate_offload earlier ...
16235 (symbol_table::finalize_compilation_unit): ... here, before
16236 early debug finalization.
16238 2017-08-24 Richard Biener <rguenther@suse.de>
16240 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
16241 and ipa-fnsummary.h.
16242 (ix86_can_inline_p): When ix86_fpmath flags do not match
16243 check whether the callee uses FP math at all.
16245 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
16247 PR middle-end/81931
16248 * tree-ssanames.c (get_nonzero_bits): Use element_precision
16249 instead of TYPE_PRECISION.
16251 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
16252 Alan Hayward <alan.hayward@arm.com>
16253 David Sherwood <david.sherwood@arm.com>
16255 * combine.c (make_extraction): Use subreg_offset_from_lsb.
16257 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
16259 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
16261 (ix86_frame::stack_realign_allocate): New field.
16262 (struct machine_frame_state): Modify comments.
16263 (machine_frame_state::sp_realigned_fp_end): New field.
16264 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
16265 layout calculation.
16266 (sp_valid_at): Add assertion to assure no attempt to access invalid
16267 offset of a realigned stack.
16268 (fp_valid_at): Likewise.
16269 (choose_baseaddr): Modify comments.
16270 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
16271 ix86_expand_prologue.
16272 (ix86_expand_prologue): Modify stack realignment and allocation.
16273 (ix86_expand_epilogue): Modify comments.
16274 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
16275 avx2_runtime, avx512f, and avx512f_runtime.
16277 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
16279 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
16280 (mstackrealign): Do not init to -1.
16281 * config/i386/i386.c (ix86_option_override_internal):
16282 Check opts_set, not opts when setting default value of
16283 opts->x_ix86_force_align_arg_pointer.
16285 2017-08-23 Richard Biener <rguenther@suse.de>
16287 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
16288 lang_hooks.decl_printable_name.
16289 * print-rtl-function.c (print_rtx_function): Likewise.
16290 * tree-pretty-print.c (dump_function_header): Likewise.
16292 2017-08-23 Richard Biener <rguenther@suse.de>
16295 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
16296 -g0 at compile-time.
16298 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16300 PR middle-end/19706
16301 * doc/sourcebuild.texi (Other hardware attributes):
16304 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16306 PR middle-end/19706
16307 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
16308 Fix single-use check.
16310 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
16312 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
16314 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16316 * doc/install.texi: Modify to add more details on running selected
16319 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16321 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
16322 is combined with -mabi=ms.
16323 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
16326 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16328 PR tree-optimization/81488
16329 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
16330 and cached_basis fields.
16331 (MAX_SPREAD): New constant.
16332 (alloc_cand_and_find_basis): Initialize new fields.
16333 (clear_visited): New function.
16334 (create_phi_basis_1): Rename from create_phi_basis, set visited
16335 and cached_basis fields.
16336 (create_phi_basis): New wrapper function.
16337 (phi_add_costs_1): Rename from phi_add_costs, add spread
16338 parameter, set visited field, short-circuit when limits reached.
16339 (phi_add_costs): New wrapper function.
16340 (record_phi_increments_1): Rename from record_phi_increments, set
16342 (record_phi_increments): New wrapper function.
16343 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
16344 (phi_incr_cost): New wrapper function.
16345 (all_phi_incrs_profitable_1): Rename from
16346 all_phi_incrs_profitable, set visited field.
16347 (all_phi_incrs_profitable): New wrapper function.
16349 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16350 Alan Hayward <alan.hayward@arm.com>
16351 David Sherwood <david.sherwood@arm.com>
16353 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
16354 that takes the outer and inner modes.
16355 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
16356 comparison as the canonical test for a paradoxical subreg.
16357 * combine.c (simplify_set): Use paradoxical_subreg_p.
16358 (make_extraction): Likewise.
16359 (force_to_mode): Likewise.
16360 (rtx_equal_for_field_assignment_p): Likewise.
16361 (gen_lowpart_for_combine): Likewise.
16362 (simplify_comparison): Likewise.
16363 * cse.c (equiv_constant): Likewise.
16364 * expmed.c (store_bit_field_1): Likewise.
16365 * final.c (alter_subreg): Likewise.
16366 * fwprop.c (propagate_rtx): Likewise.
16367 (forward_propagate_subreg): Likewise.
16368 * ira-conflicts.c (ira_build_conflicts): Likewise.
16369 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
16370 * lra-constraints.c (curr_insn_transform): Likewise.
16371 (split_reg): Likewise.
16372 * lra-eliminations.c (move_plus_up): Likewise.
16373 (lra_eliminate_regs_1): Likewise.
16374 * recog.c (general_operand): Likewise.
16375 * ree.c (combine_reaching_defs): Likewise.
16376 * reload.c (push_reload): Likewise.
16377 (find_reloads): Likewise.
16378 * reload1.c (elimination_effects): Likewise.
16379 (compute_reload_subreg_offset): Likewise.
16380 (choose_reload_regs): Likewise.
16381 * rtlanal.c (subreg_lsb_1): Likewise.
16382 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16383 (simplify_subreg): Likewise.
16384 * var-tracking.c (track_loc_p): Likewise.
16385 * emit-rtl.c (byte_lowpart_offset): Likewise.
16386 (paradoxical_subreg_p): Delete out-of-line definition.
16388 2017-08-22 Jeff Law <law@redhat.com>
16390 PR tree-optimization/81741
16391 PR tree-optimization/71947
16392 * tree-ssa-dom.c: Include tree-inline.h.
16393 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
16394 equivalences if one is more expensive to compute than the other.
16395 * tree-ssa-scopedtables.h (class const_or_copies): Make
16396 record_const_or_copy_raw method private.
16397 (class avail_exprs_stack): New method simplify_binary_operation.
16398 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
16399 avail_exprs_stack::simplify_binary_operation as needed.
16400 (avail_exprs_stack::simplify_binary_operation): New function.
16402 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
16404 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
16405 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
16406 (DOT_SYMBOLS): Likewise.
16407 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
16408 (RELOCATABLE_NEEDS_FIXUP): Likewise.
16409 (RS6000_ABI_NAME): Likewise.
16410 (TARGET_CMODEL): Likewise.
16411 (TOC_SECTION_ASM_OP): Likewise.
16412 (SET_CMODEL): New macro.
16413 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
16415 2017-08-22 Richard Biener <rguenther@suse.de>
16417 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
16418 to allow for free-lang-data replacements similar to verify_type_variant.
16420 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16421 Alan Hayward <alan.hayward@arm.com>
16422 David Sherwood <david.sherwood@arm.com>
16424 * config/aarch64/aarch64.md (casesi): Use DImode rather than
16425 VOIDmode for the LABEL_REF.
16427 2017-08-22 Richard Biener <rguenther@suse.de>
16429 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
16431 2017-08-22 Richard Biener <rguenther@suse.de>
16433 * common.opt (feliminate-dwarf2-dups): Ignore.
16434 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
16435 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
16436 same_die_p_wrap, compute_section_prefix,
16437 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
16438 (comdat_symbol_id, comdat_symbol_number): Likewise.
16439 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
16441 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
16442 (output_die): Mark unreachable path unreachable.
16443 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
16444 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
16445 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
16446 (dwarf2out_early_finish): Likewise.
16448 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
16450 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
16452 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
16455 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
16456 not VAR_P. Filter attribute warnings with OPT_Wattributes.
16457 (avr_attribute_table) <io, io_low, address>: Initialize
16458 .decl_required with true.
16460 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
16462 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
16463 undocumented debugging options.
16464 (-mvsx-scalar-double): Likewise.
16465 (-mallow-df-permute): Likewise.
16466 (-mvectorize-builtins): Likewise.
16467 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16468 (rs6000_builtin_vectorized_function): Likewise.
16469 (rs6000_builtin_md_vectorized_function): Likewise.
16470 (rs6000_opt_vars): Likewise.
16472 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
16475 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
16476 (*btrq_imm): Rename from *btrq.
16477 (*btcq_imm): Rename from *btcq.
16478 (btsc): New code attribute.
16479 (*<btsc><mode>): New insn pattern.
16480 (*btr<mode>): Ditto.
16481 (*<btsc><mode>_mask): New insn_and_split pattern.
16482 (*btr<mode>_mask): Ditto.
16484 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16485 Alan Hayward <alan.hayward@arm.com>
16486 David Sherwood <david.sherwood@arm.com>
16488 * function.c (pad_below): Simplify padding calculation.
16490 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16491 Alan Hayward <alan.hayward@arm.com>
16492 David Sherwood <david.sherwood@arm.com>
16494 * target.def (function_prologue): Remove frame size argument.
16495 (function_epilogue): Likewise.
16496 * doc/tm.texi: Regenerate.
16497 * final.c (final_start_function): Update call to function_prologue.
16498 (final_end_function): Update call to function_epilogue.
16499 (default_function_pro_epilogue): Remove frame size argument.
16500 * output.h (default_function_pro_epilogue): Likewise.
16501 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
16502 (arm_output_function_prologue): Likewise.
16503 * config/frv/frv.c (frv_function_prologue): Likewise.
16504 (frv_function_epilogue): Likewise.
16505 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
16506 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
16507 (ia64_output_function_epilogue): Likewise.
16508 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
16509 (m32r_output_function_epilogue): Likewise.
16510 * config/microblaze/microblaze.c (microblaze_function_prologue)
16511 (microblaze_function_epilogue): Likewise.
16512 * config/mips/mips.c (mips_output_function_prologue): Likewise.
16513 (mips_output_function_epilogue): Likewise.
16514 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
16515 (mmix_target_asm_function_epilogue): Likewise.
16516 * config/msp430/msp430.c (msp430_start_function): Likewise.
16517 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
16518 (nds32_asm_function_epilogue): Likewise.
16519 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
16520 * config/pa/pa.c (pa_output_function_prologue): Likewise.
16521 (pa_output_function_epilogue): Likewise.
16522 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
16523 (rs6000_output_function_epilogue): Likewise.
16524 * config/rl78/rl78.c (rl78_start_function): Likewise.
16525 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
16526 (rs6000_output_function_epilogue): Likewise.
16527 * config/rx/rx.c (rx_output_function_prologue): Likewise.
16528 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
16529 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
16530 (sparc_asm_function_epilogue): Likewise.
16532 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16534 * tree.h (type_has_mode_precision_p): New function.
16535 * convert.c (convert_to_integer_1): Use it.
16536 * expr.c (expand_expr_real_2): Likewise.
16537 (expand_expr_real_1): Likewise.
16538 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
16539 * match.pd: Likewise.
16540 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
16541 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
16542 * tree-tailcall.c (process_assignment): Likewise.
16543 * tree-vect-loop.c (vectorizable_reduction): Likewise.
16544 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
16545 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
16546 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16547 (vectorizable_assignment): Likewise.
16548 (vectorizable_shift): Likewise.
16549 (vectorizable_operation): Likewise.
16550 * tree-vrp.c (register_edge_assert_for_2): Likewise.
16552 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
16554 * match.pd: Add pow (C, x) simplification.
16556 2017-08-21 Richard Biener <rguenther@suse.de>
16558 PR tree-optimization/81900
16559 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
16560 for blocks with abnormal predecessors.
16561 (compute_antic): Do not set visited flag prematurely.
16563 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
16566 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
16568 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16570 * stor-layout.h (vector_type_mode): Move to...
16571 * tree.h (vector_type_mode): ...here.
16572 * stor-layout.c (vector_type_mode): Move to...
16573 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
16575 2017-08-21 Richard Biener <rguenther@suse.de>
16577 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
16578 register_external_die hooks.
16579 (debug_false_tree_charstarstar_uhwistar): Declare.
16580 (debug_nothing_tree_charstar_uhwi): Likewise.
16581 * debug.c (do_nothing_debug_hooks): Adjust.
16582 (debug_false_tree_charstarstar_uhwistar): New do nothing.
16583 (debug_nothing_tree_charstar_uhwi): Likewise.
16584 * dbxout.c (dbx_debug_hooks): Adjust.
16585 (xcoff_debug_hooks): Likewise.
16586 * sdbout.c (sdb_debug_hooks): Likewise.
16587 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16588 * dwarf2out.c (macinfo_label_base): New global.
16589 (dwarf2out_register_external_die): New function for the
16590 register_external_die hook.
16591 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
16592 (dwarf2_debug_hooks): Use them.
16593 (dwarf2_lineno_debug_hooks): Adjust.
16594 (struct die_struct): Add with_offset flag.
16595 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
16596 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
16597 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
16598 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
16599 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
16600 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
16601 defining section names for the early LTO debug variants.
16602 (reset_indirect_string): New helper.
16603 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
16604 (print_dw_val): Add support for offsetted symbol references.
16605 (get_ultimate_context): Split out from is_cxx.
16606 (is_cxx): Use get_ultimate_context.
16607 (is_fortran): Add decl overload.
16608 (compute_comp_unit_symbol): Split out worker from
16609 compute_section_prefix.
16610 (compute_section_prefix): Call compute_comp_unit_symbol and
16611 set comdat_type_p here.
16612 (output_die): Skip DIE symbol output for the LTO added one.
16613 Handle DIE symbol references with offset.
16614 (output_comp_unit): Guard section name mangling properly.
16615 For LTO debug sections emit a symbol at the section beginning
16616 which we use to refer to its DIEs.
16617 (add_abstract_origin_attribute): For DIEs registered via
16618 dwarf2out_register_external_die directly refer to the early
16619 DIE rather than indirectly through the shadow one we created.
16620 Remove obsolete call to dwarf2out_abstract_function for
16621 non-function/block origins.
16622 (gen_array_type_die): When generating early LTO debug do
16623 not emit DW_AT_string_length.
16624 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
16625 late when in LTO. As suggested place a gcc_unreachable for
16626 the DECL_ABSTRACT_P case.
16627 (gen_subprogram_die): Avoid another specification DIE
16628 for early built declarations/definitions for the late LTO case.
16629 (gen_variable_die): Add type references for late duplicated VLA dies
16631 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
16632 we have the abstract instance already.
16633 (process_scope_var): Adjust decl DIE contexts in LTO which
16634 first puts them in limbo.
16635 (gen_decl_die): Do not generate type DIEs late apart from
16636 types for VLAs or for decls we do not yet have a DIE. Do not
16637 call dwarf2out_abstract_function late.
16638 (dwarf2out_early_global_decl): Make sure to create DIEs
16639 for abstract instances of a decl first.
16640 (dwarf2out_late_global_decl): Adjust comment.
16641 (output_macinfo_op): With multiple macro sections use
16642 macinfo_label_base to distinguish labels.
16643 (output_macinfo): Likewise. Update macinfo_label_base.
16644 Pass in the line info label.
16645 (note_variable_value_in_expr): When generating LTO resolve
16646 all variable values here by generating DIEs as needed.
16647 (init_sections_and_labels): Add early LTO debug flag parameter
16648 and generate different sections and names if set. Add generation
16649 counter for the labels so we can have multiple of them.
16650 (reset_dies): Helper to allow DIEs to be output multiple times.
16651 (dwarf2out_finish): When outputting DIEs to the fat part of an
16652 LTO object first reset DIEs.
16653 (dwarf2out_early_finish): Output early DIEs when generating LTO.
16654 (modified_type_die): Check for decl_ultimate_origin being self
16656 (gen_type_die_with_usage): Likewise.
16657 (gen_typedef_die): Allow decl_ultimate_origin being self.
16658 (set_decl_abstract_flags): Remove.
16659 (set_block_abstract_flags): Likewise.
16660 (dwarf2out_abstract_function): Treat the early generated DIEs
16661 as the abstract copy and only add DW_AT_inline and
16662 DW_AT_artificial here and call set_decl_origin_self.
16663 If the DIE has an abstract origin don't do anything.
16664 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
16665 if we have none yet (Go fails to build one, PR78628).
16666 (variably_modified_type_p): Prevent endless recursion for Ada
16667 cyclic pointer types.
16668 * lto-streamer-in.c: Include debug.h.
16669 (dref_queue): New global.
16670 (lto_read_tree_1): Stream in DIE references.
16671 (lto_input_tree): Register DIE references.
16672 (input_function): Stream DECL_DEBUG_ARGS.
16673 * lto-streamer-out.c: Include debug.h.
16674 (lto_write_tree_1): Output DIE references.
16675 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
16676 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
16677 (output_function): Stream DECL_DEBUG_ARGS.
16678 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
16679 Stream DECL_ABSTRACT_ORIGIN.
16680 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
16681 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
16682 DECL_CONTEXT for file-scope decls.
16683 * lto-streamer.h (struct dref_entry): Declare.
16684 (dref_queue): Likewise.
16685 * cfgexpand.c (pass_expand::execute): Do not call the
16686 outlining_inline_function hook here.
16687 * lto-wrapper.c (debug_obj): New global.
16688 (tool_cleanup): Unlink it if required.
16689 (debug_objcopy): New function.
16690 (run_gcc): Handle early debug sections in the IL files by
16691 extracting them to separate files, partially linkin them and
16692 feeding the result back as result to the linker.
16693 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
16694 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
16695 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
16696 sections into a separate segment.
16697 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
16699 (darwin_asm_dwarf_section): Likewise.
16700 (darwin_asm_output_dwarf_offset): Likewise.
16701 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
16703 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16704 Alan Hayward <alan.hayward@arm.com>
16705 David Sherwood <david.sherwood@arm.com>
16707 * read-md.h (md_reader::record_potential_iterator_use): Replace
16708 pointer argument with an rtx and an index.
16709 * read-rtl.c (iterator_group::apply_iterator): Likewise.
16710 (apply_mode_iterator): Likewise.
16711 (apply_code_iterator): Likewise.
16712 (apply_int_iterator): Likewise.
16713 (apply_subst_iterator): Likewise.
16714 (record_iterator_use): Likewise.
16715 (record_attribute_use): Likewise.
16716 (md_reader::record_potential_iterator_use): Likewise. Update calls
16717 to record_iterator_use and apply_iterator.
16718 (iterator_use): Replace ptr with x and index.
16719 (attribute_use): Likewise.
16720 (apply_attribute_uses): Update calls to apply_iterator.
16721 (apply_iterators): Likewise. Update initialization of iterator_use.
16722 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
16723 and record_potential_iterator_use.
16724 (rtx_reader::read_rtx_operand): Likewise.
16726 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16727 Alan Hayward <alan.hayward@arm.com>
16728 David Sherwood <david.sherwood@arm.com>
16730 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
16733 2017-08-21 Richard Biener <rguenther@suse.de>
16735 PR middle-end/81884
16736 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
16737 at struct end conservatively when comparing common bases.
16739 2017-08-21 Richard Biener <rguenther@suse.de>
16741 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
16742 (mem_ref_in_stmt): Remove.
16743 (determine_max_movement): Use ref index to get at the reference.
16744 (invariantness_dom_walker::before_dom_children): Deal with
16745 lim data already initialized.
16746 (gather_mem_refs_stmt): Initialize lim data and record ref index.
16748 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16750 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
16751 (TARGET_ISA_ROUND): Ditto.
16752 (TARGET_ROUND): Ditto.
16753 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
16754 * config/i386/i386.md: Ditto.
16755 * config/i386/sse.md: Ditto.
16756 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
16757 with OPTION_MASK_ISA_SSE4_1.
16759 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16762 * doc/extend.texi (x86 Built-in Functions): Correct the name of
16763 __builtin_ia32_lzcnt_u16.
16765 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
16768 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
16769 (rs6000_set_current_function): Rewrite function to use it.
16771 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
16774 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
16775 and TYPE_WARN_IF_NOT_ALIGN.
16776 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
16777 (handle_warn_if_not_align): New.
16778 (place_union_field): Call handle_warn_if_not_align.
16779 (place_field): Call handle_warn_if_not_align.
16780 Copy TYPE_WARN_IF_NOT_ALIGN.
16781 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
16782 (layout_type): Likewise.
16783 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
16785 (tree_decl_common): Add warn_if_not_align.
16786 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
16787 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
16788 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
16789 (DECL_WARN_IF_NOT_ALIGN): Likewise.
16790 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
16791 * doc/extend.texi: Document warn_if_not_aligned attribute.
16792 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
16794 2017-08-17 Martin Liska <mliska@suse.cz>
16797 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
16798 (get_data_dependence): Use it as pointer type.
16799 (distribute_loop): Likewise.
16801 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16803 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
16804 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
16805 (p8_vmrgow_<mode>_direct): New define_insn.
16806 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
16807 handle endianness for vmrgew and vmrgow permute patterns.
16809 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16811 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
16812 * config/rs6000/rs6000-cpus.def: Remove comment.
16813 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
16814 (POWERPC_MASKS): Likewise.
16815 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
16816 use of TARGET_VSX_TIMODE.
16817 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
16818 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
16819 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
16820 (rs6000_option_override_internal): Remove dead code.
16821 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
16822 (rs6000_legitimize_reload_address): Likewise.
16823 (rs6000_legitimate_address_p): Likewise.
16824 (rs6000_opt_masks): Delete "vsx-timode".
16825 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
16826 from function comment.
16827 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
16828 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
16829 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
16831 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
16832 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
16833 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
16834 (VSX_TI): Likewise.
16836 (define_peephole2): Likewise.
16838 2017-08-17 Martin Sebor <msebor@redhat.com>
16841 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
16842 past the end of an array.
16843 (test_pp_format): Add test cases.
16845 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
16847 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
16848 missing ECF_NOTHROW flags.
16850 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16853 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
16854 operands residing in integer registers.
16855 (*vsx_le_perm_load_<mode>): Likewise.
16856 (*vsx_le_perm_store_<mode>): Likewise.
16857 (define_peephole2): Add peepholes to optimize the above.
16859 2017-08-17 Marek Polacek <polacek@redhat.com>
16861 PR middle-end/81814
16862 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
16863 to mimic what shorten_compare did. Change the return type to bool.
16864 (fold_cond_expr_with_comparison): Update call to
16865 operand_equal_for_comparison_p.
16866 (fold_ternary_loc): Likewise.
16868 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
16870 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
16872 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
16874 2017-08-17 Richard Biener <rguenther@suse.de>
16876 * tree-ssa-structalias.c (solve_graph): When propagating
16877 to successors update the graphs succ edges and avoid duplicate work.
16879 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
16882 * config/i386/i386.c (ix86_option_override_internal): Save target
16883 specific options after ix86_stack_protector_guard_reg was changed.
16885 2017-08-17 Richard Biener <rguenther@suse.de>
16887 PR tree-optimization/81827
16888 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
16889 (new_var_info): Initialize it conservatively.
16890 (get_call_vi): Mark register vars.
16891 (new_scalar_tmp_constraint_exp): Likewise.
16892 (handle_rhs_call): Likewise.
16893 (handle_const_call): Likewise.
16894 (create_function_info_for): Likewise.
16895 (solve_constraints): Sort varinfos to separate register from
16896 non-register vars to pack points-to solution bitmaps during
16899 2017-08-17 Marek Polacek <polacek@redhat.com>
16901 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
16903 2017-08-17 Richard Biener <rguenther@suse.de>
16905 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
16906 to true when overflow is undefined and we saturated the result.
16908 2017-08-17 Alan Modra <amodra@gmail.com>
16911 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
16912 Don't use store multiple if only one reg needs saving.
16913 (interesting_frame_related_regno): New function.
16914 (rs6000_frame_related): Don't emit frame info for regs that
16916 (rs6000_emit_epilogue): Likewise.
16918 2017-08-16 Nathan Sidwell <nathan@acm.org>
16920 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
16921 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
16922 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
16923 * tree.c (free_lang_data_in_type): Use else-if chain. Always
16924 clear TYPE_LANG_1. Remove obsolete member-function stripping.
16925 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
16926 (verify_type): Adjust for TYPE_BINFO move.
16927 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
16928 process TYPE_BINFO directly.
16929 (hash_tree): Likewise.
16930 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16932 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16935 2017-08-16 David Malcolm <dmalcolm@redhat.com>
16937 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
16939 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
16942 * config/i386/i386.md (*anddi_1_btr): Change predicates of
16943 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
16944 Add ix86_binary_operator_ok to insn constraint.
16945 (*iordi_1_bts): Ditto.
16946 (*xordi_1_btc): Ditto.
16947 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
16948 Update corresponding peephole2 pattern.
16952 2017-08-16 Bin Cheng <bin.cheng@arm.com>
16954 PR tree-optimization/81832
16955 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
16956 copy loop header which has IFN_LOOP_DIST_ALIAS call.
16958 2017-08-16 Marek Polacek <polacek@redhat.com>
16961 * fold-const.c (fold_indirect_ref_1): Restore original behavior
16962 regarding size_zero_node.
16964 2017-08-16 Martin Liska <mliska@suse.cz>
16967 * config.gcc: Respect previously set extra_objs in case
16970 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
16972 PR tree-optimization/81835
16973 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
16974 the phi SSA_NAME. Check that the condition in a COND_EXPR does
16975 not depend on the phi.
16977 2017-08-16 Alan Modra <amodra@gmail.com>
16979 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
16982 2017-08-16 Alan Modra <amodra@gmail.com>
16984 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
16985 (save_reg_p): ..into this. Update all callers.
16986 (first_reg_to_save): Simplify.
16988 2017-08-16 Alan Modra <amodra@gmail.com>
16990 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
16993 2017-08-15 Joseph Myers <joseph@codesourcery.com>
16997 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
16998 constant count if that count is less than 32.
17000 2017-08-15 Nathan Sidwell <nathan@acm.org>
17002 * gcc.c (execute): Emit friendlier message if inferior is killed
17003 by an external cause.
17005 2017-08-15 Richard Biener <rguenther@suse.de>
17007 PR tree-optimization/81790
17008 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17009 CONSTRUCTORs from simplifying and VN.
17011 2017-08-14 Martin Sebor <msebor@redhat.com>
17013 * builtin-attrs.def: Add comments.
17015 2017-08-14 Martin Sebor <msebor@redhat.com>
17018 * doc/extend.texi (attribute nonstring): Document new attribute.
17020 2017-08-14 Martin Sebor <msebor@redhat.com>
17023 * tree-diagnostic.c (default_tree_printer): Handle %G.
17024 * gimple-pretty-print.h (percent_G_format): Declare new function.
17025 * gimple-pretty-print.c (percent_G_format): Define.
17026 * tree-pretty-print.c (percent_K_format): Add argument.
17028 2017-08-14 Martin Sebor <msebor@redhat.com>
17030 PR translation/79998
17031 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17032 Remove a stray space.
17034 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17037 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17038 (*iordi_1_bts): Ditto.
17039 (*xordi_1_btc): Ditto.
17041 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17044 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17045 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17047 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17048 quoted strings, and make more translator-friendly.
17049 (darwin_rs6000_override_options): Likewise.
17050 (rs6000_option_override_internal): Likewise.
17051 (rs6000_return_in_memory): Fix overlong line.
17052 (init_cmulative_args): Use quoted strings, and make more
17053 translator-friendly.
17054 (rs6000_pass_by_reference): Fix overlong line.
17055 (def_builtin): Use quoted strings.
17056 (altivec_expand_predicate_builtin): Use quoted strings, and make
17057 more translator-friendly.
17058 (htm_expand_builtin): Use quoted strings.
17059 (cpu_expand_builtin): Use quoted strings, and make more
17060 translator-friendly.
17061 (altivec_expand_builtin): Likewise.
17062 (paired_expand_predicate_builtin): Likewise.
17063 (rs6000_invalid_builtin): Likewise.
17064 (builtin_function_type): Use quoted strings.
17065 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17066 more translator-friendly.
17067 (rs6000_trampoline_init): Likewise.
17068 (rs6000_handle_altivec_attribute): Likewise.
17069 (rs6000_inner_target_options): Use quoted strings.
17070 (rs6000_disable_incompatible_switches): Likewise.
17071 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17072 strings, and make more translator-friendly.
17073 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17075 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17077 PR tree-optimization/81799
17078 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17079 cond_expr to simple gimple operand.
17081 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17083 PR middle-end/46932
17084 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17086 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17090 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17091 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17092 TARGET_GASISR_PROLOGUES.
17093 * config/avr/avr.c (avr_option_override): Same.
17094 (avr_pass_pre_proep::execute): Same.
17096 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17099 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17100 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17102 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17104 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17105 define_insn_and_split. Split to a memory load from 0 in
17106 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17107 using PTR mode iterator.
17108 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17109 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17110 (*add_tp_<mode>): Redefine as define_insn_and_split.
17111 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17112 address space. Merge with *add_tp_x32 using PTR mode iterator.
17113 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17114 Split to an add with a memory load from 0 in
17115 DEFAULT_TLS_SEG_REG address space.
17117 2017-08-12 Andrew Pinski <apinski@cavium.com>
17119 * config/aarch64/aarch64-option-extensions.def (rdma):
17120 Fix feature string to what Linux prints out in /proc/cpuinfo.
17122 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17125 * dwarf2out.c (modified_type_die): For C typedef types that have
17126 an ultimate origin, process the ultimate origin instead of the
17128 (gen_typedef_die): Assert that input DECLs have no ultimate
17130 (gen_type_die_with_usage): For typedef variants that have an
17131 ultimate origin, just call gen_decl_die on the original DECL.
17132 (process_scope_var): Avoid creating DIEs for local typedefs and
17133 concrete static variables.
17135 2017-08-12 Alan Modra <amodra@gmail.com>
17139 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17140 match gnu-user.h startfile.
17141 (ENDFILE_LINUX_SPEC): Similarly.
17143 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17146 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17148 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17150 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17151 * config/aarch64/aarch64.md (mov<mode>): Change.
17152 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17153 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17154 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17156 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17158 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17159 for storage order barriers.
17161 2017-08-11 Martin Liska <mliska@suse.cz>
17164 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17165 variables of void type.
17167 2017-08-11 Martin Liska <mliska@suse.cz>
17169 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17170 TARGET_SUPPORTS_ALIASES.
17171 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17172 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17173 (optimize_weakref): Likewise.
17174 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17175 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17176 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17178 2017-08-11 Martin Liska <mliska@suse.cz>
17181 * config/i386/i386.c (make_resolver_func): Do complete
17182 refactoring of the function.
17184 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17187 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17188 * config/i386/i386.c (ix86_stack_protect_guard): Use
17189 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17190 * doc/invoke.texi (x86 Options): Document
17191 -mstack-protector-guard-symbol= option.
17193 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17195 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17196 * config/i386/i386.c (ix86_split_stack_guard): New function.
17197 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17198 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17199 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17200 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17201 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17202 (split_stack_space_check): Call ix86_split_stack_guard.
17204 2017-08-10 Martin Sebor <msebor@redhat.com>
17206 * print-tree.c (print_node): Print location using the established
17208 Replace spaces with colons.
17209 (debug_raw, debug): Ditto.
17211 2017-08-10 Martin Sebor <msebor@redhat.com>
17214 * pretty-print.c (pp_format): Correct the handling of %s precision.
17216 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
17219 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17221 (ix86_finalize_stack_frame_flags): This. Also clear
17222 frame_pointer_needed if -fno-omit-frame-pointer is used without
17224 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17225 with ix86_finalize_stack_frame_flags.
17226 (ix86_expand_epilogue): Likewise.
17227 (ix86_expand_split_stack_prologue): Likewise.
17228 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
17230 2017-08-10 Martin Liska <mliska@suse.cz>
17233 * c-attribs.c (handle_target_attribute):
17234 Report warning for an empty string argument of target attribute.
17236 2017-08-09 Jakub Jelinek <jakub@redhat.com>
17239 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
17241 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
17242 or DECL_NONLOCAL labels.
17243 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
17244 or DECL_NONLOCAL labels here.
17246 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
17248 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
17249 to indicate when early gimple folding has been disabled.
17250 (rs6000_gimple_fold_builtin): Add debug content.
17251 (rs6000_invalid_builtin): Fix whitespace.
17252 (rs6000_expand_builtin): Fix whitespace.
17253 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
17255 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
17258 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
17259 SAVE_MULTIPLE if not all the registers that saves, should be saved.
17261 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
17263 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
17264 (qdf24xx): Likewise.
17265 * config/aarch64/aarch64.md: Include falkor.md.
17266 * config/aarch64/falkor.md: New.
17268 2017-08-09 Marek Polacek <polacek@redhat.com>
17271 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
17272 * diagnostic.c (emit_diagnostic): Add a comment.
17273 (emit_diagnostic_valist): New function.
17275 2017-08-09 Marek Polacek <polacek@redhat.com>
17278 * input.c (make_location): New overload.
17279 * input.h (make_location): Declare.
17281 2017-08-08 Alan Modra <amodra@gmail.com>
17282 H.J. Lu <hongjiu.lu@intel.com>
17285 * gcc.c (NO_PIE_SPEC): Delete.
17286 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
17288 (LINK_PIE_SPEC): ..to here.
17289 (LINK_COMMAND_SPEC): Support -no-pie.
17290 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
17291 chain of crtbegin*.o selection, update for PIE_SPEC changes and
17293 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
17294 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
17295 (ENDFILE_CRTEND_SPEC): Similarly.
17297 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
17300 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
17301 (mstack-protector-guard-offset=): Ditto.
17302 * config/i386/i386.c (ix86_option_override): Handle
17303 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
17305 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
17306 ix86_stack_protect_guard_offset variables.
17307 (TARGET_STACK_PROTECT_GUARD): Always define.
17308 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
17309 and -mstack-protector-guard-offset= options.
17311 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17313 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
17314 boundary case for the last candidate.
17316 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17318 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
17319 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
17321 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17323 PR middle-end/19706
17324 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
17325 * config/aarch64/aarch64-builtins.c
17326 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
17327 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
17328 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
17330 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17331 Andrew Pinski <pinskia@gmail.com>
17333 PR middle-end/19706
17334 * internal-fn.def (XORSIGN): New.
17335 * optabs.def (xorsign_optab): New.
17336 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
17337 (convert_expand_mult_copysign): New.
17338 (pass_optimize_widening_mul::execute): Call
17339 convert_expand_mult_copysign.
17341 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17343 PR tree-optimization/81354
17344 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
17345 Insert on edges rather than explicitly creating landing pads.
17346 (analyze_candidates_and_replace): Commit edge inserts.
17348 2017-08-08 Richard Biener <rguenther@suse.de>
17350 PR middle-end/81719
17351 * tree-ssa-loop-niter.c: Include tree-dfa.h.
17352 (expand_simple_operations): Also look through ADDR_EXPRs with
17353 MEM_REF bases treating them as POINTER_PLUS_EXPR.
17355 2017-08-08 Richard Biener <rguenther@suse.de>
17357 PR tree-optimization/81723
17358 * tree-vect-slp.c (struct bst_traits): New hash traits.
17359 (bst_fail): New global.
17360 (vect_build_slp_tree_2): New worker, split out from ...
17361 (vect_build_slp_tree): ... this now wrapping it with using
17362 bst_fail set to cache SLP tree build fails. Properly handle
17364 (vect_analyze_slp_instance): Allocate and free bst_fail.
17366 2017-08-08 Martin Liska <mliska@suse.cz>
17369 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
17370 LABEL_DECLs that can be from a different function.
17372 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17374 PR tree-optimization/81744
17375 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
17376 loop's number of iterations.
17378 2017-08-08 Martin Liska <mliska@suse.cz>
17380 * asan.c: Include header files.
17381 * attribs.c (build_decl_attribute_variant): New function moved
17383 (build_type_attribute_qual_variant): Likewise.
17384 (cmp_attrib_identifiers): Likewise.
17385 (simple_cst_list_equal): Likewise.
17386 (omp_declare_simd_clauses_equal): Likewise.
17387 (attribute_value_equal): Likewise.
17388 (comp_type_attributes): Likewise.
17389 (build_type_attribute_variant): Likewise.
17390 (lookup_ident_attribute): Likewise.
17391 (remove_attribute): Likewise.
17392 (merge_attributes): Likewise.
17393 (merge_type_attributes): Likewise.
17394 (merge_decl_attributes): Likewise.
17395 (merge_dllimport_decl_attributes): Likewise.
17396 (handle_dll_attribute): Likewise.
17397 (attribute_list_equal): Likewise.
17398 (attribute_list_contained): Likewise.
17399 * attribs.h (lookup_attribute): New function moved from tree.[ch].
17400 (lookup_attribute_by_prefix): Likewise.
17401 * bb-reorder.c: Include header files.
17402 * builtins.c: Likewise.
17403 * calls.c: Likewise.
17404 * cfgexpand.c: Likewise.
17405 * cgraph.c: Likewise.
17406 * cgraphunit.c: Likewise.
17407 * convert.c: Likewise.
17408 * dwarf2out.c: Likewise.
17409 * final.c: Likewise.
17410 * fold-const.c: Likewise.
17411 * function.c: Likewise.
17412 * gimple-expr.c: Likewise.
17413 * gimple-fold.c: Likewise.
17414 * gimple-pretty-print.c: Likewise.
17415 * gimple.c: Likewise.
17416 * gimplify.c: Likewise.
17417 * hsa-common.c: Likewise.
17418 * hsa-gen.c: Likewise.
17419 * internal-fn.c: Likewise.
17420 * ipa-chkp.c: Likewise.
17421 * ipa-cp.c: Likewise.
17422 * ipa-devirt.c: Likewise.
17423 * ipa-fnsummary.c: Likewise.
17424 * ipa-inline.c: Likewise.
17425 * ipa-visibility.c: Likewise.
17427 * lto-cgraph.c: Likewise.
17428 * omp-expand.c: Likewise.
17429 * omp-general.c: Likewise.
17430 * omp-low.c: Likewise.
17431 * omp-offload.c: Likewise.
17432 * omp-simd-clone.c: Likewise.
17433 * opts-global.c: Likewise.
17434 * passes.c: Likewise.
17435 * predict.c: Likewise.
17436 * sancov.c: Likewise.
17437 * sanopt.c: Likewise.
17438 * symtab.c: Likewise.
17439 * toplev.c: Likewise.
17440 * trans-mem.c: Likewise.
17441 * tree-chkp.c: Likewise.
17442 * tree-eh.c: Likewise.
17443 * tree-into-ssa.c: Likewise.
17444 * tree-object-size.c: Likewise.
17445 * tree-parloops.c: Likewise.
17446 * tree-profile.c: Likewise.
17447 * tree-ssa-ccp.c: Likewise.
17448 * tree-ssa-live.c: Likewise.
17449 * tree-ssa-loop.c: Likewise.
17450 * tree-ssa-sccvn.c: Likewise.
17451 * tree-ssa-structalias.c: Likewise.
17452 * tree-ssa.c: Likewise.
17453 * tree-streamer-in.c: Likewise.
17454 * tree-vectorizer.c: Likewise.
17455 * tree-vrp.c: Likewise.
17456 * tsan.c: Likewise.
17457 * ubsan.c: Likewise.
17458 * varasm.c: Likewise.
17459 * varpool.c: Likewise.
17460 * tree.c: Remove functions moved to attribs.[ch].
17461 * tree.h: Likewise.
17462 * config/aarch64/aarch64.c: Add attrs.h header file.
17463 * config/alpha/alpha.c: Likewise.
17464 * config/arc/arc.c: Likewise.
17465 * config/arm/arm.c: Likewise.
17466 * config/avr/avr.c: Likewise.
17467 * config/bfin/bfin.c: Likewise.
17468 * config/c6x/c6x.c: Likewise.
17469 * config/cr16/cr16.c: Likewise.
17470 * config/cris/cris.c: Likewise.
17471 * config/darwin.c: Likewise.
17472 * config/epiphany/epiphany.c: Likewise.
17473 * config/fr30/fr30.c: Likewise.
17474 * config/frv/frv.c: Likewise.
17475 * config/ft32/ft32.c: Likewise.
17476 * config/h8300/h8300.c: Likewise.
17477 * config/i386/winnt.c: Likewise.
17478 * config/ia64/ia64.c: Likewise.
17479 * config/iq2000/iq2000.c: Likewise.
17480 * config/lm32/lm32.c: Likewise.
17481 * config/m32c/m32c.c: Likewise.
17482 * config/m32r/m32r.c: Likewise.
17483 * config/m68k/m68k.c: Likewise.
17484 * config/mcore/mcore.c: Likewise.
17485 * config/microblaze/microblaze.c: Likewise.
17486 * config/mips/mips.c: Likewise.
17487 * config/mmix/mmix.c: Likewise.
17488 * config/mn10300/mn10300.c: Likewise.
17489 * config/moxie/moxie.c: Likewise.
17490 * config/msp430/msp430.c: Likewise.
17491 * config/nds32/nds32-isr.c: Likewise.
17492 * config/nds32/nds32.c: Likewise.
17493 * config/nios2/nios2.c: Likewise.
17494 * config/nvptx/nvptx.c: Likewise.
17495 * config/pa/pa.c: Likewise.
17496 * config/pdp11/pdp11.c: Likewise.
17497 * config/powerpcspe/powerpcspe.c: Likewise.
17498 * config/riscv/riscv.c: Likewise.
17499 * config/rl78/rl78.c: Likewise.
17500 * config/rx/rx.c: Likewise.
17501 * config/s390/s390.c: Likewise.
17502 * config/sh/sh.c: Likewise.
17503 * config/sol2.c: Likewise.
17504 * config/sparc/sparc.c: Likewise.
17505 * config/spu/spu.c: Likewise.
17506 * config/stormy16/stormy16.c: Likewise.
17507 * config/tilegx/tilegx.c: Likewise.
17508 * config/tilepro/tilepro.c: Likewise.
17509 * config/v850/v850.c: Likewise.
17510 * config/vax/vax.c: Likewise.
17511 * config/visium/visium.c: Likewise.
17512 * config/xtensa/xtensa.c: Likewise.
17514 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17517 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
17518 constraints since the -mupper-regs-* switches have been
17520 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
17521 into a vector from a double word element that was extracted from
17522 another vector, and eliminate extra XXPERMDI instructions.
17523 (vsx_concat_<mode>_2): Likewise.
17524 (vsx_concat_<mode>_3): Likewise.
17525 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
17526 concat to allow optimizing inserts from previous extracts.
17528 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
17530 * config/i386/i386.c (ix86_stack_protect_guard): Generate
17531 memory reference to a SSP offset in TLS address space.
17532 (ix86_print_operand) <case '@'>: Remove.
17533 (ix86_print_operand_punct_valid_p): Remove '@' code.
17534 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
17535 UNSPEC_SP_TLS_TEST.
17536 (stack_tls_protect_set_<mode>): Remove.
17537 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
17538 (stack_tls_protect_test_<mode>): Remove.
17539 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
17541 2017-08-07 Olivier Hainque <hainque@adacore.com>
17544 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
17546 2017-08-07 Douglas Rupp <rupp@adacore.com>
17548 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
17549 variable was referenced as multidir in command.
17551 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17554 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
17557 2017-08-07 Martin Liska <mliska@suse.cz>
17559 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
17560 * config/rl78/rl78.c: Add include of attribs.h.
17561 * config/sh/sh.c: Likewise.
17562 * config/v850/v850.c: Likewise.
17564 2017-08-07 Tom de Vries <tom@codesourcery.com>
17566 PR middle-end/78266
17567 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
17569 2017-08-07 Martin Liska <mliska@suse.cz>
17571 * config/mips/mips.c: Include attribs.h.
17573 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
17576 * doc/invoke.texi: Document change in behvaior for -Ofast for
17579 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
17581 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
17583 (aarch64_pop_regs): Likewise.
17584 (aarch64_gen_load_pair): Likewise.
17585 (aarch64_save_callee_saves): Likewise.
17586 (aarch64_restore_callee_saves): Likewise.
17588 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17590 * config/i386/i386.c: Revert the last change.
17592 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17595 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17597 (ix86_finalize_stack_frame_flags): This. Also clear
17598 frame_pointer_needed if -fno-omit-frame-pointer is used without
17600 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17601 with ix86_finalize_stack_frame_flags.
17602 (ix86_expand_epilogue): Likewise.
17603 (ix86_expand_split_stack_prologue): Likewise.
17605 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17608 * config/i386/i386.c (get_builtin_code_for_version): Set priority
17609 to P_AES for Westmere.
17611 2017-08-07 Jonathan Yong <10walls@gmail.com>
17613 * config/i386/mingw.opt (fset-stack-executable): Removed.
17614 * config/i386/cygming.opt (fset-stack-executable): Moved
17616 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
17618 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
17620 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
17622 2017-08-07 Marek Polacek <polacek@redhat.com>
17624 PR middle-end/81737
17625 * fold-const.c (fold_indirect_ref_1): Check type_domain.
17627 2017-08-07 Martin Liska <mliska@suse.cz>
17629 * attribs.h (canonicalize_attr_name): New function.
17630 (cmp_attribs): Move from c-format.c and adjusted.
17631 (is_attribute_p): Moved from tree.h.
17632 * tree-inline.c: Add new includes.
17633 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
17634 (private_is_attribute_p): Remove.
17635 (private_lookup_attribute): Likewise.
17636 (private_lookup_attribute_by_prefix): Simplify.
17637 (remove_attribute): Use is_attribute_p.
17638 * tree.h: Remove removed declarations.
17640 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17642 PR middle-end/81698
17643 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
17644 instead of computing it in the function. Formatting fix.
17645 (expand_case): Don't rely on default_edge being the first edge,
17646 clear it if removing it, pass default_edge to
17647 emit_case_dispatch_table.
17648 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
17651 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
17653 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
17654 insn in the function, emit NOP after the insn.
17656 2017-08-06 Tom de Vries <tom@codesourcery.com>
17658 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
17661 2017-08-06 Tom de Vries <tom@codesourcery.com>
17663 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
17666 2017-08-04 Yury Gribov <tetra2005@gmail.com>
17668 PR tree-optimization/57371
17669 * match.pd: New pattern.
17671 2017-08-04 Marek Polacek <polacek@redhat.com>
17673 PR middle-end/81695
17674 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
17675 perform the computation in offset_int.
17677 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17679 PR tree-optimization/81136
17680 * tree-vectorizer.h: Include tree-hash-traits.h.
17681 (vec_base_alignments): New typedef.
17682 (vec_info): Add a base_alignments field.
17683 (vect_record_base_alignments): Declare.
17684 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
17686 (DR_IS_CONDITIONAL_IN_STMT): New macro.
17687 (create_data_ref): Add an is_conditional_in_stmt argument.
17688 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
17689 the is_conditional_in_stmt field.
17690 (data_ref_loc): Add an is_conditional_in_stmt field.
17691 (get_references_in_stmt): Set the is_conditional_in_stmt field.
17692 (find_data_references_in_stmt): Update call to create_data_ref.
17693 (graphite_find_data_references_in_stmt): Likewise.
17694 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
17695 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
17696 (vect_record_base_alignment): New function.
17697 (vect_record_base_alignments): Likewise.
17698 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
17699 for nested statements even if we fail to compute a misalignment.
17700 Use pooled base alignments for unconditional references.
17701 (vect_find_same_alignment_drs): Compare base addresses instead
17703 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
17704 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
17706 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17708 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
17709 Add an explicit name for the enum. Use auto_vec for slp_instances
17710 and grouped_stores.
17711 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
17713 (_bb_vec_info): Add a constructor and destructor.
17714 (vinfo_for_stmt): Return NULL for uids of -1 as well.
17715 (destroy_loop_vec_info): Delete.
17716 (vect_destroy_datarefs): Likewise.
17717 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
17718 (vec_info::vec_info): New function.
17719 (vec_info::~vec_info): Likewise.
17720 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
17721 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
17722 destroy_loop_vec_info.
17723 * tree-vect-loop.c (new_loop_vec_info): Replace with...
17724 (_loop_vec_info::_loop_vec_info): ...this.
17725 (destroy_loop_vec_info): Replace with...
17726 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
17727 the stmt_vec_infos. Leave handling of vec_info information to its
17728 destructor. Remove explicit vector releases.
17729 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
17730 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
17731 * tree-vect-slp.c (new_bb_vec_info): Replace with...
17732 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
17733 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
17734 (destroy_bb_vec_info): Replace with...
17735 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
17736 information to its destructor.
17737 (vect_slp_analyze_bb_1): Use new and delete instead of
17738 new_bb_vec_info and destroy_bb_vec_info.
17739 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
17742 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17744 * tree-data-ref.h (subscript): Add access_fn field.
17745 (data_dependence_relation): Add could_be_independent_p.
17746 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
17747 (same_access_functions): Move to tree-data-ref.c.
17748 * tree-data-ref.c (ref_contains_union_access_p): New function.
17749 (access_fn_component_p): Likewise.
17750 (access_fn_components_comparable_p): Likewise.
17751 (dr_analyze_indices): Add a reference to access_fn_component_p.
17752 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
17754 (constant_access_functions): Likewise.
17755 (add_other_self_distances): Likewise.
17756 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
17757 (initialize_data_dependence_relation): Use XCNEW and remove
17758 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
17759 of access functions that have the same type. Allow the
17760 subsequence to end with different bases in some circumstances.
17761 Record the chosen access functions in SUB_ACCESS_FN.
17762 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
17763 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
17764 (subscript_dependence_tester_1): Likewise dra and drb.
17765 (build_classic_dist_vector): Update calls accordingly.
17766 (subscript_dependence_tester): Likewise.
17767 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
17768 DDR_COULD_BE_INDEPENDENT_P.
17769 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
17770 comp_alias_ddrs instead of may_alias_ddrs.
17771 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17773 (vect_analyze_data_ref_dependence): Use it if
17774 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
17775 distance vectors if that fails.
17776 (dependence_distance_ge_vf): New function.
17777 (vect_prune_runtime_alias_test_list): Use it. Don't clear
17778 LOOP_VINFO_MAY_ALIAS_DDRS.
17780 2017-08-04 Richard Biener <rguenther@suse.de>
17782 PR middle-end/81705
17783 * fold-const.c (fold_binary_loc): Properly restrict
17784 minus_var0 && minus_var1 case when associating undefined overflow
17787 2017-08-04 Tom de Vries <tom@codesourcery.com>
17789 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
17791 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17793 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17794 Don't start diagnostic messages with a capital letter.
17795 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17797 (rs6000_invalid_builtin): Likewise.
17798 (rs6000_trampoline_init): Likewise.
17800 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17803 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
17804 after setting changeable df flags.
17806 2017-08-03 Richard Biener <rguenther@suse.de>
17808 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
17809 up if the use is in USE - X.
17811 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
17813 * toplev.c (dumpfile.h): New include.
17814 (internal_error_reentered): New static function. Use it...
17815 (internal_error_function): ...here to handle reentered internal_error.
17817 2017-08-03 Richard Biener <rguenther@suse.de>
17819 PR middle-end/81148
17820 * fold-const.c (split_tree): Add minus_var and minus_con
17821 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
17822 here but always use minus_*.
17823 (associate_trees): Assert we never associate with MINUS_EXPR
17824 and NULL first operand. Do not recurse for PLUS_EXPR operands
17825 when associating as MINUS_EXPR either.
17826 (fold_binary_loc): Track minus_var and minus_con.
17828 2017-08-03 Tom de Vries <tom@codesourcery.com>
17831 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
17832 ACCEL_COMPILER, apply finish_options on
17833 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
17835 2017-08-03 Tom de Vries <tom@codesourcery.com>
17838 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
17839 function_entry_patch_area_size > 0.
17841 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17844 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
17845 instead of 10??LU, perform unit multiplication in wide_int,
17846 don't change alloc_object_size_limit if the limit is larger
17849 PR tree-optimization/81655
17850 PR tree-optimization/81588
17851 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
17852 the case when ranges[i].low and high are 1 for unsigned type with
17855 PR middle-end/81052
17856 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
17857 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
17859 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17861 * tree-vrp.h: Add include guard.
17863 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
17866 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
17867 (ud2): New insn pattern.
17868 * config/i386/i386.c (ix86_expand_epilogue):
17869 For naked functions, generate ud2 instead of trap insn.
17871 2017-08-02 Marek Polacek <polacek@redhat.com>
17874 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
17876 2017-08-02 Tom de Vries <tom@codesourcery.com>
17877 Cesar Philippidis <cesar@codesourcery.com>
17879 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
17880 Add missing edge probabilities.
17882 2017-08-02 Tamar Christina <tamar.christina@arm.com>
17884 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
17885 Correct endianness.
17887 2017-08-02 Jakub Jelinek <jakub@redhat.com>
17889 PR middle-end/79499
17890 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
17891 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
17892 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
17893 prologue_seq sequences - if any.
17895 2017-08-02 Richard Biener <rguenther@suse.de>
17897 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
17898 via vectors if supported, integer extracts via punning if supported
17899 or otherwise vector extracts.
17901 2017-08-02 Richard Biener <rguenther@suse.de>
17903 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
17905 (bitmap_insert_into_set): ... this.
17907 2017-08-02 Richard Biener <rguenther@suse.de>
17909 PR tree-optimization/81633
17911 2015-08-17 Alan Hayward <alan.hayward@arm.com>
17913 PR tree-optimization/71752
17914 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
17916 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
17918 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
17919 (machine_function::call_ms2sysv_pad_out): Remove field.
17920 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
17921 (ix86_compute_frame_layout): Likewise.
17923 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
17926 * config/i386/i386.c (ix86_set_func_type): Disallow naked
17927 attribute with interrupt attribute.
17929 2017-08-01 Andrew Pinski <apinski@cavium.com>
17931 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
17932 BIT_INSERT_EXPR's operand 1
17933 to see if the types precision matches.
17935 2017-08-01 Martin Liska <mliska@suse.cz>
17937 PR middle-end/70140
17938 * builtins.c (expand_builtin_memcpy_args): Remove.
17939 (expand_builtin_memcpy): Call newly added function
17940 expand_builtin_memory_copy_args.
17941 (expand_builtin_memcpy_with_bounds): Likewise.
17942 (expand_builtin_mempcpy): Remove last argument.
17943 (expand_builtin_mempcpy_with_bounds): Likewise.
17944 (expand_builtin_memory_copy_args): New function created from
17945 expand_builtin_mempcpy_args with small modifications.
17946 (expand_builtin_mempcpy_args): Remove.
17947 (expand_builtin_stpcpy): Remove unused argument.
17948 (expand_builtin): Likewise.
17949 (expand_builtin_with_bounds): Likewise.
17951 2017-08-01 Martin Liska <mliska@suse.cz>
17954 Make mempcpy more optimal (PR middle-end/70140).
17956 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17959 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
17960 __builtin_vec_cmpne verify both arguments are compatible vectors
17961 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
17962 verify arg1_type is a pointer or array type. For __builtin_vec_st,
17963 move computation of aligned to after checking the argument types.
17967 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17968 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
17971 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
17972 Doug Rupp <rupp@adacore.com>
17973 Olivier Hainque <hainque@adacore.com>
17975 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
17976 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
17978 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
17979 for TARGET_OS_CPP_BUILTIN.
17980 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
17981 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
17983 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
17984 passing required abi options to the assembler for EABI configurations.
17985 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
17986 of .text.hot and .text.unlikely sections for kernel modules when
17987 using ARM style exceptions.
17988 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
17989 options. Add EXTRA_CC1_SPEC.
17990 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
17992 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
17994 (ARM_TARGET2_DWARF_FORMAT): Define.
17995 * config/arm/t-vxworks: Adjust multilib control to removal of the
17996 Diab command line options.
17998 2017-08-01 Martin Liska <mliska@suse.cz>
18000 PR gcov-profile/81561
18001 * gcov.c (unblock): Make unblocking safe as we need to preserve
18002 index correspondence of blocks and block_lists.
18004 2017-08-01 Richard Biener <rguenther@suse.de>
18006 PR tree-optimization/81181
18007 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18008 (compute_antic): ... end of iteration here.
18010 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18012 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18013 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18014 (ftree-slp-vectorize): Likewise.
18015 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18016 can no longer be set independent of flag_tree_loop_vectorize.
18017 * omp-general.c (emp_max_vf): Likewise.
18018 * opts.c (enable_fdo_optimizations): Remove references to
18019 flag_tree_vectorize, these are now implicit.
18020 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18021 and leave it for the options machinery.
18023 2017-08-01 Martin Liska <mliska@suse.cz>
18025 PR middle-end/70140
18026 * builtins.c (expand_builtin_memcpy_args): Remove.
18027 (expand_builtin_memcpy): Call newly added function
18028 expand_builtin_memory_copy_args.
18029 (expand_builtin_memcpy_with_bounds): Likewise.
18030 (expand_builtin_mempcpy): Remove last argument.
18031 (expand_builtin_mempcpy_with_bounds): Likewise.
18032 (expand_builtin_memory_copy_args): New function created from
18033 expand_builtin_mempcpy_args with small modifications.
18034 (expand_builtin_mempcpy_args): Remove.
18035 (expand_builtin_stpcpy): Remove unused argument.
18036 (expand_builtin): Likewise.
18037 (expand_builtin_with_bounds): Likewise.
18039 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18042 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18043 print "ds:" only for immediates in generic address space.
18045 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18048 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18049 (ix86_function_ok_for_sibcall): Return false for naked functions.
18051 2017-08-01 Richard Biener <rguenther@suse.de>
18053 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18054 (compute_antic): Seed worklist with exit block predecessors.
18055 * cfganal.c (dfs_find_deadend): For a cycle return the source
18056 of the edge closing it.
18058 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18060 * config/aarch64/aarch64.c
18061 (aarch64_can_const_movi_rtx_p): Move 0 check.
18063 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18065 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18066 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18068 * match.pd: Ditto in address comparison pattern.
18070 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18072 PR tree-optimization/81627
18073 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18074 closed ssa form for store-store chain.
18076 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18078 PR tree-optimization/81620
18079 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18080 for store-store chain.
18082 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18084 PR tree-optimization/81588
18085 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18086 ranges[i].in_p, invert comparison code ccode. For >/>=,
18087 swap rhs1 and rhs2 and comparison code unconditionally,
18088 for </<= don't do that. Don't swap rhs1/rhs2 again if
18089 ranges[i].in_p, instead invert comparison code ccode if
18090 opcode or oe->rank is BIT_IOR_EXPR.
18093 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18094 a direct optab to conversion optab.
18095 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18096 with GET_MODE_INNER as last argument instead of optab_handler.
18097 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18098 vector extraction if possible and optab is available.
18099 * expr.c (store_constructor): Use convert_optab_handler instead
18100 of optab_handler. Use vector initialization from smaller
18101 vectors if possible and optab is available.
18102 * tree-vect-stmts.c (vectorizable_load): Likewise.
18103 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18104 now have two modes.
18105 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18106 of vec_init from half-sized vectors with the same element mode.
18107 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18108 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18109 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18110 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18111 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18112 after mode in gen_vec_extract* calls.
18113 (vec_extract<mode>): Renamed to ...
18114 (vec_extract<mode><ssescalarmodelower>): ... this.
18115 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18116 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18117 element mode after mode in gen_vec_init* calls.
18118 (VEC_INIT_HALF_MODE): New mode iterator.
18119 (vec_init<mode>): Renamed to ...
18120 (vec_init<mode><ssescalarmodelower>): ... this.
18121 (vec_init<mode><ssehalfvecmodelower>): New expander.
18122 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18123 (vec_extractv2sfsf): ... this.
18124 (vec_initv2sf): Renamed to ...
18125 (vec_initv2sfsf): ... this.
18126 (vec_extractv2si): Renamed to ...
18127 (vec_extractv2sisi): ... this.
18128 (vec_initv2si): Renamed to ...
18129 (vec_initv2sisi): ... this.
18130 (vec_extractv4hi): Renamed to ...
18131 (vec_extractv4hihi): ... this.
18132 (vec_initv4hi): Renamed to ...
18133 (vec_initv4hihi): ... this.
18134 (vec_extractv8qi): Renamed to ...
18135 (vec_extractv8qiqi): ... this.
18136 (vec_initv8qi): Renamed to ...
18137 (vec_initv8qiqi): ... this.
18138 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18139 (vec_init<mode>): Renamed to ...
18140 (vec_init<mode><VEC_base_l>): ... this.
18141 (vec_extract<mode>): Renamed to ...
18142 (vec_extract<mode><VEC_base_l>): ... this.
18143 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18144 (vec_initv2sfsf): ... this.
18145 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18146 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18147 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18148 element mode after mode in gen_vec_init* calls.
18149 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18150 (vec_init<mode><Vel>): ... this.
18151 (vec_extract<mode>): Renamed to ...
18152 (vec_extract<mode><Vel>): ... this.
18153 * config/aarch64/iterators.md (Vel): New mode attribute.
18154 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18155 Add element mode after mode in gen_vec_extract* calls.
18156 * config/s390/vector.md (non_vec_l): New mode attribute.
18157 (vec_extract<mode>): Renamed to ...
18158 (vec_extract<mode><non_vec_l>): ... this.
18159 (vec_init<mode>): Renamed to ...
18160 (vec_init<mode><non_vec_l>): ... this.
18161 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18162 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18164 * config/arm/iterators.md (V_elem_l): New mode attribute.
18165 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18166 (vec_extract<mode><V_elem_l>): ... this.
18167 (vec_extractv2di): Renamed to ...
18168 (vec_extractv2didi): ... this.
18169 (vec_init<mode>): Renamed to ...
18170 (vec_init<mode><V_elem_l>): ... this.
18171 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18172 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18173 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18174 Add element mode after gen_vec_extract* calls.
18175 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18176 (vec_init<mode><unitmode>): ... this.
18177 (vec_extract<mode>): Renamed to ...
18178 (vec_extract<mode><unitmode>): ... this.
18179 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18180 (vec_init<mode><unitmode>): ... this.
18181 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18182 (vec_initv2sfsf): ... this.
18183 (vec_extractv2sf): Renamed to ...
18184 (vec_extractv2sfsf): ... this.
18185 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18186 Add element mode after gen_vec_extract* calls.
18187 * config/mips/mips.md (unitmode): New mode iterator.
18188 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18189 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18190 * config/spu/spu.md (inner_l): New mode attribute.
18191 (vec_init<mode>): Renamed to ...
18192 (vec_init<mode><inner_l>): ... this.
18193 (vec_extract<mode>): Renamed to ...
18194 (vec_extract<mode><inner_l>): ... this.
18195 * config/sparc/sparc.md (veltmode): New mode iterator.
18196 (vec_init<VMALL:mode>): Renamed to ...
18197 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18198 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18199 (vec_initv2sisi): ... this.
18200 (vec_initv2sf): Renamed to ...
18201 (vec_initv2sfsf): ... this.
18202 (vec_extractv2sf): Renamed to ...
18203 (vec_extractv2sfsf): ... this.
18204 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18205 (vec_init<mode>): Renamed to ...
18206 (vec_init<mode><VEC_base_l>): ... this.
18207 (vec_extract<mode>): Renamed to ...
18208 (vec_extract<mode><VEC_base_l>): ... this.
18209 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18210 (vec_initv2sfsf): ... this.
18211 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
18212 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18213 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
18214 gen_vec_init* calls.
18216 2017-08-01 Richard Biener <rguenther@suse.de>
18218 PR tree-optimization/81297
18219 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
18220 TREE_OVERFLOW from INTEGER_CSTs.
18222 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
18224 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
18226 2017-07-31 Carl Love <cel@us.ibm.com>
18228 * config/rs6000/rs6000-c: Add support for built-in functions
18229 vector signed char vec_xl_be (signed long long, signed char *);
18230 vector unsigned char vec_xl_be (signed long long, unsigned char *);
18231 vector signed int vec_xl_be (signed long long, signed int *);
18232 vector unsigned int vec_xl_be (signed long long, unsigned int *);
18233 vector signed long long vec_xl_be (signed long long, signed long long *);
18234 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
18235 vector signed short vec_xl_be (signed long long, signed short *);
18236 vector unsigned short vec_xl_be (signed long long, unsigned short *);
18237 vector double vec_xl_be (signed long long, double *);
18238 vector float vec_xl_be (signed long long, float *);
18239 * config/rs6000/altivec.h (vec_xl_be): Add #define.
18240 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
18241 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
18243 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
18244 (altivec_expand_builtin): Add switch statement to call
18245 altivec_expand_xl_be for each builtin.
18246 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
18247 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
18248 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
18249 __builtin_vsx_le_be_v16qi.
18250 * doc/extend.texi: Update the built-in documentation file for the
18251 new built-in functions.
18253 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18256 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
18258 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18260 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18262 * config.gcc: Add z14.
18263 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
18264 CPU model numbers for z13s and z14.
18265 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
18267 * config/s390/s390-opts.h (enum processor_type): Rename
18268 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18269 * config/s390/s390.c (processor_table): Add field for CPU name to
18270 be passed to Binutils.
18271 (s390_asm_output_machine_for_arch): Use the new field in
18272 processor_table for Binutils.
18273 (s390_expand_builtin): Replace arch12 with z14.
18274 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18275 (s390_get_sched_attrmask): Likewise.
18276 (s390_get_unit_mask): Likewise.
18277 * config/s390/s390.opt: Add z14 to processor_type enum.
18279 2017-07-31 Martin Jambor <mjambor@suse.cz>
18282 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
18283 regardless of optimization level.
18285 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
18286 Martin Liska <mliska@suse.cz>
18288 * predict.def: Remove old comment and adjust probability.
18289 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
18290 PREDICT statements.
18292 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18295 * config/i386/i386.c (ix86_function_naked): New function.
18296 (ix86_can_use_return_insn_p): Return false for naked functions.
18297 (ix86_expand_prologue): Skip prologue for naked functions.
18298 (ix86_expand_epilogue): Skip epilogue for naked functions
18299 and emit trap instruction.
18300 (ix86_warn_func_return): New function.
18301 (ix86_attribute_table): Add "naked" attribute specification.
18302 (TARGET_WARN_FUNC_RETURN): Define.
18303 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
18305 2017-07-31 Martin Liska <mliska@suse.cz>
18307 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
18308 (dump_gimple_bb_header): Always dump BB info.
18309 (pp_cfg_jump): Do not append info about BB when dumping a jump.
18311 2017-07-31 Martin Liska <mliska@suse.cz>
18314 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
18315 also with current_function_decl non-null equality.
18317 2017-07-31 Jakub Jelinek <jakub@redhat.com>
18320 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
18321 change type to the element type, instead add eltype variable and
18322 use it where we are interested in the element type.
18324 PR tree-optimization/81603
18325 * ipa-polymorphic-call.c
18326 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
18327 offset arithmetic in offset_int, bail out if the resulting bit offset
18328 doesn't fit into shwi.
18330 2017-07-31 Martin Liska <mliska@suse.cz>
18332 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
18333 (gimplify_save_expr): Fix comment.
18335 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
18338 * config/i386/i386.c (ix86_function_arg): Update arguments for
18340 (ix86_compute_frame_layout): Set the initial stack offset to
18341 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
18342 INCOMING_FRAME_SP_OFFSET.
18343 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
18344 stack before exception handler returns.
18345 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
18346 the 'ERROR_CODE' for exception handler.
18348 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
18350 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
18351 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
18352 (ASM_OUTPUT_REG_POP): Ditto.
18353 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
18354 instead of asm_fprintf to output pure string.
18356 2017-07-29 Jakub Jelinek <jakub@redhat.com>
18358 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
18359 to imported_module_or_decl hook.
18360 (debug_nothing_tree_tree_tree_bool): Remove.
18361 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
18362 * debug.c (do_nothing_debug_hooks): Use
18363 debug_nothing_tree_tree_tree_bool_bool instead of
18364 debug_nothing_tree_tree_tree_bool.
18365 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18366 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
18367 * sdbout.c (sdb_debug_hooks): Likewise.
18368 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
18369 (gen_namespace_die): Add DW_AT_export_symbols attribute if
18371 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
18372 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
18373 attribute, don't add anything.
18375 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18377 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
18378 (fold_build2_stat_loc): Likewise.
18379 (fold_build3_stat_loc): Likewise.
18380 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
18381 (fold_build1_loc): Remove macro.
18382 (fold_build2_loc): Likewise.
18383 (fold_build3_loc): Likewise.
18385 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18387 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
18388 (gimple_build_debug_bind_source_stat): Likewise.
18389 * gimple.h (gimple_build_debug_bind): Remove macro.
18390 (gimple_build_debug_bind_source): Likewise.
18392 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18394 * bitmap.c (bitmap_alloc): Adjust.
18395 (bitmap_gc_alloc): Likewise.
18396 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
18398 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18400 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
18401 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
18402 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
18403 (bitmap_gc_alloc_stat): Likewise.
18404 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
18406 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18408 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
18409 * rtl.h (shallow_copy_rtx): Remove macro.
18411 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18413 * emit-rtl.c (gen_raw_REG): Adjust.
18414 * gengenrtl.c (gendef): Likewise.
18415 * rtl.c (rtx_alloc_stat): Remove _stat from name.
18416 * rtl.h (rtx_alloc): Remove macro.
18418 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18420 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
18421 (build_tree_list_stat): Likewise.
18422 * tree.h (build_tree_list): Remove macro.
18423 (build_tree_list_vec): Likewise.
18425 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18427 * tree.c (make_vector_stat): Remove _stat from name.
18428 (build_vector_stat): Likewise.
18429 * tree.h (make_vector_stat): Remove macro.
18430 (build_vector_stat): Likewise.
18432 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18434 * tree.h (build_var_debug_value): Remove prototype.
18436 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18438 * tree.c (tree_cons_stat): Remove _stat from name.
18439 * tree.h (tree_cons): Remove macro.
18441 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18443 * tree.c (build_vl_exp_stat): Remove _stat from name.
18444 * tree.h (build_vl_exp): Remove macro.
18446 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18448 * tree.c (build_decl_stat): Remove _stat from name.
18449 * tree.h (build_decl): Remove macro.
18451 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18453 * gimple.c (gimple_build_with_ops_stat): Adjust.
18454 (gimple_alloc_stat): Remove _stat from name.
18455 * gimple.h (gimple_alloc): Remove macro.
18457 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18459 * tree.c (make_tree_vec_stat): Remove _stat from name.
18460 (grow_tree_vec_stat): Likewise.
18461 * tree.h (make_tree_vec_stat): Adjust prototype.
18462 (grow_tree_vec_stat): Likewise.
18463 (make_tree_vec): Remove macro.
18464 (grow_tree_vec): Likewise.
18466 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18468 * fold-const.c (fold_build1_stat_loc): Adjust.
18469 (fold_build2_stat_loc): Likewise.
18470 (fold_build3_stat_loc): Likewise.
18471 * tree.c (build0_stat): Remove _stat from name.
18472 (build1_stat): Likewise.
18473 (build2_stat): Likewise.
18474 (build3_stat): Likewise.
18475 (build4_stat): Likewise.
18476 (build5_stat): Likewise.
18477 * tree.h (build1_loc): Remove macro, and rename _stat function
18479 (build2_loc): Likewise.
18480 (build3_loc): Likewise.
18481 (build4_loc): Likewise.
18482 (build5_loc): Likewise.
18484 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18486 * tree.c (make_int_cst_stat): Remove _stat from name.
18487 * tree.h (make_int_cst_stat): Adjust prototype.
18488 (make_int_cst): Remove macro.
18490 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18492 * tree.c (make_tre_binfo_stat): Remove _stat from name.
18493 * tree.h (make_tree_binfo_stat): Adjust prototype.
18494 (make_tree_binfo): Remove.
18496 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18498 * tree.c (copy_node_stat): Rename to copy_node.
18499 (build_distinct_type_copy): Adjust.
18500 * tree.h (copy_node_stat): Adjust prototype.
18501 (copy_node): Remove macro.
18503 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18505 * tree.c (make_node_stat): rename to make_node.
18506 (build_tree_list_stat): Adjust.
18507 (build0_stat): Likewise.
18508 (build2_stat): Likewise.
18509 (build3_stat): Likewise.
18510 (build4_stat): Likewise.
18511 (build5_stat): Likewise.
18512 (build_decl_stat): Likewise.
18513 * tree.h (make_node_stat): Adjust prototype.
18514 (make_node): remove macro.
18516 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
18518 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
18519 (PPC_FEATURE2_SCV): Likewise.
18520 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
18522 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18524 * config/aarch64/aarch64.c
18525 (aarch64_internal_mov_immediate): Add new special pattern.
18526 * config/aarch64/aarch64.md (*movdi_aarch64):
18527 Add reg/32bit const mov case.
18529 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18530 Richard Sandiford <richard.sandiford@linaro.org>
18532 * config/aarch64/aarch64.md (mov<mode>): Generalize.
18533 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18534 Add integer and movi cases.
18535 (movi-split-hf-df-sf split, fp16): New.
18536 (enabled): Added TARGET_FP_F16INST.
18537 * config/aarch64/iterators.md (GPF_HF): New.
18538 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
18540 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18542 * config/aarch64/aarch64.c
18543 (aarch64_simd_container_mode): Add prototype.
18544 (aarch64_expand_mov_immediate): Add HI support.
18545 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
18546 (aarch64_can_const_movi_rtx_p): New.
18547 (aarch64_preferred_reload_class):
18548 Remove restrictions of using FP registers for certain SIMD operations.
18549 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
18550 (aarch64_valid_floating_const): Add integer move validation.
18551 (aarch64_simd_imm_scalar_p): Remove.
18552 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
18553 (aarch64_legitimate_constant_p): Expand list of supported cases.
18554 * config/aarch64/aarch64-protos.h
18555 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
18556 (aarch64_reinterpret_float_as_int): New.
18557 (aarch64_simd_imm_scalar_p): Remove.
18558 * config/aarch64/constraints.md (Uvi): New.
18559 (Dd): Split into Ds and new Dd.
18560 * config/aarch64/aarch64.md (*movsi_aarch64):
18562 (*movdi_aarch64): Add SIMD mov case.
18564 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18566 * tree-predcom.c: (struct chain): Handle store-store chain in which
18567 stores for elimination only store loop invariant values.
18568 (execute_pred_commoning_chain): Ditto.
18569 (prepare_initializers_chain_store_elim): Ditto.
18570 (prepare_finalizers): Ditto.
18571 (is_inv_store_elimination_chain): New function.
18572 (initialize_root_vars_store_elim_1): New function.
18574 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18576 * tree-predcom.c: Revise general description of the pass.
18577 (enum chain_type): New enum type for store elimination.
18578 (struct chain): New field supporting store elimination.
18579 (struct component): Ditto.
18580 (dump_chain): Dump store-stores chain.
18581 (release_chain): Release resources.
18582 (split_data_refs_to_components): Compute and create component
18583 contains only stores for elimination.
18584 (get_chain_last_ref_at): New function.
18585 (make_invariant_chain): Initialization.
18586 (make_rooted_chain): Specify chain type in parameter and record it.
18587 (add_looparound_copies): Skip for store-stores chain.
18588 (determine_roots_comp): Compute type of chain and pass it to
18590 (initialize_root_vars_store_elim_2): New function.
18591 (finalize_eliminated_stores): New function.
18592 (remove_stmt): Handle store for elimination.
18593 (execute_pred_commoning_chain): Execute predictive commoning on
18594 store-store chains.
18595 (determine_unroll_factor): Skip unroll for store-stores chain.
18596 (prepare_initializers_chain_store_elim): New function.
18597 (prepare_initializers_chain): Hanlde store-store chain.
18598 (prepare_finalizers_chain, prepare_finalizers): New function.
18599 (tree_predictive_commoning_loop): Return integer value indicating
18600 if loop is unrolled or lcssa form is corrupted.
18601 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
18603 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18605 * tree-predcom.c (initialize_root): Delete.
18606 (execute_pred_commoning_chain): Initialize root vars and replace
18607 reference of non-combined chain directly, rather than call above
18610 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18612 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
18613 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
18615 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18617 * tree-predcom.c (struct chain): New field init_seq.
18618 (release_chain): Release init_seq.
18619 (prepare_initializers_chain): Record intialization stmts in above
18621 (insert_init_seqs): New function.
18622 (tree_predictive_commoning_loop): Call insert_init_seqs.
18624 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18626 * tree-predcom.c (determine_roots_comp): Skip trivial components.
18628 2017-07-28 Richard Biener <rguenther@suse.de>
18630 * match.pd: Remove superfluous :c.
18631 * genmatch.c (simplify::id): Add member.
18632 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
18634 (current_id): New global.
18635 (dt_node::parent): Move from ...
18636 (dt_operand::parent): ... here. Add for_id member.
18637 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
18638 (decision_tree::find_node): Relax order requirement when
18639 merging DT_TRUE nodes to ones inbetween the current simplify
18640 and the one we try to merge with. Add diagnostic whenever
18641 we need to enforce pattern order by not merging.
18642 (decision_tree::insert): Set current_id.
18643 (decision_tree::print_node): Dump parent node and for_id.
18644 (parser::last_id): Add member.
18645 (parser::push_simplify): Assign unique id.
18646 (parser::parser): Initialize last_id.
18648 2017-07-28 Martin Liska <mliska@suse.cz>
18651 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
18652 gimple_build_debug_bind.
18654 2017-07-28 Richard Biener <rguenther@suse.de>
18656 PR tree-optimization/81502
18657 * match.pd: Add pattern combining BIT_INSERT_EXPR with
18659 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
18661 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
18662 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
18663 for BIT_FIELD_REF args.
18664 * fold-const.c (make_bit_field_ref): Likewise.
18665 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
18667 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18670 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
18671 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
18672 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
18673 Or it into SANITIZER_UNDEFINED.
18674 * ubsan.c: Include gimple-fold.h and varasm.h.
18675 (ubsan_expand_ptr_ifn): New function.
18676 (instrument_pointer_overflow): New function.
18677 (maybe_instrument_pointer_overflow): New function.
18678 (instrument_object_size): Formatting fix.
18679 (pass_ubsan::execute): Call instrument_pointer_overflow
18680 and maybe_instrument_pointer_overflow.
18681 * internal-fn.c (expand_UBSAN_PTR): New function.
18682 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
18683 * sanitizer.def (__ubsan_handle_pointer_overflow,
18684 __ubsan_handle_pointer_overflow_abort): New builtins.
18685 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
18686 * internal-fn.def (UBSAN_PTR): New internal function.
18687 * opts.c (sanitizer_opts): Add pointer-overflow.
18688 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
18689 * fold-const.c (build_range_check): Compute pointer range check in
18690 integral type if pointer arithmetics would be needed. Formatting
18693 2017-07-28 Martin Liska <mliska@suse.cz>
18696 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
18697 parameters that are of a variable-length.
18699 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
18701 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
18703 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
18704 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
18705 (CRT_CALL_STATIC_FUNCTION): Likewise.
18706 (ASM_DEFAULT_SPEC): New define.
18707 (ASM_SPEC32): Likewise.
18708 (ASM_SPEC64): Likewise.
18709 (ASM_SPEC_COMMON): Likewise.
18710 (ASM_SPEC): Likewise.
18711 (INVALID_64BIT): Likewise.
18712 (LINK_OS_DEFAULT_SPEC): Likewise.
18713 (LINK_OS_SPEC32): Likewise.
18714 (LINK_OS_SPEC64): Likewise.
18715 (POWERPC_LINUX): Likewise.
18716 (PTRDIFF_TYPE): Likewise.
18717 (RESTORE_FP_PREFIX): Likewise.
18718 (RESTORE_FP_SUFFIX): Likewise.
18719 (SAVE_FP_PREFIX): Likewise.
18720 (SAVE_FP_SUFFIX): Likewise.
18721 (SIZE_TYPE): Likewise.
18722 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
18723 (TARGET_64BIT): Likewise.
18724 (TARGET_64BIT): Likewise.
18725 (TARGET_AIX): Likewise.
18726 (WCHAR_TYPE_SIZE): Likewise.
18727 (WCHAR_TYPE): Undefine.
18728 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
18729 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
18730 (CPP_OS_RTEMS_SPEC): Delete.
18731 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
18732 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
18734 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
18736 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18738 PR tree-optimization/81578
18739 * tree-parloops.c (build_new_reduction): Bail out if
18740 reduction_code isn't one of the standard OpenMP reductions.
18741 Move the details printing after that decision.
18743 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18745 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
18746 related to reload_in_progress.
18747 (splat_input_operand): Likewise.
18748 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
18750 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
18752 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
18753 (TARGET_INSTANTIATE_DECLS): Likewise.
18754 (legitimate_indexed_address_p): Delete reload_in_progress code.
18755 (rs6000_debug_legitimate_address_p): Likewise.
18756 (rs6000_eliminate_indexed_memrefs): Likewise.
18757 (rs6000_emit_le_vsx_store): Likewise.
18758 (rs6000_emit_move_si_sf_subreg): Likewise.
18759 (rs6000_emit_move): Likewise.
18760 (register_to_reg_type): Likewise.
18761 (rs6000_pre_atomic_barrier): Likewise.
18762 (rs6000_machopic_legitimize_pic_address): Likewise.
18763 (rs6000_allocate_stack_temp): Likewise.
18764 (rs6000_address_for_fpconvert): Likewise.
18765 (rs6000_address_for_altivec): Likewise.
18766 (rs6000_secondary_memory_needed_rtx): Delete function.
18767 (rs6000_check_sdmode): Likewise.
18768 (rs6000_alloc_sdmode_stack_slot): Likewise.
18769 (rs6000_instantiate_decls): Likewise.
18770 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
18771 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
18772 Delete reload_in_progress.
18773 (*vec_reload_and_plus_<mptrsize>): Likewise.
18774 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
18775 (vsx_div_v2di): Likewise.
18776 (vsx_udiv_v2di): Likewise.
18778 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18780 * config/rs6000/rs6000.opt (mlra): Replace with stub.
18781 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
18782 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
18783 (rs6000_debug_reg_global): Delete print of LRA status.
18784 (rs6000_option_override_internal): Delete dead LRA related code.
18785 (rs6000_lra_p): Delete function.
18786 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
18788 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
18790 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
18791 * config/riscv/rtems.h: New file.
18793 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18794 Sudakshina Das <sudi.das@arm.com>
18796 * config/aarch64/aarch64.md
18797 (define_split for and<mode>3nr_compare): Move
18798 non aarch64_logical_operand to a register.
18799 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
18800 register immediate operand to a register.
18801 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
18803 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18805 PR middle-end/81564
18806 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
18808 2017-07-27 Richard Biener <rguenther@suse.de>
18810 PR tree-optimization/81573
18811 PR tree-optimization/81494
18812 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
18813 multi defuse cycle case.
18815 2017-07-27 Richard Biener <rguenther@suse.de>
18817 PR tree-optimization/81571
18818 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
18821 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
18823 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
18824 earlier and only if MASK_FPU is set. Adjust formatting.
18826 2017-07-27 Martin Liska <mliska@suse.cz>
18828 * opt-functions.awk: Add validation of value of Init.
18829 * optc-gen.awk: Pass new argument.
18831 2017-07-27 Martin Liska <mliska@suse.cz>
18833 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
18834 Fix wrong condition.
18836 2017-07-27 Martin Liska <mliska@suse.cz>
18838 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
18839 BBs and edges seen by autoFDO.
18841 2017-07-27 Richard Biener <rguenther@suse.de>
18843 PR tree-optimization/81502
18844 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
18845 with incompatible but same sized type.
18846 (execute_update_addresses_taken): Likewise.
18848 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
18850 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
18851 flag_tree_loop_vectorize rather than flag_tree_vectorize.
18853 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18856 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
18857 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
18858 Change s_operand to memory_operand.
18860 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
18862 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
18863 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
18864 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
18865 Emit instructions rather than returning an expression. Handle TFmode
18866 and KFmode by casting to TImode.
18867 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
18868 (rs6000_emit_le_vsx_store): Likewise.
18869 * config/rs6000/vsx.md (VSX_TI): New iterator.
18870 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
18871 (*vsx_le_undo_permute_<mode>): Likewise.
18872 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
18873 emit the split sequence.
18874 (*vsx_le_perm_store_<mode>): Likewise.
18876 2017-07-27 Jakub Jelinek <jakub@redhat.com>
18878 PR tree-optimization/81555
18879 PR tree-optimization/81556
18880 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
18881 if true, force CHANGED for the recursive invocation.
18882 (reassociate_bb): Remember original length of ops array, pass
18883 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
18885 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
18886 attributes for noipa attribute. For naked attribute use
18887 lookup_attribute first before lookup_attribute_spec.
18888 * final.c (rest_of_handle_final): Disable IPA RA for functions with
18890 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
18891 for functions with noipa attribute.
18892 (cgraph_externally_visible_p): Return true for functions with noipa
18894 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
18895 for functions with noipa attribute.
18896 * doc/extend.texi: Document noipa function attribute.
18897 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
18898 also for functions with noipa attribute.
18899 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
18901 2017-07-26 Andrew Pinski <apinski@cavium.com>
18903 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
18904 vec_unalign_load_cost and vec_unalign_store_cost.
18906 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
18908 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
18909 -mvsx-small-integer option.
18910 (ISA_3_0_MASKS_IEEE): Likewise.
18911 (OTHER_VSX_VECTOR_MASKS): Likewise.
18912 (POWERPC_MASKS): Likewise.
18913 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
18914 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
18915 code, only testing for DImode being allowed in non-VSX floating
18917 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
18918 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
18920 (rs6000_option_override_internal): Delete -mvsx-small-integer.
18921 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
18922 TARGET_P8_VECTOR test.
18923 (rs6000_secondary_reload_simple_move): Likewise.
18924 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
18925 since TARGET_P9_VECTOR was already tested.
18926 (rs6000_opt_masks): Remove -mvsx-small-integer.
18927 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
18928 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18930 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
18931 test for TARGET_VEXTRACTUB was used, and that uses
18933 (p9 extract splitter): Likewise.
18934 (vsx_extract_<mode>_di_p9): Likewise.
18935 (vsx_extract_<mode>_store_p9): Likewise.
18936 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
18937 for TARGET_P9_VECTOR was used. Delete code that is now dead with
18938 the elimination of TARGET_VSX_SMALL_INTEGER.
18939 (vsx_extract_<mode>_p8): Likewise.
18940 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
18941 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
18942 (vsx_set_<mode>_p9): Likewise.
18943 (vsx_set_v4sf_p9): Likewise.
18944 (vsx_set_v4sf_p9_zero): Likewise.
18945 (vsx_insert_extract_v4sf_p9): Likewise.
18946 (vsx_insert_extract_v4sf_p9_2): Likewise.
18947 * config/rs6000/rs6000.md (sign extend splitter): Change
18948 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
18949 (floatsi<mode>2_lfiwax_mem): Likewise.
18950 (floatunssi<mode>2_lfiwzx_mem): Likewise.
18951 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
18952 since a test for TARGET_P9_VECTOR was used.
18953 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18954 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
18955 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18956 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18957 TARGET_P8_VECTOR test.
18958 (fix_trunc<mode>si2_stfiwx): Likewise.
18959 (fix_trunc<mode>si2_internal): Likewise.
18960 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
18961 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18963 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18964 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18965 TARGET_P8_VECTOR test.
18966 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18967 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
18968 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18970 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18971 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
18972 since a test for TARGET_P9_VECTOR was used.
18973 (splitter for loading small constants): Likewise.
18975 2017-07-26 Andrew Pinski <apinski@cavium.com>
18977 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
18980 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
18983 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
18984 (fp_valid_at): Likewise.
18986 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18988 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
18989 (qdf24xx_addrcost_table): Likewise.
18990 (cortexa57_tunings): Update to use generic_branch_cost.
18991 (cortexa72_tunings): Likewise.
18992 (cortexa73_tunings): Likewise.
18993 (qdf24xx_tunings): Likewise.
18995 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18997 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
18998 (thunderx2t99_branch_cost): Likewise.
18999 (cortexa35_tunings): Update to use generic_branch_cost.
19000 (cortexa53_tunings): Likewise.
19001 (cortexa57_tunings): Likewise.
19002 (cortexa72_tunings): Likewise.
19003 (cortexa73_tunings): Likewise.
19004 (thunderx2t99_tunings): Likewise.
19006 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19008 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19009 (sparc_option_override): Honour MASK_FSMULD.
19010 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19011 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19012 * config/sparc/sparc.opt (mfsmuld): New option.
19013 * doc/invoke.texi (mfsmuld): Document option.
19015 2017-07-26 Marek Polacek <polacek@redhat.com>
19017 PR middle-end/70992
19018 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19020 2017-07-26 Richard Biener <rguenther@suse.de>
19022 * gimple-match-head.c (do_valueize): Return OP if valueize
19024 (get_def): New helper to get at the def stmt of a SSA name
19025 if valueize allows.
19026 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19027 do_valueize to get at the def stmt.
19028 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19030 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19032 PR middle-end/46932
19033 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19035 2017-07-26 Martin Liska <mliska@suse.cz>
19038 * function.c (expand_function_start): Make expansion of
19039 nonlocal_goto_save_area after parm_birth_insn.
19041 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19043 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19044 from all CPU target flags enable members.
19046 2017-07-26 Richard Biener <rguenther@suse.de>
19048 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19049 (decision_tree::gen): Make 'type' const.
19050 (write_predicate): Likewise.
19052 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19054 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19055 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19056 (rs6000_option_override_internal): Likewise.
19057 (rs6000_expand_vector_set): Likewise.
19058 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19059 (TARGET_UPPER_REGS_SF): Likewise.
19060 (TARGET_UPPER_REGS_DI): Likewise.
19061 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19062 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19063 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19064 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19065 (Splitters for DI constants in Altivec registers): Likewise.
19066 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19067 (vsx_set_v4sf_p9): Likewise.
19068 (vsx_set_v4sf_p9_zero): Likewise.
19069 (vsx_insert_extract_v4sf_p9): Likewise.
19070 (vsx_insert_extract_v4sf_p9_2): Likewise.
19072 2017-07-25 Carl Love <cel@us.ibm.com>
19074 * doc/extend.texi: Update the built-in documentation file for the
19075 existing built-in functions
19076 vector signed char vec_cnttz (vector signed char);
19077 vector unsigned char vec_cnttz (vector unsigned char);
19078 vector signed short vec_cnttz (vector signed short);
19079 vector unsigned short vec_cnttz (vector unsigned short);
19080 vector signed int vec_cnttz (vector signed int);
19081 vector unsigned int vec_cnttz (vector unsigned int);
19082 vector signed long long vec_cnttz (vector signed long long);
19083 vector unsigned long long vec_cnttz (vector unsigned long long);
19085 2017-07-25 Andrew Pinski <apinski@cavium.com>
19087 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19088 accesses where the use is for the first operand of a BIT_INSERT.
19090 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19093 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19094 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19096 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19098 * config/i386/gstabs.h: Delete.
19099 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19101 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19103 * config/i386/i386.c (ix86_decompose_address): Do not check for
19104 register RTX when looking at index_reg or base_reg.
19105 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19107 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19109 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19110 to update EH info here.
19112 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19114 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19116 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19118 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19120 2017-07-25 Torsten Duwe <duwe@suse.de>
19122 * common.opt: Introduce -fpatchable-function-entry
19123 command line option, and its variables function_entry_patch_area_size
19124 and function_entry_patch_area_start.
19125 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19126 including a two-value parser.
19127 * target.def (print_patchable_function_entry): New target hook.
19128 * targhooks.h (default_print_patchable_function_entry): New function.
19129 * targhooks.c (default_print_patchable_function_entry): Likewise.
19130 * toplev.c (process_options): Switch off IPA-RA if
19131 patchable function entries are being generated.
19132 * varasm.c (assemble_start_function): Look at the
19133 patchable-function-entry command line switch and current
19134 function attributes and maybe generate NOP instructions by
19135 calling the print_patchable_function_entry hook.
19136 * doc/extend.texi: Document patchable_function_entry attribute.
19137 * doc/invoke.texi: Document -fpatchable_function_entry
19138 command line option.
19139 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19141 * doc/tm.texi: Re-generate.
19143 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19146 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19147 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19149 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19151 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19153 2017-07-25 Richard Biener <rguenther@suse.de>
19155 PR tree-optimization/81455
19156 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19157 not walk in cycles when looking for guards.
19159 2017-07-25 Richard Biener <rguenther@suse.de>
19161 PR tree-optimization/81529
19162 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19163 when optimizing backedge uses.
19165 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19167 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19169 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19170 to dl_section_ref. On AIX, append an expression to subtract
19171 the size of the section length to dl_section_ref.
19173 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19175 * configure.ac: If any of the config.* scripts fail, exit 1.
19176 * configure: Regenerate.
19178 2017-07-25 Richard Biener <rguenther@suse.de>
19180 PR middle-end/81546
19181 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19182 of immediate uses, be more verbose on errors.
19184 2017-07-25 Richard Biener <rguenther@suse.de>
19186 PR tree-optimization/81510
19187 * tree-vect-loop.c (vect_is_simple_reduction): When the
19188 reduction stmt is not inside the loop bail out.
19190 2017-07-25 Richard Biener <rguenther@suse.de>
19192 PR tree-optimization/81303
19193 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19194 profitability check against LOOP_VINFO_NITERSM1.
19196 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19198 * domwalk.c (cmp_bb_postorder): Simplify.
19199 (sort_bbs_postorder): New function. Use it...
19200 (dom_walker::walk): ...here to optimize common cases.
19202 2017-07-25 Martin Liska <mliska@suse.cz>
19205 * ipa-visibility.c (function_and_variable_visibility): Make the
19206 redirection just on target that supports aliasing.
19207 Fix GNU coding style.
19209 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
19212 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
19214 * config/arm/rtems.h: Likewise.
19215 * config/bfin/rtems.h: Likewise.
19216 * config/i386/rtemself.h: Likewise.
19217 * config/lm32/rtems.h: Likewise.
19218 * config/m32c/rtems.h: Likewise.
19219 * config/m68k/rtemself.h: Likewise.
19220 * config/microblaze/rtems.h: Likewise.
19221 * config/mips/rtems.h: Likewise.
19222 * config/moxie/rtems.h: Likewise.
19223 * config/nios2/rtems.h: Likewise.
19224 * config/powerpcspe/rtems.h: Likewise.
19225 * config/rs6000/rtems.h: Likewise.
19226 * config/rtems.h: Likewise.
19227 * config/sh/rtems.h: Likewise.
19228 * config/sh/rtemself.h: Likewise.
19229 * config/sparc/rtemself.h: Likewise.
19231 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19234 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
19235 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
19236 * tree-ssa-structalias.c (alias_get_name): Same.
19238 2017-07-25 Bin Cheng <bin.cheng@arm.com>
19241 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
19242 instructions if no du chain is found.
19244 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19246 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
19248 2017-07-25 Richard Biener <rguenther@suse.de>
19250 PR middle-end/81505
19251 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
19254 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19256 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
19257 upper-regs options.
19258 (ISA_2_7_MASKS_SERVER): Likewise.
19259 (ISA_3_0_MASKS_IEEE): Likewise.
19260 (OTHER_P8_VECTOR_MASKS): Likewise.
19261 (OTHER_VSX_VECTOR_MASKS): Likewise.
19262 (POWERPC_MASKS): Likewise.
19263 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
19264 duplicate list of options.
19265 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
19266 explicit -mupper-regs options.
19267 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
19268 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
19269 alias for -mupper-regs-df.
19270 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19271 (rs6000_init_hard_regno_mode_ok): Likewise.
19272 (rs6000_option_override_internal): Likewise.
19273 (rs6000_opt_masks): Likewise.
19274 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
19275 options in terms of whether -mvsx or -mpower8-vector was used.
19276 (TARGET_UPPER_REGS_DI): Likewise.
19277 (TARGET_UPPER_REGS_SF): Likewise.
19278 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
19279 -mupper-regs-* options.
19281 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19283 * passes.c (emergency_dump_function): Print some empty lines and a
19284 header before the RTL dump.
19286 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19288 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
19290 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
19293 * config/aarch64/aarch64.c (aarch64_classify_symbol):
19294 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
19296 2017-07-24 Carl Love <cel@us.ibm.com>
19298 * config/rs6000/rs6000-c.c: Add support for built-in functions
19299 vector float vec_extract_fp32_from_shorth (vector unsigned short);
19300 vector float vec_extract_fp32_from_shortl (vector unsigned short);
19301 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
19302 vec_extract_fp_from_shortl): Add defines for the two builtins.
19303 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
19304 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
19306 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
19307 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
19308 * doc/extend.texi: Update the built-in documentation file for the
19309 new built-in function.
19311 2017-07-24 Jakub Jelinek <jakub@redhat.com>
19314 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
19316 * doc/generic.texi: Likewise.
19317 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19318 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19320 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
19322 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
19323 (aarch64_mls_elt_merge<mode>): Likewise.
19325 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
19327 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19328 having __cxa_atexit.
19330 2017-07-23 Michael Collison <michael.collison@arm.com>
19332 * config/arm/arm.c (arm_option_override): Deprecate
19333 use of -mstructure-size-boundary.
19334 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
19335 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
19337 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19340 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19341 Reduce cost estimate for direct moves.
19343 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
19346 * config/i386/i386.c (ix86_option_override_internal): Disable
19347 BMI, BMI2 and TBM instructions for -m16.
19349 2017-07-21 Carl Love <cel@us.ibm.com>
19351 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19352 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19353 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19354 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19355 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19356 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19357 VMULOSW): New enum "unspec" values.
19358 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19359 altivec_vmulosw): New patterns.
19360 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19361 VMULOSW): Add definitions.
19363 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
19365 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
19366 (qdf24xx): Likewise.
19367 * config/aarch64/aarch64-options-extensions.def (rdma); New.
19368 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
19369 (AARCH64_FL_V8_1): Renumber.
19370 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
19371 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
19372 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
19373 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
19374 rdma to feature modifiers list.
19376 2017-07-21 Yury Gribov <tetra2005@gmail.com>
19378 PR middle-end/56727
19379 * ipa-visibility (function_and_variable_visibility): Convert
19380 recursive PLT call to direct call if appropriate.
19382 2017-07-21 Andrew Pinski <apinski@cavium.com>
19384 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
19385 operand 1 to see if the types precision matches.
19386 * fold-const.c (operand_equal_p): Likewise.
19388 2017-07-21 Richard Biener <rguenther@suse.de>
19390 PR tree-optimization/81303
19391 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
19392 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
19393 (vect_peeling_hash_get_lowest_cost): Adjust.
19394 (vect_enhance_data_refs_alignment): Likewise. Use
19395 vect_get_peeling_costs_all_drs to compute the penalty for no
19396 peeling to match up costs.
19398 2017-07-21 Richard Biener <rguenther@suse.de>
19400 PR tree-optimization/81500
19401 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
19402 we didn't identify a reduction path.
19404 2017-07-21 Tom de Vries <tom@codesourcery.com>
19405 Cesar Philippidis <cesar@codesourcery.com>
19407 PR gcov-profile/81442
19408 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
19411 2017-07-21 Tom de Vries <tom@codesourcery.com>
19414 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
19416 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
19417 nvptx_override_options_after_change.
19419 2017-07-21 Ulrich Drepper <drepper@redhat.com>
19421 * dwarf2out.c (output_file_names): Avoid double testing for
19422 dwarf_version >= 5.
19424 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
19426 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
19428 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19430 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
19432 (try_crossjump_to_edge): Do not punt on partitioned functions.
19434 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19436 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
19437 Put all BBs reachable only via paths crossing cold region to cold
19439 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
19441 2016-07-21 Richard Biener <rguenther@suse.de>
19443 PR tree-optimization/81303
19444 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
19445 into account prologue and epilogue iterations when raising
19446 min_profitable_iters to sth at least covering one vector iteration.
19448 2017-07-21 Tamar Christina <tamar.christina@arm.com>
19450 * config/arm/arm.c (arm_test_cpu_arch_dat):
19453 2017-07-20 Nathan Sidwell <nathan@acm.org>
19455 Remove TYPE_METHODS.
19456 * tree.h (TYPE_METHODS): Delete.
19457 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
19458 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
19459 (dbxout_type_methods): Scan TYPE_FIELDS.
19460 (dbxout_type): Don't check TYPE_METHODS here.
19461 * function.c (use_register_for_decl): Always ignore register for
19462 class types when not optimizing.
19463 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
19464 * tree.c (free_lang_data_in_type): Stitch out member functions and
19465 templates from TYPE_FIELDS.
19466 (build_distinct_type_copy, verify_type_variant,
19467 verify_type): Member fns are on TYPE_FIELDS.
19468 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
19469 * tree-pretty-print.c (dump_generic_node): Likewise.
19471 2017-07-20 Jakub Jelinek <jakub@redhat.com>
19474 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
19475 V2TImode and V4TImode.
19476 (ix86_expand_vector_extract): Likewise.
19477 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
19478 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
19479 (ssescalarmode): Handle V4TImode and V2TImode.
19480 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
19481 (*vec_extractv2ti, *vec_extractv4ti): New insns.
19482 (VEXTRACTI128_MODE): New mode iterator.
19483 (splitter for *vec_extractv?ti first element): New.
19484 (VEC_INIT_MODE): New mode iterator.
19485 (vec_init<mode>): Consolidate 3 expanders into one using
19486 VEC_INIT_MODE mode iterator.
19488 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19490 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
19491 non_spilled_static_chain_regno_p.
19493 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19495 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
19497 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
19499 * bb-reorder.c (connect_traces): Allow copying of blocks within
19502 2017-07-20 Richard Biener <rguenther@suse.de>
19504 * gimple.h (gimple_phi_result): Add gphi * overload.
19505 (gimple_phi_result_ptr): Likewise.
19506 (gimple_phi_arg): Likewise. Adjust index assert to only
19507 allow actual argument accesses rather than all slots available
19509 (gimple_phi_arg_def): Add gphi * overload.
19510 * tree-phinodes.c (make_phi_node): Initialize only actual
19512 (resize_phi_node): Clear memory not covered by old node,
19513 do not initialize excess argument slots.
19514 (reserve_phi_args_for_new_edge): Initialize new argument slot
19517 2017-07-20 Bin Cheng <bin.cheng@arm.com>
19519 PR tree-optimization/81388
19521 2016-07-21 Bin Cheng <bin.cheng@arm.com>
19523 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
19524 by removing computation of may_be_zero.
19526 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19527 Tom de Vries <tom@codesourcery.com>
19529 PR middle-end/81030
19530 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
19531 when gimple level profile disagrees with what RTL expander did.
19533 2017-07-20 Richard Biener <rguenther@suse.de>
19535 PR tree-optimization/61171
19536 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
19537 (vect_analyze_stmt): Add slp instance parameter.
19538 (vectorizable_reduction): Likewise.
19539 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
19540 (vect_is_simple_reduction): Deal with chains not detected
19541 as SLP reduction chain, specifically not properly associated
19542 chains containing a mix of plus/minus.
19543 (get_reduction_op): Remove.
19544 (get_initial_defs_for_reduction): Simplify, pass in whether
19545 this is a reduction chain, pass in the SLP node for the PHIs.
19546 (vect_create_epilog_for_reduction): Get the SLP instance as
19548 (vectorizable_reduction): Get the SLP instance as arg.
19549 During analysis remember the SLP node with the PHIs in the
19550 instance. Simplify getting at the vectorized reduction PHIs.
19551 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
19552 through SLP instance.
19553 (vect_slp_analyze_operations): Likewise.
19554 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
19555 (vect_transform_stmt): Likewise.
19557 2017-07-20 Tom de Vries <tom@codesourcery.com>
19559 PR tree-optimization/81489
19560 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
19561 read of phi arg location to before loop that modifies phi.
19563 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19565 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
19568 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19570 PR middle-end/81331
19571 * except.c (execute): Fix ordering issue.
19573 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19575 PR rtl-optimization/81423
19576 * combine.c (make_compound_operation_int): Don't try to optimize
19577 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
19579 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19581 PR rtl-optimization/81423
19582 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
19583 with a constant that is -1 in the truncated to mode.
19585 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19587 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
19588 (determine_unlikely_bbs): ... here.
19589 * predict.h (propagate_unlikely_bbs_forward): Declare.
19590 * cfgexpand.c (pass_expand::execute): Use it.
19591 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
19593 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
19594 propagate_unlikely_bbs_forward.
19596 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19598 PR middle-end/81331
19599 * except.c (maybe_add_nop_after_section_switch): New function.
19602 2017-07-19 Tom de Vries <tom@codesourcery.com>
19604 * gimple.h (gimple_phi_set_arg): Make assert more strict.
19606 2017-07-19 Tom de Vries <tom@codesourcery.com>
19608 * gimple.h (gimple_phi_arg): Make assert more strict.
19610 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
19612 * config.gcc (powerpc*-*-*): Add mmintrin.h.
19613 * config/rs6000/mmintrin.h: New file.
19614 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
19616 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19618 PR tree-optimization/81346
19619 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
19621 2017-07-19 Tom de Vries <tom@codesourcery.com>
19623 * config/nvptx/nvptx.md (VECIM): Add V2DI.
19625 2017-07-19 Tom de Vries <tom@codesourcery.com>
19627 * config/nvptx/nvptx-modes.def: Add V2DImode.
19628 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
19629 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
19630 (nvptx_output_mov_insn): Handle lack of mov.b128.
19631 (nvptx_print_operand): Handle 'H' and 'L' codes.
19632 (nvptx_vector_mode_supported): Allow V2DImode.
19633 (nvptx_preferred_simd_mode): New function.
19634 (nvptx_data_alignment): New function.
19635 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
19636 nvptx_preferred_simd_mode.
19637 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
19639 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
19641 2017-07-19 Tom de Vries <tom@codesourcery.com>
19643 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
19644 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
19645 (nvptx_vector_mode_supported): New function. Allow V2SImode.
19646 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
19647 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
19648 (mov<VECIM>_insn): New define_insn.
19649 (define_expand "mov<VECIM>): New define_expand.
19651 2017-07-19 Tom de Vries <tom@codesourcery.com>
19653 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
19655 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19657 PR tree-optimization/81346
19658 * fold-const.h (fold_div_compare, range_check_type): Declare.
19659 * fold-const.c (range_check_type): New function.
19660 (build_range_check): Use range_check_type.
19661 (fold_div_compare): No longer static, rewritten into
19662 a match.pd helper function.
19663 (fold_comparison): Don't call fold_div_compare here.
19664 * match.pd (X / C1 op C2): New optimization using fold_div_compare
19665 as helper function.
19667 2017-07-19 Nathan Sidwell <nathan@acm.org>
19669 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
19670 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
19671 * tree.c (find_decls_types_r, verify_type): Use
19672 TYPE_{MIN,MAX}_VALUE_RAW.
19673 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
19674 (hash_tree): Likewise.
19675 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19677 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19680 2017-07-18 Tom de Vries <tom@codesourcery.com>
19682 PR middle-end/81464
19683 * omp-expand.c (expand_omp_for_static_chunk): Handle
19684 equal-argument loop exit phi.
19686 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
19689 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
19690 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
19691 operand 2 predicate.
19692 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
19693 operand 2 predicate.
19694 (ror,rol -> rorx splitters): Use const_int_operand as
19695 operand 2 predicate.
19697 2017-06-18 Richard Biener <rguenther@suse.de>
19699 PR tree-optimization/81410
19700 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
19701 the gap in the ! slp_perm SLP case after each group.
19703 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19705 PR middle-end/81463
19706 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
19709 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19711 PR middle-end/81462
19712 * predict.c (set_even_probabilities): Cleanup; do not affect
19713 probabilities that are already known.
19714 (combine_predictions_for_bb): Call even when count is set.
19716 2017-07-18 Nathan Sidwell <nathan@acm.org>
19718 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
19721 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19724 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
19725 optimization for loop niter analysis.
19727 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
19730 * config/avr/avr.c (avr_optimize_casesi): Don't use
19731 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
19733 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
19735 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
19736 body_cost_vec from _vect_peel_extended_info.
19737 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
19738 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
19741 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19743 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
19745 2017-07-18 Richard Biener <rguenther@suse.de>
19747 PR tree-optimization/80620
19748 PR tree-optimization/81403
19749 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
19750 info when re-using a VN table entry.
19752 2017-07-18 Richard Biener <rguenther@suse.de>
19754 PR tree-optimization/81418
19755 * tree-vect-loop.c (vectorizable_reduction): Properly compute
19756 vectype_in. Verify that with lane-reducing reduction operations
19757 we have a single def-use cycle.
19759 2017-07-17 Carl Love <cel@us.ibm.com>
19761 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
19763 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19764 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19765 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19766 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19767 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19768 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19769 VMULOSW): New enum "unspec" values.
19770 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19771 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19772 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19773 altivec_vmulosw): New patterns.
19774 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19775 VMULOSW): Add definitions.
19776 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
19778 * config/alpha/alpha.c: Include predict.h.
19780 2017-07-17 Yury Gribov <tetra2005@gmail.com>
19782 * tree-vrp.c (compare_assert_loc): Fix comparison function
19783 to return predictable results.
19785 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19787 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
19789 (subdi3): Likewise.
19790 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
19791 * doc/invoke.texi (mexpand-adddi): Update text.
19793 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19795 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
19796 that also clobbers the CC register. The old expand code is moved
19798 (*arc_clzsi2): ... here.
19799 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
19800 the CC register. The old expand code is moved to ...
19801 (arc_ctzsi2): ... here.
19803 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19805 * config/arc/arc.opt (mindexed-loads): Use initial value
19806 TARGET_INDEXED_LOADS_DEFAULT.
19807 (mauto-modify-reg): Use initial value
19808 TARGET_AUTO_MODIFY_REG_DEFAULT.
19809 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19810 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19811 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19812 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19814 2017-07-17 Martin Liska <mliska@suse.cz>
19817 * opts.c (finish_options): Do not allow -fgnu-tm
19818 w/ -fsanitize={kernel-,}address. Say sorry.
19820 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19823 * tree-loop-distribution.c (classify_partition): Only assert on
19824 numer of iterations.
19825 (merge_dep_scc_partitions): Delete prameter. Update function call.
19826 (distribute_loop): Remove code handling loop with unknown niters.
19827 (pass_loop_distribution::execute): Skip loop with unknown niters.
19829 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19832 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
19833 function sort_partitions_by_post_order.
19835 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19837 PR tree-optimization/81374
19838 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
19839 the max index of basic blocks, rather than number of basic blocks.
19841 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19843 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
19845 (arc_legitimate_pic_operand_p): Likewise.
19846 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
19848 (arc_needs_pcl_p): Likewise.
19849 (arc_legitimate_pc_offset_p): Likewise.
19850 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
19851 function is also used in constrains.md.
19852 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
19853 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
19854 PLUS. Only return true/false in known cases, otherwise assert.
19855 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
19856 is already called in arc_legitimate_constant_p.
19857 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
19859 (LEGITIMATE_PIC_OPERAND_P): Use
19860 arc_raw_symbolic_reference_mentioned_p function.
19861 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
19866 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19867 Andrew Burgess <andrew.burgess@embecosm.com>
19869 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
19870 (arc_return_address_register): New function.
19871 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
19872 (arc_handle_fndecl_attribute): Add naked attribute.
19873 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19874 (TARGET_WARN_FUNC_RETURN): Likewise.
19875 (arc_allocate_stack_slots_for_args): New function.
19876 (arc_warn_func_return): Likewise.
19877 (machine_function): Change type fn_type.
19878 (arc_compute_function_type): Consider new naked function type,
19879 change function return type.
19880 (arc_must_save_register): Adapt to handle new
19881 arc_compute_function_type's return type.
19882 (arc_expand_prologue): Likewise.
19883 (arc_expand_epilogue): Likewise.
19884 (arc_return_address_regs): Delete.
19885 (arc_return_address_register): New function.
19886 (arc_epilogue_uses): Use above function.
19887 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
19888 (arc_function_type): Change encoding, add naked type.
19889 (ARC_INTERRUPT_P): Change to handle the new encoding.
19890 (ARC_FAST_INTERRUPT_P): Likewise.
19891 (ARC_NORMAL_P): Define.
19892 (ARC_NAKED_P): Likewise.
19893 (arc_compute_function_type): Delete prototype.
19894 * config/arc/arc.md (in_ret_delay_slot): Use
19895 arc_return_address_register function.
19896 (simple_return): Likewise.
19897 (p_return_i): Likewise.
19899 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19901 PR tree-optimization/81428
19902 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
19903 can't be built for those types.
19905 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19907 Remove stuff dead since r239246.
19909 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
19910 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
19911 (avr_inform_devices): Remove dead stuff.
19913 2017-07-17 Tamar Christina <tamar.christina@arm.com>
19915 * config/arm/arm_neon.h: Fix softp typo.
19917 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19919 PR tree-optimization/81365
19920 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
19921 aggregate moves onto bb predecessor edges, make sure there are no
19922 loads that could alias the lhs in between the start of bb and the
19925 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19928 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
19929 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
19930 [LSHIFTRT, outer_code = TRUNCATE]: Same.
19932 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19934 PR tree-optimization/81396
19935 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
19936 (init_symbolic_number): Initialize it to 1.
19937 (perform_symbolic_merge): Add n_ops from both operands into the new
19939 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
19940 without base_addr as useless if they need more than one operation.
19941 (bswap_replace): Handle !bswap case for NULL base_addr.
19943 2017-07-17 Tom de Vries <tom@codesourcery.com>
19946 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
19949 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
19951 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
19952 conditional builtin define __FIX_LEON3FT_B2BST.
19954 2017-07-17 Daniel Cederman <cederman@gaisler.com>
19956 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
19957 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
19960 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19962 PR rtl-optimization/81424
19963 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
19964 to remove potential trapping from operands if -fnon-call-exceptions.
19966 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19968 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
19969 profile_proability for scalling.
19970 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
19972 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19974 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
19976 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19978 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
19979 fixpoint arithmetics.
19981 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19983 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
19984 fixpoint arithmetics.
19986 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19988 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
19989 fixpoint arithmetics.
19991 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19993 * profile-count.h (profile_probability::from_reg_br_prob_note,
19994 profile_probability::to_reg_br_prob_note): New functions.
19995 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
19996 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
19997 * predict.c (probability_reliable_p): Update.
19998 (edge_probability_reliable_p): Update.
19999 (br_prob_note_reliable_p): Update.
20000 (invert_br_probabilities): Update.
20001 (add_reg_br_prob_note): New function.
20002 (combine_predictions_for_insn): Update.
20003 * asan.c (asan_clear_shadow): Update.
20004 * cfgbuild.c (compute_outgoing_frequencies): Update.
20005 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20006 (update_br_prob_note): Update.
20007 (rtl_verify_edges): Update.
20008 (purge_dead_edges): Update.
20009 (fixup_reorder_chain): Update.
20010 * emit-rtl.c (try_split): Update.
20011 * ifcvt.c (cond_exec_process_insns): Update.
20012 (cond_exec_process_if_block): Update.
20013 (dead_or_predicable): Update.
20014 * internal-fn.c (expand_addsub_overflow): Update.
20015 (expand_neg_overflow): Update.
20016 (expand_mul_overflow): Update.
20017 * loop-doloop.c (doloop_modify): Update.
20018 * loop-unroll.c (compare_and_jump_seq): Update.
20019 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20020 * predict.h: Update.
20021 * reorg.c (mostly_true_jump): Update.
20023 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20024 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20025 * config/arc/arc.c: (emit_unlikely_jump): Update.
20026 * config/arm/arm.c: (emit_unlikely_jump): Update.
20027 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20028 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20029 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20030 (ix86_print_operand): Update.
20031 (ix86_split_fp_branch): Update.
20032 (predict_jump): Update.
20033 * config/ia64/ia64.c (ia64_print_operand): Update.
20034 * config/mmix/mmix.c (mmix_print_operand): Update.
20035 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20036 (rs6000_expand_split_stack_prologue): Update.
20037 * config/rs6000/rs6000.c: Update.
20038 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20039 (s390_expand_vec_movstr): Update.
20040 (s390_expand_cs_tdsi): Update.
20041 (s390_expand_split_stack_prologue): Update.
20042 * config/sh/sh.c (sh_print_operand): Update.
20043 (expand_cbranchsi4): Update.
20044 (expand_cbranchdi4): Update.
20045 * config/sparc/sparc.c (output_v9branch): Update.
20046 * config/spu/spu.c (get_branch_target): Update.
20047 (ea_load_store_inline): Update.
20048 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20049 * config/tilepro/tilepro.c: Update.
20051 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20053 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20054 (gimplify_save_expr): Likewise.
20056 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20058 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20060 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20062 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20064 * timevar.def (TV_IPA_FNSUMMARY): Define.
20066 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20068 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20069 to back store errata sensitive sequence from being generated.
20070 (sqrtdf2_fix): Likewise.
20072 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20074 * tree-ssa-threadupdate.c (compute_path_counts,
20075 update_joiner_offpath_counts): Use profile_probability.
20077 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20080 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20082 * config/arm/arm-c.c (arm_cpu_builtins): Define
20083 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20085 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20087 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20088 array entries to represent __ieee128 versions of the
20089 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20090 scalar_extract_sig, and scalar_insert_exp built-in functions.
20091 (altivec_resolve_overloaded_builtin): Add special case handling
20092 for the __builtin_scalar_insert_exp function, as represented by
20093 the P9V_BUILTIN_VEC_VSIEDP constant.
20094 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20095 exponent support for __ieee128 argument.
20096 (VSESQP): Add scalar extract signature support for __ieee128
20098 (VSTDCNQP): Add scalar test negative support for __ieee128
20100 (VSIEQP): Add scalar insert exponent support for __int128 argument
20101 with __ieee128 result.
20102 (VSIEQPF): Add scalar insert exponent support for __ieee128
20103 argument with __ieee128 result.
20104 (VSTDCQP): Add scalar test data class support for __ieee128
20106 (VSTDCNQP): Add overload support for scalar test negative with
20107 __ieee128 argument.
20108 (VSTDCQP): Add overload support for scalar test data class
20109 __ieee128 argument.
20110 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20111 UNSPEC_VSX_SXSIGDP.
20112 (UNSPEC_VSX_SIEXPQP): New constant.
20113 (xsxexpqp): New insn for VSX scalar extract exponent quad
20115 (xsxsigqp): New insn for VSX scalar extract significand quad
20117 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20118 precision with floating point argument.
20119 (xststdcqp): New expand for VSX scalar test data class quad
20121 (xststdcnegqp): New expand for VSX scalar test negative quad
20123 (xststdcqp): New insn to match expansions for VSX scalar test data
20124 class quad precision and VSX scalar test negative quad precision.
20125 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20126 special case operand checking to enforce that second operand of
20127 VSX scalar test data class with quad precision argument is a 7-bit
20129 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20130 prototypes and descriptions of __ieee128 versions of
20131 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20132 scalar_test_data_class, and scalar_test_neg built-in functions.
20134 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20136 PR tree-optimization/81162
20137 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20138 replace a negate with an add.
20140 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20142 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20144 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20146 * config/arm/arm-c.c (arm_cpu_builtins): Define
20147 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20149 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20151 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20152 (armv8-r): Set ARM Cortex-R52 as default CPU.
20153 * config/arm/arm-tables.opt: Regenerate.
20154 * config/arm/arm-tune.md: Regenerate.
20155 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20157 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20158 extension for -mcpu=cortex-r52.
20160 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20162 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20163 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20164 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20165 (fp-armv8): Define it as FP_ARMv8 only.
20166 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20167 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20169 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20170 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20171 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20172 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20173 than TARGET_FPU_ARMV8.
20174 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20175 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20176 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20178 * config/arm/neon.md (neon_vrint): Likewise.
20179 (neon_vcvt): Likewise.
20180 (neon_<fmaxmin_op><mode>): Likewise.
20181 (<fmaxmin><mode>3): Likewise.
20182 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20183 * config/arm/predicates.md (arm_cond_move_operator): Check against
20184 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20186 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20188 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20189 to top of function.
20191 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20193 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20194 loop in comment with memset.
20196 2017-07-14 Martin Liska <mliska@suse.cz>
20198 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20199 * dwarf2out.c (is_java): Remove the function.
20200 (output_pubname): Remove usage of the function.
20201 (lower_bound_default): Remove usage of DW_LANG_Java.
20202 (gen_compile_unit_die): Likewise.
20203 * gcc.c: Remove compiler defaults for .java and .zip files.
20204 * gimple-expr.c (remove_suffix): Change as there's no longer
20205 extension than 4-letter one.
20206 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20207 (gimplify_save_expr): Likewise.
20208 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20209 as it's possible even for other languages than Java.
20210 * langhooks.h (struct lang_hooks): Remove Java from a comment.
20211 * lto-opts.c (lto_write_options): Remove reference to Java.
20212 * opts.c (strip_off_ending): Update file extension handling.
20213 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
20214 * tree-eh.c (lower_resx): Likewise.
20215 * tree.c (free_lang_data_in_type): Remove dead code.
20216 (find_decls_types_r): Likewise.
20217 (build_common_builtin_nodes): Remove Java from a comment.
20218 (verify_type): Remove dead code.
20219 * varasm.c (assemble_external): Remove Java from a comment.
20221 2017-07-14 Martin Liska <mliska@suse.cz>
20223 * opts.c (finish_options): Add quotes.
20224 (common_handle_option): Likewise.
20226 2017-07-14 Martin Liska <mliska@suse.cz>
20228 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
20229 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
20230 Remove N_SO_PASCAL.
20231 * dwarf2out.c (lower_bound_default): Do not handle
20233 (gen_compile_unit_die): Likewise.
20234 * gcc.c: Remove default extension binding for GNU Pascal.
20235 * stmt.c: Remove Pascal language from a comment.
20236 * xcoffout.c: Likewise.
20238 2017-07-13 David Malcolm <dmalcolm@redhat.com>
20241 * diagnostic-show-locus.c (fixit_cmp): New function.
20242 (layout::layout): Sort m_fixit_hints.
20243 (column_range::column_range): Assert that the values are valid.
20244 (struct char_span): New struct.
20245 (correction::overwrite): New method.
20246 (struct source_line): New struct.
20247 (line_corrections::add_hint): Add assertions. Reimplement memcpy
20248 calls in terms of classes source_line and char_span, and
20249 correction::overwrite.
20250 (selftest::test_overlapped_fixit_printing_2): New function.
20251 (selftest::diagnostic_show_locus_c_tests): Call it.
20253 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
20255 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
20256 early if there is no lhs.
20258 2017-07-13 Martin Liska <mliska@suse.cz>
20260 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
20261 (gen_reference_type_die): Likewise.
20262 * stor-layout.c: Remove Pascal-related comment.
20264 2017-07-13 Martin Liska <mliska@suse.cz>
20266 * opts.c (finish_options): Add quotes to error messages.
20267 (parse_sanitizer_options): Likewise.
20269 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
20271 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
20273 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
20275 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
20277 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
20279 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
20281 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
20283 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
20286 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
20287 provides the hardware capability bits, define the macro
20288 __BUILTIN_CPU_SUPPORTS__.
20289 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
20290 if GLIBC does not provide the hardware capability bits. Add a
20291 gcc_unreachable call if the built-in cpu function is neither
20292 __builtin_cpu_is nor __builtin_cpu_supports.
20293 (rs6000_get_function_versions_dispatcher): Change the warning
20294 that an old GLIBC is used which does not export the capability
20295 bits to be an error.
20296 * doc/extend.texi (target_clones attribute): Document the
20297 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
20298 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
20299 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
20300 the macros defined by GCC if the newer GLIBC is available.
20302 2017-07-12 Jeff Law <law@redhat.com>
20304 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
20305 remaining includes slightly.
20306 * config/riscv/riscv-builtins.c: Include profile-count.h.
20308 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20311 * config/avr/avr.c (avr_set_current_function): In diagnostic
20312 messages: Quote keywords and (parts of) identifiers.
20313 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
20316 2017-07-12 Carl Love <cel@us.ibm.com>
20318 * config/rs6000/rs6000-c.c: Add support for built-in functions
20319 vector bool char vec_revb (vector bool char);
20320 vector bool short vec_revb (vector short char);
20321 vector bool int vec_revb (vector bool int);
20322 vector bool long long vec_revb (vector bool long long);
20323 * doc/extend.texi: Update the built-in documentation file for the
20324 new built-in functions.
20326 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20328 * config/s390/s390.md: Remove movcc splitter.
20330 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20332 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
20333 load/store on condition.
20335 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20338 * config/avr/avr.c (avr_encode_section_info)
20339 [progmem && !TREE_READONLY]: Error if progmem object needs
20342 2017-07-11 Michael Collison <michael.collison@arm.com>
20344 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
20347 2017-07-11 Carl Love <cel@us.ibm.com>
20349 * config/rs6000/rs6000-c.c: Add support for builtins
20350 vector unsigned int vec_parity_lsbb (vector signed int);
20351 vector unsigned int vec_parity_lsbb (vector unsigned int);
20352 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
20353 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
20354 vector unsigned long long vec_parity_lsbb (vector signed long long);
20355 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
20356 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
20357 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
20358 * doc/extend.texi: Update the built-in documentation file for the
20359 new built-in functions.
20361 2017-07-11 David Malcolm <dmalcolm@redhat.com>
20363 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
20364 (layout::m_primary_loc): New field.
20365 (layout::layout): Initialize new field. Move location filtering
20366 logic from here to...
20367 (layout::maybe_add_location_range): ...this new method. Add
20368 support for filtering to just the lines already specified by other
20370 (layout::will_show_line_p): New method.
20371 (gcc_rich_location::add_location_if_nearby): New method.
20372 (selftest::test_add_location_if_nearby): New test function.
20373 (selftest::diagnostic_show_locus_c_tests): Call it.
20374 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
20377 2017-07-11 Tom de Vries <tom@codesourcery.com>
20379 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
20380 (bb_first_real_insn): New function.
20381 (nvptx_single): Add extra initialization of broadcasted condition
20384 2017-07-11 Nathan Sidwell <nathan@acm.org>
20386 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
20388 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
20390 * doc/extend.texi (AVR Function Attributes): Remove weblink to
20391 Binutils doc as TEXI will mess them up.
20392 * doc/invoke.texi (AVR Options): Same here.
20394 2017-07-11 Daniel Cederman <cederman@gaisler.com>
20396 * config/sparc/sparc.opt (mfix-ut700): New option.
20397 (mfix-gr712rc): Likewise.
20398 (sparc_fix_b2bst): New variable.
20399 * doc/invoke.texi (SPARC options): Document them.
20400 (ARM options): Fix warnings.
20401 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
20402 instructions to prevent sequences that can trigger the store-store
20403 errata for certain LEON3FT processors.
20404 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
20405 (sparc_option_override): Set sparc_fix_b2bst appropriately.
20406 * config/sparc/sparc.md (fix_b2bst): New attribute.
20407 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
20409 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
20412 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
20414 (*rsqrtsf2_sse): Ditto.
20416 (div<mode>3): Macroize insn from divdf3 and divsf3
20417 using MODEF mode iterator.
20419 2017-07-10 Martin Sebor <msebor@redhat.com>
20421 PR tree-optimization/80397
20422 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
20423 instead of testing for equality to INTEGER_TYPE.
20425 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
20427 * config.gcc: Remove uclibc from arc target spec.
20429 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
20431 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
20433 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20436 * lto-wrapper.c (remove_option): New function.
20437 (merge_and_complain): Merge PIC/PIE options more realistically.
20439 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20441 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
20445 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
20446 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
20447 * config.in: Regenerate.
20448 * configure: Regenerate.
20449 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
20450 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
20451 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
20452 (TARGET_GASISR_PROLOGUES): ...target mask.
20453 * common/config/avr/avr-common.c
20454 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
20455 Set -mgas-isr-prologues.
20456 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
20457 INSERT_PASS_BEFORE for it.
20458 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
20459 * config/avr/avr.c (avr_option_override)
20460 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
20461 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
20462 (avr_attribute_table) <no_gccisr>: Add new function attribute.
20463 (avr_set_current_function) <is_no_gccisr>: Init machine field.
20464 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
20466 (make_avr_pass_pre_proep): New function.
20467 (emit_push_sfr) <treg>: Add argument to function and use it
20468 instead of TMP_REG.
20469 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
20470 and set machine->gasisr.yes.
20471 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
20472 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
20473 __gcc_isr.n_pushed to .L__stack_usage.
20474 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
20475 (avr_asm_final_postscan_insn): ...this new static function.
20476 * config/avr/avr.h (machine_function)
20477 <is_no_gccisr, use_L__stack_usage>: New fields.
20478 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
20479 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
20480 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
20481 (gasisr, *gasisr): New expander and insn.
20482 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20483 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
20484 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
20486 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
20488 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
20491 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20493 Move jump-tables out of .text again.
20496 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
20497 (ASM_OUTPUT_ADDR_VEC): New function.
20498 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
20499 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
20500 INSN_ADDRESSes as asm comment.
20501 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
20502 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
20503 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
20504 * config/avr/avr.md (*tablejump): Adjust comment.
20505 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
20506 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
20508 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
20509 (avr_output_addr_vec): New proto.
20510 (avr_log_t) <insn_addresses>: New field.
20512 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
20515 * config/i386/i386.c (ix86_function_arg_advance): Set
20516 outgoing_args_on_stack to true if there are outgoing arguments
20518 (ix86_function_arg): Likewise.
20519 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
20520 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
20521 * config/i386/i386.h (machine_function): Add
20522 outgoing_args_on_stack.
20524 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
20526 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20527 supporting pthreds.
20528 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
20530 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
20532 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
20533 (REAL_H): Remove $(MACHMODE_H).
20534 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
20536 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
20537 $(MACHMODE_H) and double-int.h.
20538 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
20540 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
20543 2017-07-07 Andrew Pinski <apinski@cavium.com>
20545 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
20546 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
20548 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20550 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
20551 Add warning if GCC was not configured to link against a GLIBC that
20552 exports the hardware capability bits.
20553 (make_resolver_func): Make resolver function private and not a
20554 COMDAT function. Create the name with clone_function_name instead
20555 of make_unique_name.
20558 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
20559 correct operand in doing the split.
20561 2017-07-07 Carl Love <cel@us.ibm.com>
20563 * config/rs6000/rs6000-c: Add support for built-in function
20564 vector unsigned short vec_pack_to_short_fp32 (vector float,
20566 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
20567 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
20568 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
20569 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
20570 (convert_4f32_8i16): Add define_expand.
20571 * doc/extend.texi: Update the built-in documentation file for the
20572 new built-in function.
20574 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20576 * config/sparc/m8.md: New file.
20577 * config/sparc/sparc.md: Include m8.md.
20579 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20581 * config/sparc/sparc.opt: New option -mvis4b.
20582 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
20583 (sparc_option_override): Handle VIS4B.
20584 (enum sparc_builtins): Define
20585 SPARC_BUILTIN_DICTUNPACK{8,16,32},
20586 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
20587 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
20588 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
20589 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
20590 (check_constant_argument): New function.
20591 (sparc_vis_init_builtins): Define builtins
20592 __builtin_vis_dictunpack{8,16,32},
20593 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
20594 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
20595 __builtin_vis_fpcmpde{8,16,32}shl and
20596 __builtin_vis_fpcmpur{8,16,32}shl.
20597 (sparc_expand_builtin): Check that the constant operands to
20598 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
20599 constant and in range.
20600 * config/sparc/sparc-c.c (sparc_target_macros): Handle
20602 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
20603 (SPARC_IMM5_P): Likewise.
20604 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
20605 (enabled): Handle vis4b.
20606 (UNSPEC_DICTUNPACK): New unspec.
20607 (UNSPEC_FPCMPSHL): Likewise.
20608 (UNSPEC_FPUCMPSHL): Likewise.
20609 (UNSPEC_FPCMPDESHL): Likewise.
20610 (UNSPEC_FPCMPURSHL): Likewise.
20611 (cpu_feature): New CPU feature `vis4b'.
20612 (dictunpack{8,16,32}): New insns.
20613 (FPCSMODE): New mode iterator.
20614 (fpcscond): New code iterator.
20615 (fpcsucond): Likewise.
20616 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
20617 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
20618 (fpcmpde{8,16,32}{si,di}shl): Likewise.
20619 (fpcmpur{8,16,32}{si,di}shl): Likewise.
20620 * config/sparc/constraints.md: Define constraints `q' for unsigned
20621 2-bit integer constants and `t' for unsigned 5-bit integer
20623 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
20625 (imm5_operand_dictunpack16): Likewise.
20626 (imm5_operand_dictunpack32): Likewise.
20627 (imm2_operand): Likewise.
20628 * doc/invoke.texi (SPARC Options): Document -mvis4b.
20629 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
20630 ditunpack* and fpcmp*shl builtins.
20632 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20634 * config.gcc: Handle m8 in --with-{cpu,tune} options.
20635 * config.in: Add HAVE_AS_SPARC6 define.
20636 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
20638 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
20640 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
20641 (CPP_CPU_SPEC): Handle m8.
20642 (ASM_CPU_SPEC): Likewise.
20643 * config/sparc/sparc-opts.h (enum processor_type): Add
20645 * config/sparc/sparc.c (m8_costs): New struct.
20646 (sparc_option_override): Handle TARGET_CPU_m8.
20647 (sparc32_initialize_trampoline): Likewise.
20648 (sparc64_initialize_trampoline): Likewise.
20649 (sparc_issue_rate): Likewise.
20650 (sparc_register_move_cost): Likewise.
20651 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
20652 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
20653 (ASM_CPU64_DEFAULT_SPEC): Likewise.
20654 (CPP_CPU_SPEC): Handle M8.
20655 (ASM_CPU_SPEC): Likewise.
20656 (AS_M8_FLAG): Define.
20657 * config/sparc/sparc.md: Add m8 to the cpu attribute.
20658 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
20659 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
20661 * configure: Regenerate.
20662 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
20665 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20667 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
20669 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
20670 ("*movdi_insn_sp32"): Do not set v3pipe.
20671 ("*movsi_insn"): Likewise.
20672 ("*movdi_insn_sp64"): Likewise.
20673 ("*movsf_insn"): Likewise.
20674 ("*movdf_insn_sp32"): Likewise.
20675 ("*movdf_insn_sp64"): Likewise.
20676 ("*zero_extendsidi2_insn_sp64"): Likewise.
20677 ("*sign_extendsidi2_insn"): Likewise.
20678 ("*mov<VM32:mode>_insn"): Likewise.
20679 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20680 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20681 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20682 ("<vlop:code><VL:mode>3"): Likewise.
20683 ("*not_<vlop:code><VL:mode>3"): Likewise.
20684 ("*nand<VL:mode>_vis"): Likewise.
20685 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
20686 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
20687 ("one_cmpl<VL:mode>2"): Likewise.
20688 ("faligndata<VM64:mode>_vis"): Likewise.
20689 ("alignaddrsi_vis"): Likewise.
20690 ("alignaddrdi_vis"): Likweise.
20691 ("alignaddrlsi_vis"): Likewise.
20692 ("alignaddrldi_vis"): Likewise.
20693 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20694 ("bmaskdi_vis"): Likewise.
20695 ("bmasksi_vis"): Likewise.
20696 ("bshuffle<VM64:mode>_vis"): Likewise.
20697 ("cmask8<P:mode>_vis"): Likewise.
20698 ("cmask16<P:mode>_vis"): Likewise.
20699 ("cmask32<P:mode>_vis"): Likewise.
20700 ("pdistn<P:mode>_vis"): Likewise.
20701 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20703 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20705 * config/sparc/sparc.md ("subtype"): New insn attribute.
20706 ("*wrgsr_sp64"): Set insn subtype.
20707 ("*rdgsr_sp64"): Likewise.
20708 ("alignaddrsi_vis"): Likewise.
20709 ("alignaddrdi_vis"): Likewise.
20710 ("alignaddrlsi_vis"): Likewise.
20711 ("alignaddrldi_vis"): Likewise.
20712 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20713 ("fexpand_vis"): Likewise.
20714 ("fpmerge_vis"): Likewise.
20715 ("faligndata<VM64:mode>_vis"): Likewise.
20716 ("bshuffle<VM64:mode>_vis"): Likewise.
20717 ("cmask8<P:mode>_vis"): Likewise.
20718 ("cmask16<P:mode>_vis"): Likewise.
20719 ("cmask32<P:mode>_vis"): Likewise.
20720 ("fchksm16_vis"): Likewise.
20721 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
20722 ("fmean16_vis"): Likewise.
20723 ("fp<plusminus_insn>64_vis"): Likewise.
20724 ("<plusminus_insn>v8qi3"): Likewise.
20725 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20726 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
20727 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
20728 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
20729 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
20730 ("*movqi_insn"): Likewise.
20731 ("*movhi_insn"): Likewise.
20732 ("*movsi_insn"): Likewise.
20733 ("movsi_pic_gotdata_op"): Likewise.
20734 ("*movdi_insn_sp32"): Likewise.
20735 ("*movdi_insn_sp64"): Likewise.
20736 ("movdi_pic_gotdata_op"): Likewise.
20737 ("*movsf_insn"): Likewise.
20738 ("*movdf_insn_sp32"): Likewise.
20739 ("*movdf_insn_sp64"): Likewise.
20740 ("*zero_extendhisi2_insn"): Likewise.
20741 ("*zero_extendqihi2_insn"): Likewise.
20742 ("*zero_extendqisi2_insn"): Likewise.
20743 ("*zero_extendqidi2_insn"): Likewise.
20744 ("*zero_extendhidi2_insn"): Likewise.
20745 ("*zero_extendsidi2_insn_sp64"): Likewise.
20746 ("ldfsr"): Likewise.
20747 ("prefetch_64"): Likewise.
20748 ("prefetch_32"): Likewise.
20749 ("tie_ld32"): Likewise.
20750 ("tie_ld64"): Likewise.
20751 ("*tldo_ldub_sp32"): Likewise.
20752 ("*tldo_ldub1_sp32"): Likewise.
20753 ("*tldo_ldub2_sp32"): Likewise.
20754 ("*tldo_ldub_sp64"): Likewise.
20755 ("*tldo_ldub1_sp64"): Likewise.
20756 ("*tldo_ldub2_sp64"): Likewise.
20757 ("*tldo_ldub3_sp64"): Likewise.
20758 ("*tldo_lduh_sp32"): Likewise.
20759 ("*tldo_lduh1_sp32"): Likewise.
20760 ("*tldo_lduh_sp64"): Likewise.
20761 ("*tldo_lduh1_sp64"): Likewise.
20762 ("*tldo_lduh2_sp64"): Likewise.
20763 ("*tldo_lduw_sp32"): Likewise.
20764 ("*tldo_lduw_sp64"): Likewise.
20765 ("*tldo_lduw1_sp64"): Likewise.
20766 ("*tldo_ldx_sp64"): Likewise.
20767 ("*mov<VM32:mode>_insn"): Likewise.
20768 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20769 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20771 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20773 * config/sparc/sparc.md ("type"): New insn type viscmp.
20774 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
20776 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
20777 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
20778 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20779 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
20781 ("n7_vis_logical_11cycle"): Likewise.
20782 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
20783 * config/sparc/niagara2.md ("niag3_vis": Likewise.
20784 * config/sparc/niagara.md ("niag_vis"): Likewise.
20785 * config/sparc/ultra3.md ("us3_fga"): Likewise.
20786 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
20788 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20790 * config/sparc/sparc.md: New instruction type `bmask'.
20791 (bmaskdi_vis): Use the `bmask' type.
20792 (bmasksi_vis): Likewise.
20793 * config/sparc/ultra3.md (us3_array): Likewise.
20794 * config/sparc/niagara7.md (n7_array): Likewise.
20795 * config/sparc/niagara4.md (n4_array): Likewise.
20796 * config/sparc/niagara2.md (niag2_vis): Likewise.
20797 (niag3_vis): Likewise.
20798 * config/sparc/niagara.md (niag_vis): Likewise.
20800 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20802 * ipa-comdats.c: Remove optimize check from gate.
20803 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
20804 for functions not optimized.
20805 (ipa_fn_summary_read): Skip optimize check.
20806 (ipa_fn_summary_write): Likewise.
20807 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
20809 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
20811 (can_inline_edge_p): Check flag_pcc_struct_return for match.
20812 (check_callers): Give up on caller which is not optimized.
20813 (inline_small_functions): Likewise.
20814 (ipa_inline): Do not give up when not optimizing.
20815 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
20816 away unoptimizes cdtors.
20817 (whole_program_function_and_variable_visibility): Do
20818 ipa_discover_readonly_nonaddressable_vars in LTO mode.
20819 * ipa.c (process_references): Do not check optimize.
20820 (symbol_table::remove_unreachable_nodes): Update optimize check.
20821 (set_writeonly_bit): Update optimize check.
20822 (pass_ipa_cdtor_merge::gate): Do not check optimize.
20823 (pass_ipa_single_use::gate): Remove.
20825 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20827 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
20828 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
20829 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
20830 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
20831 permute_load, permute_store, adjust_extract, adjust_splat,
20832 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
20833 replace_swap_with_copy, dump_swap_insn_table,
20834 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
20835 recombine_lvx_pattern, recombine_stvx_pattern,
20836 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
20837 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
20838 to file rs6000-p8swap.c.
20839 * config/rs6000/rs6000-p8swap.c: New file.
20840 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
20841 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
20842 and rs6000*-*-* targets.
20844 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20846 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
20848 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20850 * lto-wrapper.c (merge_and_complain): Do not merge
20851 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
20852 fsigned_zeros, ftrapping_math, fwrapv.
20853 (append_compiler_options): Do not track these options.
20854 (append_linker_options): Likewie
20856 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20858 * cgraphunit.c (cgraph_node::finalize_function): When
20859 !flag_toplevel_reorde set no_reorder flag.
20860 (varpool_node::finalize_decl): Likewise.
20861 (symbol_table::compile): Drop no toplevel reorder path.
20863 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20865 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
20866 edges; zero probability is not better than uninitialized.
20868 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20870 * asan.h (asan_sanitize_allocas_p): Declare.
20871 * asan.c (asan_sanitize_allocas_p): New function.
20872 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
20873 (handle_builtin_alloca): Likewise.
20874 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
20875 if !asan_sanitize_allocas_p.
20876 * params.def (asan-instrument-allocas): Add new option.
20877 * params.h (ASAN_PROTECT_ALLOCAS): Define.
20878 * opts.c (common_handle_option): Disable allocas sanitization for
20881 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20883 * asan.c: Include gimple-fold.h.
20884 (get_last_alloca_addr): New function.
20885 (handle_builtin_stackrestore): Likewise.
20886 (handle_builtin_alloca): Likewise.
20887 (asan_emit_allocas_unpoison): Likewise.
20888 (get_mem_refs_of_builtin_call): Add new parameter, remove const
20889 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
20890 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
20891 (instrument_builtin_call): Pass gimple iterator to
20892 get_mem_refs_of_builtin_call.
20893 (last_alloca_addr): New global.
20894 * asan.h (asan_emit_allocas_unpoison): Declare.
20895 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
20896 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
20897 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
20898 if function calls alloca.
20899 * gimple-fold.c (replace_call_with_value): Remove static keyword.
20900 * gimple-fold.h (replace_call_with_value): Declare.
20901 * internal-fn.c: Include asan.h.
20902 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
20903 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
20905 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20907 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
20908 (C_SELFTEST_FLAGS): New.
20909 (CPP_SELFTEST_FLAGS): New.
20910 (SELFTEST_DEPS): New, from deps of s-selftest.
20911 (C_SELFTEST_DEPS): New, from deps of s-selftest.
20912 (CPP_SELFTEST_DEPS): New.
20913 (selftest): Add dependency on s-selftest-c++.
20914 (s-selftest): Rename to...
20915 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
20916 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
20917 than SELFTEST_FLAGS.
20918 (selftest-gdb): Rename to...
20919 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
20921 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
20922 (selftest-valgrind): Rename to...
20923 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
20925 (selftest-valgrind): Reintroduce as an alias for
20926 selftest-c-valgrind.
20927 (s-selftest-c++): New.
20928 (selftest-c++-gdb): New.
20929 (selftest-c++-valgrind): New.
20931 2017-07-06 Olivier Hainque <hainque@adacore.com>
20933 * gcc.c (process_command): When deciding if undefined variables
20934 should be ignored when processing specs, accept "gcc -v" as well.
20936 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20938 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
20939 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
20941 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20943 * config/arm/arm-cpus.in (armv8-r): Add new entry.
20944 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
20945 * config/arm/arm-tables.opt: Regenerate.
20946 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
20948 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
20950 2017-07-06 Carl Love <cel@us.ibm.com>
20952 * ChangeLog: Clean up from mid air collision
20954 2017-07-06 Carl Love <cel@us.ibm.com>
20956 * config/rs6000/rs6000-c.c: Add support for built-in functions
20957 vector signed int vec_subc (vector signed int, vector signed int);
20958 vector signed __int128 vec_subc (vector signed __int128,
20959 vector signed __int128);
20960 vector unsigned __int128 vec_subc (vector unsigned __int128,
20961 vector unsigned __int128);
20962 vector signed int vec_sube (vector signed int, vector signed int,
20963 vector signed int);
20964 vector unsigned int vec_sube (vector unsigned int,
20965 vector unsigned int,
20966 vector unsigned int);
20967 vector signed __int128 vec_sube (vector signed __int128,
20968 vector signed __int128,
20969 vector signed__int128);
20970 vector unsigned __int128 vec_sube (vector unsigned __int128,
20971 vector unsigned __int128,
20972 vector unsigned __int128);
20973 vector signed int vec_subec (vector signed int, vector signed int,
20974 vector signed int);
20975 vector unsigned int vec_subec (vector unsigned int,
20976 vector unsigned int,
20977 vector unsigned int);
20978 vector signed __int128 vec_subec (vector signed __int128,
20979 vector signed __int128,
20980 vector signed__int128);
20981 vector unsigned __int128 vec_subec (vector unsigned __int128,
20982 vector unsigned __int128,
20983 vector unsigned __int128);
20984 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
20985 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
20986 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
20987 BU_ALTIVEC_OVERLOAD_X definitions.
20988 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
20989 * doc/extend.texi: Update the built-in documentation file for the new
20990 built-in functions.
20992 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20995 * diagnostic-show-locus.c (layout::layout): Use start and finish
20996 spelling location for the start and finish of each range.
20997 * genmatch.c (linemap_client_expand_location_to_spelling_point):
20998 Add unused aspect param.
20999 * input.c (expand_location_1): Add "aspect" param, and use it
21000 to access the correct part of the location.
21001 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21003 (expand_location_to_spelling_point): Likewise.
21004 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21005 param, and pass it to expand_location_1.
21007 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21009 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21010 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21011 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21012 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21013 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21014 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21015 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21016 _mm_maskz_getmant_ss): New intrinsics.
21017 (__builtin_ia32_getexpss128_mask): Changed to ...
21018 __builtin_ia32_getexpss128_round ... this.
21019 (__builtin_ia32_getexpsd128_mask): Changed to ...
21020 __builtin_ia32_getexpsd128_round ... this.
21021 * config/i386/i386-builtin-types.def
21022 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21023 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21024 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21025 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21026 __builtin_ia32_getmantss_mask_round): New builtins.
21027 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21028 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21029 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21030 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21031 * config/i386/sse.md
21032 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21033 avx512f_sgetexp<mode><mask_scalar_name>
21034 <round_saeonly_scalar_name> ... this.
21035 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21036 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21037 vgetexp<ssescalarmodesuffix>
21038 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21039 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21040 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21041 avx512f_vgetmant<mode><mask_scalar_name>
21042 <round_saeonly_scalar_name> ... this.
21043 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21044 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21045 vgetmant<ssescalarmodesuffix>
21046 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21047 %0<mask_scalar_operand4>, %1, %2
21048 <round_saeonly_scalar_mask_op4>, %3} ... this.
21049 * config/i386/subst.md (mask_scalar_operand4,
21050 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21051 round_saeonly_scalar_nimm_predicate): New subst attributes.
21053 2017-07-06 Julia Koval <julia.koval@intel.com>
21055 * config/i386/i386.c (ix86_erase_embedded_rounding):
21056 Remove code for old rounding pattern.
21058 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21060 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21062 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21064 * doc/sourcebuild.texi (Test Directives, Variants of
21065 dg-require-support): Add documentation for dg-require-stack-check.
21067 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21069 * config/i386/subst.md (mask_scalar, round_scalar,
21070 round_saeonly_scalar): New meta-templates.
21071 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21072 round_scalar_mask_operand3, round_scalar_mask_op3,
21073 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21074 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21075 round_saeonly_scalar_constraint,
21076 round_saeonly_scalar_prefix): New subst attribute.
21077 * config/i386/sse.md
21078 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21079 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21080 <round_scalar_name> ... this.
21081 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21082 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21083 <round_scalar_name> ... this.
21084 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21085 <sse>_vm<code><mode>3<mask_scalar_name>
21086 <round_saeonly_scalar_name> ... this.
21087 (v<plusminus_mnemonic><ssescalarmodesuffix>
21088 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21089 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21090 v<plusminus_mnemonic><ssescalarmodesuffix>
21091 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21092 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21093 (v<multdiv_mnemonic><ssescalarmodesuffix>
21094 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21095 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21096 v<multdiv_mnemonic><ssescalarmodesuffix>
21097 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21098 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21099 (v<maxmin_float><ssescalarmodesuffix>
21100 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21101 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21102 v<maxmin_float><ssescalarmodesuffix>
21103 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21104 %0<mask_scalar_operand3>, %1, %<iptr>2
21105 <round_saeonly_scalar_mask_op3>} ... this.
21107 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21109 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21110 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21112 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21113 Alan Hayward <alan.hayward@arm.com>
21114 David Sherwood <david.sherwood@arm.com>
21116 * combine.c (simplify_if_then_else): Remove "enum" before
21118 * compare-elim.c (can_eliminate_compare): Likewise.
21119 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21121 (aarch64_lookup_simd_builtin_type): Likewise.
21122 (aarch64_simd_builtin_type): Likewise.
21123 (aarch64_init_simd_builtin_types): Likewise.
21124 (aarch64_simd_expand_args): Likewise.
21125 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21127 (aarch64_reverse_mask): Likewise.
21128 (aarch64_simd_emit_reg_reg_move): Likewise.
21129 (aarch64_gen_adjusted_ldpstp): Likewise.
21130 (aarch64_ccmp_mode_to_code): Likewise.
21131 (aarch64_operands_ok_for_ldpstp): Likewise.
21132 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21133 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21135 (aarch64_min_divisions_for_recip_mul): Likewise.
21136 (aarch64_reassociation_width): Likewise.
21137 (aarch64_get_condition_code_1): Likewise.
21138 (aarch64_simd_emit_reg_reg_move): Likewise.
21139 (aarch64_simd_attr_length_rglist): Likewise.
21140 (aarch64_reverse_mask): Likewise.
21141 (aarch64_operands_ok_for_ldpstp): Likewise.
21142 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21143 (aarch64_gen_adjusted_ldpstp): Likewise.
21144 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21146 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21147 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21148 (arm_lookup_simd_builtin_type): Likewise.
21149 (arm_simd_builtin_type): Likewise.
21150 (arm_init_simd_builtin_types): Likewise.
21151 (arm_expand_builtin_args): Likewise.
21152 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21153 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21154 (ft32_setup_incoming_varargs): Likewise.
21155 (ft32_function_arg): Likewise.
21156 (ft32_function_arg_advance): Likewise.
21157 (ft32_pass_by_reference): Likewise.
21158 (ft32_arg_partial_bytes): Likewise.
21159 (ft32_valid_pointer_mode): Likewise.
21160 (ft32_addr_space_pointer_mode): Likewise.
21161 (ft32_addr_space_legitimate_address_p): Likewise.
21162 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21164 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21165 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21166 (iamcu_alignment): Likewise.
21167 (canonicalize_vector_int_perm): Likewise.
21168 (ix86_noce_conversion_profitable_p): Likewise.
21169 (ix86_mpx_bound_mode): Likewise.
21170 (ix86_operands_ok_for_move_multiple): Likewise.
21171 * config/microblaze/microblaze-protos.h
21172 (microblaze_expand_conditional_branch_reg): Likewise.
21173 * config/microblaze/microblaze.c
21174 (microblaze_expand_conditional_branch_reg): Likewise.
21175 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21177 (rs6000_reassociation_width): Likewise.
21178 (rs6000_invalid_binary_op): Likewise.
21179 (fusion_p9_p): Likewise.
21180 (emit_fusion_p9_load): Likewise.
21181 (emit_fusion_p9_store): Likewise.
21182 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21184 (riscv_hard_regno_mode_ok_p): Likewise.
21185 (riscv_address_insns): Likewise.
21186 (riscv_split_symbol): Likewise.
21187 (riscv_legitimize_move): Likewise.
21188 (riscv_function_value): Likewise.
21189 (riscv_hard_regno_nregs): Likewise.
21190 (riscv_expand_builtin): Likewise.
21191 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21192 (riscv_build_integer): Likewise.
21193 (riscv_split_integer): Likewise.
21194 (riscv_legitimate_constant_p): Likewise.
21195 (riscv_cannot_force_const_mem): Likewise.
21196 (riscv_regno_mode_ok_for_base_p): Likewise.
21197 (riscv_valid_base_register_p): Likewise.
21198 (riscv_valid_offset_p): Likewise.
21199 (riscv_valid_lo_sum_p): Likewise.
21200 (riscv_classify_address): Likewise.
21201 (riscv_legitimate_address_p): Likewise.
21202 (riscv_address_insns): Likewise.
21203 (riscv_load_store_insns): Likewise.
21204 (riscv_force_binary): Likewise.
21205 (riscv_split_symbol): Likewise.
21206 (riscv_force_address): Likewise.
21207 (riscv_legitimize_address): Likewise.
21208 (riscv_move_integer): Likewise.
21209 (riscv_legitimize_const_move): Likewise.
21210 (riscv_legitimize_move): Likewise.
21211 (riscv_address_cost): Likewise.
21212 (riscv_subword): Likewise.
21213 (riscv_output_move): Likewise.
21214 (riscv_canonicalize_int_order_test): Likewise.
21215 (riscv_emit_int_order_test): Likewise.
21216 (riscv_function_arg_boundary): Likewise.
21217 (riscv_pass_mode_in_fpr_p): Likewise.
21218 (riscv_pass_fpr_single): Likewise.
21219 (riscv_pass_fpr_pair): Likewise.
21220 (riscv_get_arg_info): Likewise.
21221 (riscv_function_arg): Likewise.
21222 (riscv_function_arg_advance): Likewise.
21223 (riscv_arg_partial_bytes): Likewise.
21224 (riscv_function_value): Likewise.
21225 (riscv_pass_by_reference): Likewise.
21226 (riscv_setup_incoming_varargs): Likewise.
21227 (riscv_print_operand): Likewise.
21228 (riscv_elf_select_rtx_section): Likewise.
21229 (riscv_save_restore_reg): Likewise.
21230 (riscv_for_each_saved_reg): Likewise.
21231 (riscv_register_move_cost): Likewise.
21232 (riscv_hard_regno_mode_ok_p): Likewise.
21233 (riscv_hard_regno_nregs): Likewise.
21234 (riscv_class_max_nregs): Likewise.
21235 (riscv_memory_move_cost): Likewise.
21236 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
21237 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
21238 (rl78_addr_space_address_mode): Likewise.
21239 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21241 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21242 (rs6000_reassociation_width): Likewise.
21243 (rs6000_invalid_binary_op): Likewise.
21244 (fusion_p9_p): Likewise.
21245 (emit_fusion_p9_load): Likewise.
21246 (emit_fusion_p9_store): Likewise.
21247 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
21248 (ok_for_simple_move_operands): Likewise.
21249 (ok_for_simple_move_strict_operands): Likewise.
21250 (ok_for_simple_arith_logic_operands): Likewise.
21251 (visium_legitimize_reload_address): Likewise.
21252 (visium_select_cc_mode): Likewise.
21253 (output_cbranch): Likewise.
21254 (visium_split_double_move): Likewise.
21255 (visium_expand_copysign): Likewise.
21256 (visium_expand_int_cstore): Likewise.
21257 (visium_expand_fp_cstore): Likewise.
21258 * config/visium/visium.c (visium_pass_by_reference): Likewise.
21259 (visium_function_arg): Likewise.
21260 (visium_function_arg_advance): Likewise.
21261 (visium_libcall_value): Likewise.
21262 (visium_setup_incoming_varargs): Likewise.
21263 (visium_legitimate_constant_p): Likewise.
21264 (visium_legitimate_address_p): Likewise.
21265 (visium_legitimize_address): Likewise.
21266 (visium_secondary_reload): Likewise.
21267 (visium_register_move_cost): Likewise.
21268 (visium_memory_move_cost): Likewise.
21269 (prepare_move_operands): Likewise.
21270 (ok_for_simple_move_operands): Likewise.
21271 (ok_for_simple_move_strict_operands): Likewise.
21272 (ok_for_simple_arith_logic_operands): Likewise.
21273 (visium_function_value_1): Likewise.
21274 (rtx_ok_for_offset_p): Likewise.
21275 (visium_legitimize_reload_address): Likewise.
21276 (visium_split_double_move): Likewise.
21277 (visium_expand_copysign): Likewise.
21278 (visium_expand_int_cstore): Likewise.
21279 (visium_expand_fp_cstore): Likewise.
21280 (visium_split_cstore): Likewise.
21281 (visium_select_cc_mode): Likewise.
21282 (visium_split_cbranch): Likewise.
21283 (output_cbranch): Likewise.
21284 (visium_print_operand_address): Likewise.
21285 * expmed.c (flip_storage_order): Likewise.
21286 * expmed.h (emit_cstore): Likewise.
21287 (flip_storage_order): Likewise.
21288 * genrecog.c (validate_pattern): Likewise.
21289 * hsa-gen.c (gen_hsa_addr): Likewise.
21290 * internal-fn.c (expand_arith_overflow): Likewise.
21291 * ira-color.c (allocno_copy_cost_saving): Likewise.
21292 * lra-assigns.c (find_hard_regno_for_1): Likewise.
21293 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
21294 (process_invariant_for_inheritance): Likewise.
21295 * lra-eliminations.c (move_plus_up): Likewise.
21296 * omp-low.c (lower_oacc_reductions): Likewise.
21297 * simplify-rtx.c (simplify_subreg): Likewise.
21298 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
21299 (TARGET_CHKP_BOUND_MODE): Likewise..
21300 * targhooks.c (default_chkp_bound_mode): Likewise.
21301 (default_setup_incoming_vararg_bounds): Likewise.
21302 * targhooks.h (default_chkp_bound_mode): Likewise.
21303 (default_setup_incoming_vararg_bounds): Likewise.
21304 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
21305 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
21306 (have_whole_vector_shift): Likewise.
21307 * tree-vect-stmts.c (vectorizable_load): Likewise.
21308 * doc/tm.texi: Regenerate.
21310 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21312 Graceful degrade if Binutils PR21472 is not available.
21315 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
21316 .rodata in flash test fails.
21317 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
21318 * confgure: Regenerate.
21319 * config.in: Regenerate.
21320 * config/avr/avr.c (avr_asm_named_section)
21321 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
21322 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
21323 (avr_asm_init_sections): Same.
21325 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21327 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
21328 (fma<VH:mode>4_intrinsic): Likewise.
21329 (*fmsub<VCVTF:mode>4): Likewise.
21330 (*fmsub<VH:mode>4_intrinsic): Likewise.
21332 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21335 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
21336 Don't depend on "optimize > 0".
21337 (out_movhi_r_mr, out_movqi_mr_r): Same.
21338 (out_movhi_mr_r, out_movqi_r_mr): Same.
21339 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
21340 io_address_operand on "optimize > 0".
21342 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21344 * tree-loop-distribution.c: Add general explanantion on the pass.
21345 (generate_loops_for_partition): Mark distributed loop.
21346 (pg_add_dependence_edges): New parameter. Handle alias data
21347 dependence specially and record it in the parameter if asked.
21348 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
21349 (init_partition_graph_vertices, add_partition_graph_edge): New.
21350 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
21351 (free_partition_graph_vdata, build_partition_graph): New.
21352 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
21353 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
21354 (data_ref_segment_size, latch_dominated_by_data_ref): New.
21355 (compute_alias_check_pairs, version_loop_by_alias_check): New.
21356 (version_for_distribution_p, finalize_partitions): New.
21357 (distribute_loop): Handle alias data dependence specially. Factor
21358 out loop fusion code as functions and call these functions.
21360 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21362 * tree-loop-distribution.c (classify_partition): New parameter and
21363 better handle reduction statement.
21364 (rdg_build_partitions): Revise comment.
21365 (distribute_loop): Compute statements in all partitions and pass it
21366 to classify_partition.
21368 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21370 * tree-loop-distribution.c (enum partition_type): New.
21371 (struct partition): New field type.
21372 (partition_merge_into): Add parameter. Update partition type.
21373 (data_dep_in_cycle_p, update_type_for_merge): New functions.
21374 (build_rdg_partition_for_vertex): Compute partition type.
21375 (rdg_build_partitions): Dump partition type.
21376 (distribute_loop): Update calls to partition_merge_into.
21378 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21380 * tree-loop-distribution.c (struct ddr_hasher): New.
21381 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
21383 (classify_partition): Call get_data_dependence.
21384 (pg_add_dependence_edges): Ditto.
21385 (distribute_loop): Release data dependence hash table.
21387 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21389 * tree-loop-distribution.c (ref_base_address): Delete.
21390 (similar_memory_accesses): Rename ...
21391 (share_memory_accesses): ... to this. Check if partitions access
21392 the same memory reference.
21393 (distribute_loop): Call share_memory_accesses.
21395 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21397 * tree-loop-distribution.c (struct partition): New field recording
21398 its data reference.
21399 (partition_alloc, partition_free): Init and release data refs.
21400 (partition_merge_into): Merge data refs.
21401 (build_rdg_partition_for_vertex): Collect data refs for partition.
21402 (pg_add_dependence_edges): Change parameters from vector to bitmap.
21404 (distribute_loop): Remve data refs from vertice data of partition
21407 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21409 * tree-loop-distribution.c (params.h): Include header file.
21410 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
21411 (datarefs_vec): New global var.
21412 (create_rdg_vertices): Use datarefs_vec directly.
21413 (free_rdg): Don't free data references.
21414 (build_rdg): Update use. Don't free data references.
21415 (distribute_loop): Compute global variable for data references.
21416 Bail out if there are too many data references.
21418 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21420 * tree-loop-distribution.c (loop_nest): New global var.
21421 (build_rdg): Use loop directly, rather than loop nest.
21422 (pg_add_dependence_edges): Remove loop nest parameter. Use global
21424 (distribute_loop): Compute global variable loop nest. Update use.
21426 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21428 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
21429 (partition_merge_into): New parameter. Dump reason for fusion.
21430 (distribute_loop): Update use of partition_merge_into.
21432 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21434 * tree-loop-distribution.c (bb_top_order_index): New.
21435 (bb_top_order_index_size, bb_top_order_cmp): New.
21436 (stmts_from_loop): Use topological order.
21437 (pass_loop_distribution::execute): Compute and release topological
21438 order for basic blocks.
21440 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21442 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
21445 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21447 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
21448 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
21449 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
21450 * internal-fn.def (LOOP_DIST_ALIAS): New.
21451 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
21452 (fold_loop_internal_call): ... this.
21453 (vect_loop_dist_alias_call): New function.
21454 (set_uid_loop_bbs): Call fold_loop_internal_call.
21455 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
21458 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21461 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
21462 Require dead FLAGS_REG at the beginning of a peephole.
21464 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21467 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
21468 arguments in the call to __builtin_ia32_sbb_u32.
21469 (_subborrow_u64): Swap _X and _Y arguments in the call to
21470 __builtin_ia32_sbb_u64.
21472 2017-07-04 Jakub Jelinek <jakub@redhat.com>
21475 * tree-vrp.c (compare_assert_loc): Turn into a function template
21476 with stable template parameter. Only test if a->e is NULL,
21477 !a->e == !b->e has been verified already. Use e == NULL or
21478 e != NULL instead of e or ! e tests. If stable is true, don't use
21479 iterative_hash_expr, on the other side allow a or b or both NULL
21480 and sort the NULLs last.
21481 (process_assert_insertions): Sort using compare_assert_loc<false>
21482 instead of compare_assert_loc, later sort using
21483 compare_assert_loc<true> before calling process_assert_insertions_for
21484 in a loop. Use break instead of continue once seen NULL pointer.
21486 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21488 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21489 Cortex-R7 and Cortex-R8 processors.
21491 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21493 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
21494 uninitialized while src is not.
21496 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
21498 * common/config/arm/arm-common.c: Adjust include path for
21500 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
21501 (arm-cpu.h): Create in build directory. Adjust dependency rules.
21502 (arm-cpu-data.h): Likewise.
21503 (arm-cpu-cdata.h): Likewise.
21504 * config/arm/arm-cpu.h: Delete.
21505 * config/arm/arm-cpu-cdata.h: Delete.
21506 * config/arm/arm-cpu-data.h: Delete.
21508 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
21510 * config/arm/arm-cpus.in (cortex-a55): New.
21511 (cortex-a75): Likewise.
21512 (cortex-a75.cortex-a55): Likewise.
21513 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
21515 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
21516 * config/arm/arm-cpu-cdata.h: Regenerate.
21517 * config/arm/arm-cpu-data.h: Regenerate.
21518 * config/arm/arm-cpu.h: Regenerate.
21519 * config/arm/arm-tables.opt: Regenerate.
21520 * config/arm/arm-tune.md: Regenerate.
21522 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21524 * haifa-sched.c (sched_create_recovery_edges): Update profile.
21526 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21528 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
21531 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
21533 PR tree-optimization/81292
21534 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
21535 full_string_p, also call adjust_related_strinfos if the adjustment
21536 is simple, otherwise invalidate related strinfos.
21538 2017-07-04 Martin Liska <mliska@suse.cz>
21541 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
21542 newly created variable as DECL_IGNORED_P.
21544 2017-07-04 Martin Liska <mliska@suse.cz>
21547 * ipa-inline.c (inline_small_functions):
21548 Use xstrdup_for_dump.
21550 2017-07-04 Tom de Vries <tom@codesourcery.com>
21552 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
21554 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21557 * config/darwin.c (darwin_function_switched_text_sections):
21560 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
21562 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
21564 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21566 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
21568 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21570 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
21571 min_profitable_iters, and th as inclusive lower bounds.
21572 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
21573 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
21574 for min_profitable_iters and min_profitable_estimate.
21575 (vect_transform_loop): Treat th as an inclusive lower bound.
21576 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21578 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21581 * config/darwin.c (darwin_function_switched_text_sections):
21582 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
21583 in two pieces, and suppress the use of buf.
21585 2017-07-03 Nathan Sidwell <nathan@acm.org>
21587 * hash-table.h (hash_table_mod1): Fix indentation.
21589 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21591 PR middle-end/81290
21592 * predict.c (force_edge_cold): Be more careful about propagation
21594 * profile-count.h (profile_probability::guessed,
21595 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
21597 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
21599 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
21601 * doc/invoke.texi (rcpc architecture extension): Document it.
21603 2017-07-03 Richard Biener <rguenther@suse.de>
21605 PR tree-optimization/60510
21606 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
21607 the scalar reduction PHI and use it.
21608 (vectorizable_reduction): Properly guard the single_defuse_cycle
21609 path for non-SLP reduction chains where we cannot use it.
21610 Rework reduc_def/index and vector type deduction. Rework
21611 vector operand gathering during reduction op code-gen.
21612 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
21613 chains dissolve the chain and leave it to non-SLP reduction
21616 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21618 * tree-data-ref.h (dr_alignment): Declare.
21619 * tree-data-ref.c (dr_alignment): New function.
21620 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
21621 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
21623 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
21625 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21627 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
21628 and base_misalignment fields.
21629 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
21630 * tree-data-ref.c: Include builtins.h.
21631 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
21632 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
21633 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
21634 * tree-vect-data-refs.c: Include tree-cfg.h.
21635 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
21636 fields instead of calculating an alignment here.
21637 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
21638 innermost_loop_behavior fields.
21640 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21642 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
21644 (DR_STEP_ALIGNMENT): New macro.
21645 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
21646 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
21647 (create_data_ref): Print it.
21648 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
21649 to tell whether the step preserves vector (mis)alignment.
21650 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21651 Move the check for an integer step and generalise to all INTEGER_CST.
21652 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
21653 Print the outer step alignment.
21655 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21657 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
21658 with offset_alignment.
21659 (DR_ALIGNED_TO): Delete.
21660 (DR_OFFSET_ALIGNMENT): New macro.
21661 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
21662 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
21663 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
21664 (create_data_ref): Likewise.
21665 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21666 (vect_analyze_data_refs): Likewise.
21667 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
21668 creating dummy innermost behavior.
21670 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21672 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
21673 with a "innermost_loop_behavior *" and refeence tree.
21674 * tree-data-ref.c (dr_analyze_innermost): Likewise.
21675 (create_data_ref): Update call accordingly.
21676 * tree-predcom.c (find_looparound_phi): Likewise.
21678 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21680 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
21681 fields with dr_wrt_vec_loop.
21682 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
21683 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
21684 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
21685 (vect_dr_behavior): New function.
21686 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21687 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
21688 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
21689 track whether the step preserves the misalignment.
21690 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21691 Use vect_dr_behavior.
21692 (vect_setup_realignment): Update call accordingly.
21693 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
21694 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
21695 call to vect_create_addr_base_for_vector_ref.
21696 (vect_create_cond_for_align_checks): Likewise.
21697 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
21698 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
21699 (vect_recog_mask_conversion_pattern): Likewise.
21700 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
21701 (new_stmt_vec_info): Remove redundant zeroing.
21703 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21705 * common/config/arm/arm-common.c (arm_be8_option): New function.
21706 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
21707 (ISA_ARMv6): Add isa_bit_be8.
21708 * config/arm/arm.h (arm_be8_option): Add prototype.
21709 (BE8_SPEC_FUNCTION): New define.
21710 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
21711 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
21712 (mlittle-endian): Similarly.
21713 (mbe8, mbe32): New options.
21714 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
21715 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
21717 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21719 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
21721 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21723 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
21724 (cleanup_tree_cfg_bb): Use it.
21725 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
21727 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
21729 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21732 * loop-doloop.c (add_test): Update profile.
21734 2017-07-03 Martin Liska <mliska@suse.cz>
21737 * sanopt.c (rewrite_usage_of_param): New function.
21738 (sanitize_rewrite_addressable_params): Likewise.
21739 (pass_sanopt::execute): Call rewrite_usage_of_param.
21741 2017-07-03 Richard Biener <rguenther@suse.de>
21743 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
21744 back to using VIEW_CONVERT_EXPR.
21746 2017-07-03 Martin Liska <mliska@suse.cz>
21749 * doc/extend.texi: Document when a resolver function is
21750 generated for target_clones.
21752 2017-07-03 Martin Liska <mliska@suse.cz>
21754 * asan.c (asan_emit_stack_protection): Unpoison just red zones
21755 and shadow memory of auto variables which are subject of
21756 use-after-scope sanitization.
21757 (asan_expand_mark_ifn): Add do set only when is_poison.
21759 2016-07-03 Richard Biener <rguenther@suse.de>
21761 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
21763 (vect_force_simple_reduction): Record reduction def -> phi mapping.
21764 (vectorizable_reduction): Perform reduction PHI creation when
21765 visiting a reduction PHI and adjust and simplify code generation
21766 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
21767 (vect_transform_loop): Visit reduction PHIs.
21768 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
21769 defs into the SLP tree.
21770 (vect_build_slp_tree): Reduction defs terminate the recursion.
21771 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
21773 (vect_get_vec_defs_for_stmt_copy): Export.
21774 (vect_get_vec_defs): Likewise.
21775 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
21777 (vect_get_vec_defs_for_stmt_copy): Declare.
21778 (vect_get_vec_defs): Likewise.
21780 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21782 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
21783 parameter with a "loop" parameter and use it instead of the
21784 loop containing DR_STMT. Don't check simple_iv when doing
21785 BB analysis. Describe the two analysis modes in the comment.
21787 2017-07-03 Tom de Vries <tom@codesourcery.com>
21789 PR tree-optimization/69468
21790 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
21791 (find_same_succ_bb): Handle ignore_edge_flags.
21793 2017-07-03 Tom de Vries <tom@codesourcery.com>
21795 PR tree-optimization/81192
21796 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
21798 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
21800 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
21802 2017-07-03 Tom de Vries <tom@codesourcery.com>
21804 PR tree-optimization/81192
21805 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
21806 BB_SAME_SUCC (bb) == NULL.
21808 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21810 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
21813 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21815 * dumpfile.c: Include profile-count.h
21816 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
21818 (insert_cond_bb): Update profile.
21819 * tree-cfg.h (insert_cond_bb): Update prototype.
21820 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
21821 * tree-dump.c: Do not include tree-cfg.
21823 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21825 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
21827 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21829 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
21832 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21834 * tree-complex.c (expand_complex_div_wide): update profile.
21836 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21837 Alan Hayward <alan.hayward@arm.com>
21838 David Sherwood <david.sherwood@arm.com>
21840 * Makefile.in (MACHMODE_H): Remove insn-modes.h
21841 (CORETYPES_H): New define.
21842 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
21843 (insn-modes-inline.h, s-modes-inline-h): New rules.
21844 (generated_files): Add insn-modes-inline.h.
21845 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
21846 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
21847 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
21848 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
21849 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
21850 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
21851 (build/gencodes.o, build/genconditions.o): Likewise.
21852 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
21853 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
21854 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
21855 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
21856 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
21857 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
21858 * coretypes.h: Include everything up to real.h for generators.
21859 Include insn-modes.h first. Include wide-int-print.h after
21860 wide-int.h. Include insn-modes-inline.h and then machmode.h.
21861 * machmode.h: Don't include insn-modes.h here.
21862 * function-tests.c: Remove includes of signop.h, machmode.h,
21863 double-int.h and wide-int.h.
21865 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
21867 * optc-save-gen.awk: Likewise.
21868 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
21869 * godump.c: Remove include of wide-int-print.h.
21870 * pretty-print.h: Likewise.
21871 * wide-int-print.cc: Likewise.
21872 * wide-int.cc: Likewise.
21873 * hash-map-tests.c: Remove include of signop.h.
21874 * hash-set-tests.c: Likewise.
21875 * rtl-tests.c: Likewise.
21876 * mkconfig.sh: Remove include of machmode.h.
21877 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
21879 (emit_insn_modes_inline_h): ...this new function. Emit the code
21880 into an insn-modes-inline.h header file, adding appropriate
21881 include guards and end comments.
21882 (emit_insn_modes_c_header): Remove include of machmode.h.
21883 (emit_min_insn_modes_c_header): Include coretypes.h rather than
21885 (main): Handle -i flag and call emit_insn_modes_inline_h when
21888 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21890 * tree-ssa-strlen.c (strinfo): Rename the length field to
21891 nonzero_chars. Add a full_string_p field.
21892 (compare_nonzero_chars, zero_length_string_p): New functions.
21893 (get_addr_stridx): Add an offset_out parameter.
21894 Use compare_nonzero_chars.
21895 (get_stridx): Update accordingly. Use compare_nonzero_chars.
21896 (new_strinfo): Update after above changes to strinfo.
21897 (set_endptr_and_length): Set full_string_p.
21898 (get_string_length): Update after above changes to strinfo.
21899 (unshare_strinfo): Update call to new_strinfo.
21900 (maybe_invalidate): Likewise.
21901 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
21902 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
21903 as a uhwi instead of an shwi. Update after above changes to
21904 strinfo and new_strinfo.
21905 (zero_length_string): Assert that chainsi contains full strings.
21906 Use zero_length_string_p. Update call to new_strinfo.
21907 (adjust_related_strinfos): Update after above changes to strinfo.
21908 Copy full_string_p from origsi.
21909 (adjust_last_stmt): Use zero_length_string_p.
21910 (handle_builtin_strlen): Update after above changes to strinfo and
21911 new_strinfo. Install the lhs as the string length if the previous
21912 entry didn't describe a full string.
21913 (handle_builtin_strchr): Update after above changes to strinfo
21915 (handle_builtin_strcpy): Likewise.
21916 (handle_builtin_strcat): Likewise.
21917 (handle_builtin_malloc): Likewise.
21918 (handle_pointer_plus): Likewise.
21919 (handle_builtin_memcpy): Likewise. Track nonzero characters
21920 that aren't necessarily followed by a nul terminator.
21921 (handle_char_store): Likewise.
21923 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21925 PR tree-optimization/80769
21926 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
21927 for malloc and calloc. Document the new invariant that all related
21928 strinfos have delayed lengths or none do.
21929 (verify_related_strinfos): Move earlier in file.
21930 (set_endptr_and_length): New function, split out from...
21931 (get_string_length): ...here. Also set the lengths of related
21933 (zero_length_string): Assert that chainsi has known (rather than
21935 (adjust_related_strinfos): Likewise.
21937 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21939 PR tree-optimization/81136
21940 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
21941 assert that two references with the same misalignment have the same
21942 compile-time misalignment if those compile-time misalignments
21945 2017-07-01 Andi Kleen <ak@linux.intel.com>
21947 * print-tree.c (print_node): Print all attributes.
21949 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21951 * cfg.c (scale_bbs_frequencies): New function.
21952 * cfg.h (scale_bbs_frequencies): Declare it.
21953 * cfgloopanal.c (single_likely_exit): Cleanup.
21954 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
21956 (scale_loop_profile): Likewise.
21957 (loop_version): Likewise.
21958 (create_empty_loop_on_edge): Update.
21959 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
21960 scale_loop_frequencies, scale_loop_profile, loopify,
21961 loop_version): Update prototypes.
21962 * modulo-sched.c (sms_schedule): Update.
21963 * predict.c (unlikely_executed_edge_p): Also check probability.
21964 (probably_never_executed_edge_p): Fix typo.
21965 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21966 * tree-parloops.c (gen_parallel_loop): Update.
21967 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
21968 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21969 * tree-ssa-loop-split.c (split_loop): Update.
21970 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21971 * tree-vect-loop-manip.c (vect_do_peeling): Update.
21972 (vect_loop_versioning): Update.
21973 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21975 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21977 * trans-mem.c (split_bb_make_tm_edge): Update profile.
21979 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21981 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
21982 to keep profile consistent.
21984 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21986 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
21987 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
21988 * profile-count.h (max_safe_multiplier): Make unsigned.
21989 (profile_count::guessed_zero): New.
21991 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21993 * bb-reorder.c (fix_up_crossing_landing_pad,
21994 fix_crossing_conditional_branches): Use make_single_succ_edge
21995 to keep profile consistent.
21997 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21999 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22002 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22005 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22006 the right scopes, make sure cond_jump isn't preserved between multiple
22007 iterations. Search for fallthru edge whenever there are 3+ edges and
22008 use find_fallthru_edge for it.
22010 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22012 Patch by Alexander Monakov <amonakov@ispras.ru>
22013 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22014 probabilities consistently.
22016 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22018 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22019 emit_cmp_and_jump_insns.
22021 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22024 * tree-inline.c (expand_call_inline): Combine profile statuses.
22026 2017-06-30 Andrew Pinski <apinski@cavium.com>
22028 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22029 fold_stmt returned true.
22031 2017-06-30 Nathan Sidwell <nathan@acm.org>
22033 * ggc.h (empty_string): Delete.
22034 * cfgexpand.c (expand_asm_stmt): Use plain "".
22035 * optabs.c (expand_asm_memory_barrier): Likewise.
22036 * stringpool.c (empty_string): Delete.
22037 (digit_vector, digit_string): Delete.
22038 (ggc_alloc_string): Use plain "", don't optimize single digit
22039 strings. Use ggc_alloc_atomic.
22041 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22043 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22044 comparison set and one other set, use the cost of the non-comparison
22047 2017-06-30 Nathan Sidwell <nathan@acm.org>
22049 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22052 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22054 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22055 loops. Remove now unneeded calls to gimple_switch_set_label() that
22056 just set removed labels to NULL_TREE.
22058 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22060 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22061 (set_range_info): ...here. Only call set_range_info_raw if domain
22063 (set_nonzero_bits): Call set_range_info_raw.
22064 * tree-ssanames.h (set_range_info_raw): New.
22066 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22069 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22070 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22071 of nonimmediate_operand and <store_mask_constraint> instead of m
22072 for the input operand. For V8FI iterator, always split if input
22073 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22074 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22075 <store_mask_predicate> instead of register_operand and
22076 <store_mask_constraint> instead of v for the input operand. Make
22077 sure both operands aren't MEMs for if not <mask_applied>.
22079 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22081 * lto-wrapper.c (copy_file) Close both file descriptors before
22084 2017-06-30 Martin Liska <mliska@suse.cz>
22087 * multiple_target.c (create_dispatcher_calls): Make ifunc
22088 also for function that don't have calls or are not referenced.
22090 2017-06-30 Richard Biener <rguenther@suse.de>
22092 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22093 analyze the first scalar stmt. Move vector type computation
22094 for the BB case here from ...
22095 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22096 live operation processing in the SLP case properly.
22098 2017-06-30 Richard Biener <rguenther@suse.de>
22100 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22102 2017-06-30 Martin Liska <mliska@suse.cz>
22105 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22106 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22108 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22110 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22113 2017-06-30 Richard Biener <rguenther@suse.de>
22115 PR tree-optimization/81249
22116 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22117 condition reduction result to original scalar type.
22119 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22121 * profile-count.h (enum profile_quality): Fix typos and whitespace
22124 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22126 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22127 type for branch probabilities.
22129 2017-06-29 Julian Brown <julian@codesourcery.com>
22130 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22132 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22133 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22134 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22135 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22137 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22139 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22140 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22141 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22142 CC usage from generic code to here.
22143 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22144 CC usage into the target macros.
22146 2017-06-29 Maya Rashish <coypu@sdf.org>
22148 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22151 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22153 * arm/arm-builtins.c: Include profile-count.h
22154 * except.c (sjlj_emit_function_enter): Use
22155 profile_probability::unlikely.
22157 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22159 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22160 and tocrel_offset be pointer args rather than implicitly using
22162 (legitimate_constant_pool_address_p, rs6000_emit_move,
22163 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22164 tocrel_offset and use in toc_relative_expr_p call.
22165 (print_operand, print_operand_address): Use static tocrel_base_oac
22166 and tocrel_offset_oac.
22167 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22170 2017-06-29 Maya Rashish <coypu@sdf.org>
22172 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22174 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22176 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22177 objects, take into account only the alignment of 'op0' and 'mode1' if
22180 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22182 * ccmp.c (ccmp_tree_comparison_p): New function.
22183 (ccmp_candidate_p): Update to use above function.
22184 (get_compare_parts): New function.
22185 (expand_ccmp_next): Update to use new functions.
22186 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22188 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22189 take mode as argument.
22190 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22191 * expr.c (expand_expr_real_1): Pass mode as argument.
22193 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22195 * combine.c (combine_instructions): Print insns to dump_file, together
22198 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22200 * asan.c (asan_emit_stack_protection): Update.
22201 (create_cond_insert_point): Update.
22202 * auto-profile.c (afdo_propagate_circuit): Update.
22203 * basic-block.h (struct edge_def): Turn probability to
22204 profile_probability.
22205 (EDGE_FREQUENCY): Update.
22206 * bb-reorder.c (find_traces_1_round): Update.
22207 (better_edge_p): Update.
22208 (sanitize_hot_paths): Update.
22209 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22210 (make_single_succ_edge): Update.
22211 (check_bb_profile): Update.
22212 (dump_edge_info): Update.
22213 (update_bb_profile_for_threading): Update.
22214 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
22215 probabilitycount to 0.
22216 * cfgbuild.c (compute_outgoing_frequencies): Update.
22217 * cfgcleanup.c (try_forward_edges): Update.
22218 (outgoing_edges_match): Update.
22219 (try_crossjump_to_edge): Update.
22220 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
22221 (expand_gimple_tailcall): Update.
22222 (construct_init_block): Use make_single_succ_edge.
22223 (construct_exit_block): Use make_single_succ_edge.
22224 * cfghooks.c (verify_flow_info): Update.
22225 (redirect_edge_succ_nodup): Update.
22226 (split_edge): Update.
22227 (account_profile_record): Update.
22228 * cfgloopanal.c (single_likely_exit): Update.
22229 * cfgloopmanip.c (scale_loop_profile): Update.
22230 (set_zero_probability): Remove.
22231 (duplicate_loop_to_header_edge): Update.
22232 * cfgloopmanip.h (loop_version): Update prototype.
22233 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
22234 (force_nonfallthru_and_redirect): Update.
22235 (update_br_prob_note): Update.
22236 (rtl_verify_edges): Update.
22237 (purge_dead_edges): Update.
22238 (rtl_lv_add_condition_to_bb): Update.
22239 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
22240 * cgraphunit.c (init_lowered_empty_function): Update.
22241 (cgraph_node::expand_thunk): Update.
22242 * cilk-common.c: Include profile-count.h
22243 * dojump.c (inv): Remove.
22244 (jumpifnot): Update.
22245 (jumpifnot_1): Update.
22246 (do_jump_1): Update.
22248 (do_jump_by_parts_greater_rtx): Update.
22249 (do_compare_rtx_and_jump): Update.
22250 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
22251 do_jump_1. do_compare_rtx_and_jump): Update prototype.
22252 * dwarf2cfi.c: Include profile-count.h
22253 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
22254 (sjlj_emit_dispatch_table): Likewise.
22255 * explow.c: Include profile-count.h
22256 * expmed.c (emit_store_flag_force): Update.
22257 (do_cmp_and_jump): Update.
22258 * expr.c (compare_by_pieces_d::generate): Update.
22259 (compare_by_pieces_d::finish_mode): Update.
22260 (emit_block_move_via_loop): Update.
22261 (store_expr_with_bounds): Update.
22262 (store_constructor): Update.
22263 (expand_expr_real_2): Update.
22264 (expand_expr_real_1): Update.
22265 * expr.h (try_casesi, try_tablejump): Update prototypes.
22266 * gimple-pretty-print.c (dump_probability): Update.
22267 (dump_profile): New.
22268 (dump_gimple_label): Update.
22269 (dump_gimple_bb_header): Update.
22270 * graph.c (draw_cfg_node_succ_edges): Update.
22271 * hsa-gen.c (convert_switch_statements): Update.
22272 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
22273 (find_if_case_1): Update.
22274 (find_if_case_2): Update.
22275 * internal-fn.c (expand_arith_overflow_result_store): Update.
22276 (expand_addsub_overflow): Update.
22277 (expand_neg_overflow): Update.
22278 (expand_mul_overflow): Update.
22279 (expand_vector_ubsan_overflow): Update.
22280 * ipa-cp.c (good_cloning_opportunity_p): Update.
22281 * ipa-split.c (split_function): Use make_single_succ_edge.
22282 * ipa-utils.c (ipa_merge_profiles): Update.
22283 * loop-doloop.c (add_test): Update.
22284 (doloop_modify): Update.
22285 * loop-unroll.c (compare_and_jump_seq): Update.
22286 (unroll_loop_runtime_iterations): Update.
22287 * lra-constraints.c (lra_inheritance): Update.
22288 * lto-streamer-in.c (input_cfg): Update.
22289 * lto-streamer-out.c (output_cfg): Update.
22290 * mcf.c (adjust_cfg_counts): Update.
22291 * modulo-sched.c (sms_schedule): Update.
22292 * omp-expand.c (expand_omp_for_init_counts): Update.
22293 (extract_omp_for_update_vars): Update.
22294 (expand_omp_ordered_sink): Update.
22295 (expand_omp_for_ordered_loops): Update.
22296 (expand_omp_for_generic): Update.
22297 (expand_omp_for_static_nochunk): Update.
22298 (expand_omp_for_static_chunk): Update.
22299 (expand_cilk_for): Update.
22300 (expand_omp_simd): Update.
22301 (expand_omp_taskloop_for_outer): Update.
22302 (expand_omp_taskloop_for_inner): Update.
22303 * omp-simd-clone.c (simd_clone_adjust): Update.
22304 * optabs.c (expand_doubleword_shift): Update.
22305 (expand_abs): Update.
22306 (emit_cmp_and_jump_insn_1): Update.
22307 (expand_compare_and_swap_loop): Update.
22308 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
22309 * predict.c (predictable_edge_p): Update.
22310 (edge_probability_reliable_p): Update.
22311 (set_even_probabilities): Update.
22312 (combine_predictions_for_insn): Update.
22313 (combine_predictions_for_bb): Update.
22314 (propagate_freq): Update.
22315 (estimate_bb_frequencies): Update.
22316 (force_edge_cold): Update.
22317 * profile-count.c (profile_count::dump): Add missing space into dump.
22318 (profile_count::debug): Add newline.
22319 (profile_count::differs_from_p): Explicitly convert to unsigned.
22320 (profile_count::stream_in): Update.
22321 (profile_probability::dump): New member function.
22322 (profile_probability::debug): New member function.
22323 (profile_probability::differs_from_p): New member function.
22324 (profile_probability::differs_lot_from_p): New member function.
22325 (profile_probability::stream_in): New member function.
22326 (profile_probability::stream_out): New member function.
22327 * profile-count.h (profile_count_quality): Rename to ...
22328 (profile_quality): ... this one.
22329 (profile_probability): New.
22330 (profile_count): Update.
22331 * profile.c (compute_branch_probabilities): Update.
22332 * recog.c (peep2_attempt): Update.
22333 * sched-ebb.c (schedule_ebbs): Update.
22334 * sched-rgn.c (find_single_block_region): Update.
22335 (compute_dom_prob_ps): Update.
22336 (schedule_region): Update.
22337 * sel-sched-ir.c (compute_succs_info): Update.
22338 * stmt.c (struct case_node): Update.
22339 (do_jump_if_equal): Update.
22340 (get_outgoing_edge_probs): Update.
22341 (conditional_probability): Update.
22342 (emit_case_dispatch_table): Update.
22343 (expand_case): Update.
22344 (expand_sjlj_dispatch_table): Update.
22345 (emit_case_nodes): Update.
22346 * targhooks.c: Update.
22347 * tracer.c (better_p): Update.
22348 (find_best_successor): Update.
22349 * trans-mem.c (expand_transaction): Update.
22350 * tree-call-cdce.c: Update.
22351 * tree-cfg.c (gimple_split_edge): Upate.
22352 (move_sese_region_to_fn): Upate.
22353 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
22354 * tree-eh.c (lower_resx): Upate.
22355 (cleanup_empty_eh_move_lp): Upate.
22356 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22357 * tree-inline.c (copy_edges_for_bb): Update.
22358 (copy_cfg_body): Update.
22359 * tree-parloops.c (gen_parallel_loop): Update.
22360 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
22361 (gimple_gen_time_profiler): Update.
22362 * tree-ssa-dce.c (remove_dead_stmt): Update.
22363 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22364 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
22365 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
22366 (unloop_loops): Update.
22367 (try_peel_loop): Update.
22368 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22369 * tree-ssa-loop-split.c (connect_loops): Update.
22370 (split_loop): Update.
22371 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22372 (hoist_guard): Update.
22373 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
22374 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
22375 (value_replacement): Update.
22376 * tree-ssa-reassoc.c (branch_fixup): Update.
22377 * tree-ssa-tail-merge.c (replace_block_by): Update.
22378 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
22379 (create_edge_and_update_destination_phis): Update.
22380 (compute_path_counts): Update.
22381 (recompute_probabilities): Update.
22382 (update_joiner_offpath_counts): Update.
22383 (freqs_to_counts_path): Update.
22384 (duplicate_thread_path): Update.
22385 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
22386 (struct switch_conv_info): Update.
22387 (gen_inbound_check): Update.
22388 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22389 (vect_do_peeling): Update.
22390 (vect_loop_versioning): Update.
22391 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22392 (optimize_mask_stores): Update.
22393 * ubsan.c (ubsan_expand_null_ifn): Update.
22394 * value-prof.c (gimple_divmod_fixed_value): Update.
22395 (gimple_divmod_fixed_value_transform): Update.
22396 (gimple_mod_pow2): Update.
22397 (gimple_mod_pow2_value_transform): Update.
22398 (gimple_mod_subtract): Update.
22399 (gimple_mod_subtract_transform): Update.
22400 (gimple_ic): Update.
22401 (gimple_stringop_fixed_value): Update.
22402 (gimple_stringops_transform): Update.
22403 * value-prof.h: Update.
22405 2017-06-29 Carl Love <cel@us.ibm.com>
22407 * config/rs6000/rs6000-c.c: Add support for built-in functions
22408 vector signed int vec_signed (vector float);
22409 vector signed long long vec_signed (vector double);
22410 vector signed int vec_signed2 (vector double, vector double);
22411 vector signed int vec_signede (vector double);
22412 vector signed int vec_signedo (vector double);
22413 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
22414 instruction generator.
22415 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22416 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
22417 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
22419 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
22420 vunsignede_v2df): Add define_expands.
22421 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
22422 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
22423 VEC_UNSIGNEDO): Add definitions.
22424 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22425 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
22426 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
22427 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
22428 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
22429 * config/rs6000/altivec.h (vec_signed, vec_signed2,
22430 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
22431 vec_unsignede, vec_unsignedo): Add builtin defines.
22432 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
22434 * doc/extend.texi: Update the built-in documentation file for the
22435 new built-in functions.
22437 2017-06-29 Richard Biener <rguenther@suse.de>
22439 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
22440 reduction chains to LOOP_VINFO_REDUCTIONS.
22441 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
22442 SLP reductions after processing reduction chains.
22444 2017-06-29 Nathan Sidwell <nathan@acm.org>
22446 * builtins.c (fold_builtin_FUNCTION): Use
22447 lang_hooks.decl_printable_name.
22449 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
22451 PR middle-end/81194
22452 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
22453 with only one label.
22454 * stmt.c (expand_case): Assert NCASES is greater than one.
22456 2017-06-29 Richard Biener <rguenther@suse.de>
22458 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
22460 (group_case_labels): Likewise.
22461 (find_taken_edge): Push sanity checking on val to workers...
22462 (find_taken_edge_cond_expr): ... here
22463 (find_taken_edge_switch_expr): ... and here, handle cases
22464 with just a default label.
22465 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
22466 (group_case_labels): Likewise.
22467 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
22468 group_case_labels does anything cleanup the CFG again.
22470 2017-06-29 Bin Cheng <bin.cheng@arm.com>
22472 PR tree-optimization/81196
22473 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
22474 exit condition comparing two IVs.
22476 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
22478 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
22479 profile to the dummy entry at the end of the list of architectures.
22480 * config/arm/arm-cpu-cdata.h: Regenerated.
22482 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22483 Michael Collison <michael.collison@arm.com>
22486 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
22488 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
22489 (*aarch64_reg_<mode>3_minus_mask): New pattern.
22490 (*aarch64_<optab>_reg_di3_mask2): New pattern.
22491 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
22492 of shift when the shift amount is masked with constant equal to
22493 the size of the mode.
22494 * config/aarch64/predicates.md (subreg_lowpart_operator): New
22497 2017-06-29 Martin Liska <mliska@suse.cz>
22499 * config/i386/i386.opt: Change range from [1,5] to [0,5].
22501 2017-06-29 Yury Gribov <tetra2005@gmail.com>
22504 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
22507 (edge_growth_cache_entry::edge_growth_cache_entry): New
22509 (reset_edge_growth_cache): Update to use constructor.
22511 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22513 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
22514 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22515 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22517 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
22519 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
22520 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
22522 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
22524 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
22525 (*-linux-uclibc*): Add t-uclibc tmake_file.
22526 * config/t-musl: New.
22527 * config/t-uclibc: New.
22529 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
22531 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
22533 (gen_comm_data): Emit architectural setting of arch_prof.
22534 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
22536 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
22537 (armv8-m.base, armv8-m.main): Likewise.
22538 * arm-protos.h (arm_build_target): Add profile field.
22539 (arch_option): Likewise.
22540 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
22542 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
22543 arm_active_target.profile.
22545 2017-06-28 Richard Biener <rguenther@suse.de>
22547 PR middle-end/81227
22548 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
22549 TYPE_OVERFLOW_WRAPS.
22550 * match.pd (negate_expr_p): Likewise.
22551 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
22552 fold_build2, not fold_binary.
22554 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22556 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
22557 Convert memory address to Pmode.
22558 (aarch64_print_operand): Assert MEM operands are always Pmode.
22560 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22563 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
22564 Remove redundant if.
22565 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
22566 * config/arm/aarch-common-protos.h
22567 (aarch_forward_to_shift_is_not_shifted_re): Remove.
22568 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
22570 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
22573 * multiple_target.c (create_dispatcher_calls): Set the default
22574 clone to be static, not public.
22576 2017-06-28 Richard Biener <rguenther@suse.de>
22578 * tree-vect-loop.c (vectorizable_reduction): Move special
22579 cond reduction IV var creation ...
22580 (vect_create_epilog_for_reduction): ... here. Remove induction_index
22581 parameter. Use STMT_VINFO_VECTYPE.
22582 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
22585 2017-06-28 Martin Liska <mliska@suse.cz>
22588 * ipa-visibility.c (non_local_p): Handle visibility.
22590 2017-06-28 Martin Liska <mliska@suse.cz>
22593 * common.opt: Add IntegerRange to various options.
22594 * opt-functions.awk (integer_range_info): New function.
22595 * optc-gen.awk: Add integer_range_info to cl_options struct.
22596 * opts-common.c (decode_cmdline_option): Handle
22597 CL_ERR_INT_RANGE_ARG.
22598 (cmdline_handle_error): Likewise.
22599 * opts.c (print_filtered_help): Show valid interval in
22600 when --help is provided.
22601 * opts.h (struct cl_option): Add range_min and range_max fields.
22602 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
22604 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
22606 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
22607 (x * C EQ/NE y * C): New transformation.
22609 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
22611 * genmultilib (combination_space): Accept '+' in option names.
22613 2017-06-28 Martin Liska <mliska@suse.cz>
22616 * asan.c (instrument_derefs): Bail out inner references
22617 that are hard register variables.
22619 2017-06-28 Jakub Jelinek <jakub@redhat.com>
22622 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
22623 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
22625 2017-06-28 Richard Biener <rguenther@suse.de>
22627 * tree-vectorizer.h (vect_get_vec_defs): Remove.
22628 (vect_get_slp_defs): Adjust.
22629 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
22631 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
22633 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
22634 get_initial_defs_for_reduction instead of vect_get_vec_defs.
22635 (vectorizable_reduction): Adjust.
22636 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
22638 (vect_get_slp_defs): Likewise.
22639 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
22640 (vectorizable_bswap): Adjust.
22641 (vectorizable_call): Likewise.
22642 (vectorizable_conversion): Likewise.
22643 (vectorizable_assignment): Likewise.
22644 (vectorizable_shift): Likewise.
22645 (vectorizable_operation): Likewise.
22646 (vectorizable_store): Likewise.
22647 (vectorizable_condition): Likewise.
22648 (vectorizable_comparison): Likewise.
22650 2017-06-28 Michael Collison <michael.collison@arm.com>
22653 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
22655 (arm_gen_movmemqi): Removed unused variable 'i'.
22656 Convert 'for' loop into 'while' loop.
22657 (arm_expand_prologue): Remove last unnecessary set of insn.
22658 (thumb_pop): Remove unused variable 'pushed_words'.
22659 (thumb_exit): Remove last unnecessary set of regs_to_pop.
22661 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22663 * config/s390/predicates.md: Use s390_rel_address_ok_p.
22664 * config/s390/s390-protos.h: Add prototype of
22665 s390_rel_address_ok_p.
22666 * config/s390/s390.c (s390_got_symbol): New function.
22667 (s390_rel_address_ok_p): New function.
22668 (legitimize_pic_address): Use s390_rel_address_ok_p.
22669 (s390_load_got): Use s390_got_symbol.
22670 (s390_option_override): Issue error if
22671 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
22672 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
22674 * config/s390/s390.opt: New option mpic-data-is-text-relative.
22676 2017-06-27 Andrew Pinski <apinski@cavium.com>
22678 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
22679 (X * copysign (1.0, X)): New pattern.
22680 (X * copysign (1.0, -X)): New pattern.
22681 (copysign (-1.0, CST)): New pattern.
22683 2017-06-27 Joseph Myers <joseph@codesourcery.com>
22685 * genmultilib (combination_space): Remove variable.
22686 Validate reuse rules against regular expression for any sequence
22687 of multilib options in any order.
22689 2017-06-27 Michael Collison <michael.collison@arm.com>
22691 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
22692 call aarch64_split_simd_combine.
22693 * (aarch64_combine_internal<mode>): Delete pattern.
22694 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
22695 Allow register and subreg operands.
22697 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22699 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
22700 specific need, just fallback on defaults.
22701 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
22703 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22704 Olivier Hainque <hainque@adacore.com>
22706 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
22708 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
22709 targets. Pick a default if no particular attempt applied.
22710 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
22713 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22715 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
22716 (x86_64-wrs-vxworks7): Likewise.
22718 2017-06-27 Marek Polacek <polacek@redhat.com>
22721 * ubsan.c (instrument_null): Check get_base_address's result for null.
22723 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22725 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
22727 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22729 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
22730 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
22731 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
22732 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
22733 New function types.
22734 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
22735 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
22736 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
22737 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
22738 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
22739 BUILT_IN_FEUPDATEENV): New builtins.
22740 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
22741 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
22742 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
22743 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
22745 (builtin_structptr_types): Adjust size.
22746 * tree.c (builtin_structptr_types): Add four entries.
22748 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22749 Olivier Hainque <hainque@adacore.com>
22751 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
22752 (TLS_SYM): New local macro, forcing reference to __tls__ on
22753 link command lines for VxWorks 7 RTPs, triggering initialization
22755 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
22756 OS features TLS support, true for RTPs on VxWorks 7.
22757 * config/vxworks.c (vxworks_override_options): Setup emutls
22760 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22762 * predict.c (test_prediction_value_range): Use -1U instead of -1
22763 to avoid narrowing conversion warning.
22764 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
22765 to avoid narrowing conversion warning.
22766 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
22768 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
22770 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22772 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
22773 64bit configurations.
22774 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
22775 (SIZE_TYPE): Likewise.
22776 * config/vxworks.c (vxworks_emutls_var_fields): Use
22777 long_unsigned_type_node instead of unsigned_type_node as the offset
22778 field type, which is "pointer" mode in emutls.c.
22780 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22783 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
22785 PR middle-end/81207
22786 * gimple-fold.c (replace_call_with_call_and_fold): Handle
22787 gimple_vuse copying separately from gimple_vdef copying.
22789 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22791 * value-prof.c (free_hist): Remove call to memset and the enclosing if
22794 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22795 Olivier Hainque <hainque@adacore.com>
22797 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
22798 for all vxworks7 targets.
22799 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
22800 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
22801 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
22802 variations for VX6/VX7 and 32/64bits later on in ...
22803 (VXWORKS_LIB_SPEC): Leverage new macros.
22804 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
22805 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
22807 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22809 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
22810 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
22812 2017-06-26 Carl Love <cel@us.ibm.com>
22814 * config/rs6000/rs6000-c.c: Add support for built-in functions
22815 vector bool char vec_reve (vector bool char);
22816 vector signed char vec_reve (vector signed char);
22817 vector unsigned char vec_reve (vector unsigned char);
22818 vector bool int vec_reve (vector bool int);
22819 vector signed int vec_reve (vector signed int);
22820 vector unsigned int vec_reve (vector unsigned int);
22821 vector bool long long vec_reve (vector bool long long);
22822 vector signed long long vec_reve (vector signed long long);
22823 vector unsigned long long vec_reve (vector unsigned long long);
22824 vector bool short vec_reve (vector bool short);
22825 vector signed short vec_reve (vector signed short);
22826 vector double vec_reve (vector double);
22827 vector float vec_reve (vector float);
22828 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
22829 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
22830 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
22831 (altivec_vreve): New pattern.
22832 * config/rs6000/altivec.h (vec_reve): New define.
22833 * doc/extend.texi (vec_rev): Update the built-in documentation file
22834 for the new built-in functions.
22836 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22838 PR tree-optimization/71815
22839 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
22841 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
22843 (slsr_process_phi): Likewise.
22844 (replace_uncond_cands_and_profitable_phis): Don't replace a
22845 multiply candidate with a stride of 1 (copy or cast).
22846 (phi_incr_cost): Call uses_consumed_by_stmt rather than
22848 (lowest_cost_path): Likewise.
22849 (total_savings): Likewise.
22851 2017-06-26 Richard Biener <rguenther@suse.de>
22854 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
22855 Use def_builtin_pure for all gather builtins.
22857 2017-06-26 Richard Biener <rguenther@suse.de>
22859 PR tree-optimization/81203
22860 * tree-tailcall.c (find_tail_calls): Do not move stmts into
22861 non-dominating BBs.
22863 2017-06-26 Marek Polacek <polacek@redhat.com>
22866 * doc/invoke.texi: Document -Wmultistatement-macros.
22868 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
22870 * doc/sourcebuild.texi (ARM-specific attributes): Document new
22871 arm_neon_ok_no_float_abi effective target.
22873 2017-06-26 Richard Biener <rguenther@suse.de>
22875 PR tree-optimization/80928
22876 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
22877 (copy_bbs): Set BB_DUPLICATED flag early.
22878 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
22880 (execute_on_shrinking_pred): Likewise.
22881 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
22882 BB_DUPLICATED blocks.
22883 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
22884 iterate over all PHIs considering removal of *gsi.
22886 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
22888 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
22891 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22893 * config/rs6000/rs6000-string.c: (expand_block_clear,
22894 do_load_for_compare, select_block_compare_mode,
22895 compute_current_alignment, expand_block_compare,
22896 expand_strncmp_align_check, expand_strn_compare,
22897 expand_block_move, rs6000_output_load_multiple)
22898 Move functions related to string/block move/compare
22899 to a separate file.
22900 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
22901 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
22902 for this function which is now used in two files.
22903 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
22904 * config.gcc: Add rs6000-string.o to extra_objs for
22905 targets powerpc*-*-* and rs6000*-*-*.
22907 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
22910 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
22911 32-bit, since indexed is not valid for DImode.
22912 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
22913 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
22914 (define_peephole2 for Altivec d-form load): Add 32-bit support.
22915 (define_peephole2 for Altivec d-form store): Likewise.
22918 * multiple_target.c (create_dispatcher_calls): Only create the
22919 dispatcher call if the function is the default clone of a
22920 versioned function.
22922 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
22924 PR middle-end/80902
22925 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
22926 a call, force the call to not be a tail call.
22928 2017-06-23 Jeff Law <law@redhat.com>
22930 * doc/contrib.texi: Add entry for Steven Pemberton's work on
22933 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
22935 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
22936 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
22937 handling for early expansion of vector shifts (sl,sr,sra,rl).
22938 (builtin_function_type): Add vector shift right instructions
22939 to the unsigned argument list.
22941 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
22943 rtl-optimizatoin/79286
22944 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
22945 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
22946 trap. PIC register plus a const unspec without offset can never trap.
22948 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
22950 * tree.h (builtin_structptr_type): New type.
22951 (builtin_structptr_types): Declare new array.
22952 * tree.c (builtin_structptr_types): New array.
22953 (free_lang_data, build_common_tree_nodes): Use it.
22955 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
22958 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
22961 2017-06-22 Matt Turner <mattst88@gmail.com>
22963 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
22964 Lake models to skylake case. Assume skylake for unknown
22965 models with clflushopt.
22967 2017-06-22 Jeff Law <law@redhat.com>
22969 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
22970 frame sizes that do not satisfy aarch64_uimm12_shift.
22972 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
22974 * profile-count.h (apply_probability,
22975 apply_scale, probability_in): Fix checks for zero.
22977 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22979 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
22980 * doc/cppdiropts.texi (-I @var{dir}): Document it.
22982 2016-06-22 Richard Biener <rguenther@suse.de>
22984 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22985 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
22986 REDUC_MAX_EXPR support.
22987 (vectorizable_reduction): Likewise.
22988 (vect_create_epilog_for_reduction): Likewise.
22990 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
22992 * match.pd (A / (1 << B) -> A >> B): New.
22993 * generic-match-head.c: Include optabs-tree.h.
22994 * gimple-match-head.c: Likewise.
22995 * optabs-tree.h (target_supports_op_p): New.
22996 * optabs-tree.c (target_supports_op_p): New.
22998 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23000 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23001 $gcc_cv_ld --help output.
23002 (gcc_cv_ld_demangle): Likewise.
23003 (gcc_cv_ld_eh_frame_hdr): Likewise.
23004 (gcc_cv_ld_pie): Likewise.
23005 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23006 (gcc_cv_ld_buildid): Likewise.
23007 (gcc_cv_ld_sysroot): Likewise.
23008 (ld_bndplt_support): Likewise.
23009 (ld_pushpopstate_support): Likewise.
23010 * configure: Regenerate.
23011 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23013 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23016 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23017 operands indexes to avoid gap between operands and match_dups.
23019 2017-06-21 Andrew Pinski <apinski@cavium.com>
23021 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23022 Increment Arith_shift and Arith_shift_reg by 1.
23023 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23025 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23026 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23027 (aarch64_strip_extend): Add new argument and test for it.
23028 (aarch64_cheap_mult_shift_p): New function.
23029 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23030 add a cost if it is true.
23031 Update calls to aarch64_strip_extend.
23032 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23034 2017-06-21 Andrew Pinski <apinski@cavium.com>
23036 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23038 (thunderxt88): Likewise.
23039 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23040 (thunderx_prefetch_tune): New variable.
23041 (thunderx2t99_prefetch_tune): Update for the correct values.
23042 (thunderxt88_tunings): New variable.
23043 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23044 generic_prefetch_tune.
23045 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23047 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23049 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23050 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23051 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23052 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23053 (aarch64_atomic_cas<mode>, GPI): Likewise.
23055 2017-06-21 Martin Liska <mliska@suse.cz>
23057 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23058 statements on cold and hot labels.
23059 * predict.c (tree_estimate_probability_bb): Remove the
23060 prediction from this place.
23062 2017-06-21 Martin Liska <mliska@suse.cz>
23064 PR tree-optimization/79489
23065 * gimplify.c (maybe_add_early_return_predict_stmt): New
23067 (gimplify_return_expr): Call the function.
23068 * predict.c (tree_estimate_probability_bb): Remove handling
23070 * predict.def: Update comment about early return predictor.
23071 * gimple-predict.h (is_gimple_predict): New function.
23072 * predict.def: Change default value of early return to 66.
23073 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23075 * passes.def: Put pass_strip_predict_hints to the beginning of
23078 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23080 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23081 FUNCTION_DECL declarations.
23082 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23084 (dwaf2out_decl): Likewise.
23085 * godump.c (go_early_global_decl): Skip call to the real debug hook
23086 for FUNCTION_DECL declarations.
23087 * passes.c (rest_of_decl_compilation): Skip call to the
23088 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23089 -fdump-go-spec is passed.
23091 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23093 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23094 Reorder for compactness.
23095 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23096 (def_builtin_pure, def_builtin_pure2): New functions.
23097 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23099 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23101 * match.pd (nop_convert): New predicate.
23102 ((A +- CST1) +- CST2): Allow some NOP conversions.
23104 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23107 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23108 with ctors/dtors if GOVD_SHARED is set.
23110 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23112 * config/aarch64/aarch64.md (movti_aarch64):
23113 Emit mov rather than orr.
23114 (movtf_aarch64): Likewise.
23115 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23116 Emit mov rather than orr.
23118 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23120 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23121 Swap alternatives, make integer dup more expensive.
23123 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23125 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23126 Return true for non-tls symbols.
23128 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23130 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23131 (cortex-a75): Likewise.
23132 (cortex-a75.cortex-a55): Likewise.
23133 * config/aarch64/aarch64-tune.md: Regenerate.
23134 * doc/invoke.texi (-mtune): Document new values for -mtune.
23136 2017-06-21 Tom de Vries <tom@codesourcery.com>
23138 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23139 stack_size feature.
23140 (Effective-Target Keywords, Other attributes): Suggest using
23141 dg-add-options stack_size feature to get stack limit in stack_size
23142 effective target documentation.
23144 2017-06-21 Julian Brown <julian@codesourcery.com>
23145 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23147 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23148 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23149 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23151 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23152 attribute type list for neon_multiply.
23153 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23154 attribute type list for neon_multiply.
23155 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23156 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23157 attribute type list for neon_multiply.
23158 * config/arm/types.md (crypto_pmull): Add.
23159 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23160 attribute type list.
23162 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23164 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23167 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23169 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23170 to make sure not to dereference a NULL cost_classes_ptr pointer.
23172 2017-06-20 Carl Love <cel@us.ibm.com>
23174 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23175 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23176 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23177 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23178 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23179 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23180 VMULOSW): New enum "unspec" values.
23181 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23182 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23183 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23184 altivec_vmulosw): New patterns.
23185 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23186 VMULOSW): Add definitions.
23188 2017-06-20 Julia Koval <julia.koval@intel.com>
23190 * config/i386/i386.c: Fix rounding expand for new pattern.
23191 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23193 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23195 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23196 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23198 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23200 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23203 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23205 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23206 architecture, then by implementer ID.
23207 * config/aarch64/aarch64-tune.md: Regenerate.
23209 2017-06-20 Richard Biener <rguenther@suse.de>
23211 PR middle-end/81097
23212 * fold-const.c (split_tree): Fold to type before negating.
23214 2017-06-20 David Malcolm <dmalcolm@redhat.com>
23216 * diagnostic-show-locus.c
23217 (selftest::test_fixit_deletion_affecting_newline): New function.
23218 (selftest::diagnostic_show_locus_c_tests): Call it.
23220 2017-06-20 Andreas Schwab <schwab@suse.de>
23223 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
23226 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23228 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
23229 __ARM_FEATURE_COPROC according to support.
23231 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23233 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23234 Rewritten to avoid overflow for > 32-bit pointers.
23237 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
23238 by removing enum keyword.
23239 (ubsan_type_descriptor): Likewise. Formatting fix.
23242 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
23243 splitter): Require TARGET_SSE2 in the condition.
23245 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
23248 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
23249 for doing vector set of SFmode on ISA 3.0.
23250 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
23251 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
23253 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
23254 SFmode value into a V4SF variable that was extracted from another
23255 V4SF variable without converting the element to double precision
23256 and back to single precision vector format.
23257 (vsx_insert_extract_v4sf_p9_2): Likewise.
23259 2017-06-19 Jakub Jelinek <jakub@redhat.com>
23261 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
23262 in UWHI to avoid undefined overflow.
23265 * ubsan.h (enum ubsan_encode_value_phase): New.
23266 (ubsan_encode_value): Change second argument to
23267 enum ubsan_encode_value_phase with default value of
23268 UBSAN_ENCODE_VALUE_GENERIC.
23269 * ubsan.c (ubsan_encode_value): Change second argument to
23270 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
23271 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
23272 create_tmp_var_raw instead of create_tmp_var and use a
23274 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
23275 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
23276 ubsan_encode_value callers.
23279 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
23280 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
23281 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
23283 2017-06-19 Richard Biener <rguenther@suse.de>
23285 PR middle-end/81118
23286 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
23287 estimates if we changed anything.
23289 2017-06-19 Richard Biener <rguenther@suse.de>
23291 PR tree-optimization/80887
23292 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
23293 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
23294 simplified lookups, then reset mprts_hook.
23295 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
23297 (try_to_simplify): Likewise.
23299 2017-06-19 Martin Liska <mliska@suse.cz>
23302 * gimplify.c (gimplify_switch_expr):
23303 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
23305 2017-06-19 Martin Liska <mliska@suse.cz>
23307 * doc/install.texi: Document that PGO runs in 4 stages.
23309 2017-06-19 Martin Liska <mliska@suse.cz>
23312 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
23313 to dispatcher function name.
23314 * multiple_target.c (replace_function_decl): New function.
23315 (create_dispatcher_calls): Redirect both edges and references.
23317 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
23319 * profile-count.c (profile_count::dump): Dump quality.
23320 (profile_count::differs_from_p): Update for unsigned val.
23321 * profile-count.h (profile_count_quality): New enum.
23322 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
23324 2017-06-19 Richard Biener <rguenther@suse.de>
23326 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
23327 struct function as arg.
23328 (estimate_numbers_of_iterations): Export overload with loop arg.
23329 (free_numbers_of_iterations_estimates_loop): Use an overload of
23330 free_numbers_of_iterations_estimates instead.
23331 * tree-cfg.c (remove_bb): Adjust.
23332 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
23333 * tree-parloops.c (gen_parallel_loop): Likewise.
23334 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
23336 (tree_unroll_loops_completely): Likewise.
23337 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
23338 Use an overload instead and export.
23339 (estimated_loop_iterations): Adjust.
23340 (max_loop_iterations): Likewise.
23341 (likely_max_loop_iterations): Likewise.
23342 (estimate_numbers_of_iterations): Take struct function as arg
23344 (loop_exits_before_overflow): Adjust.
23345 (free_numbers_of_iterations_estimates_loop): Use an overload.
23346 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
23347 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
23349 2017-06-19 Richard Biener <rguenther@suse.de>
23352 * ipa-prop.c (find_constructor_constant_at_offset): Handle
23353 RANGE_EXPR conservatively.
23355 2017-06-16 Carl Love <cel@us.ibm.com>
23357 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23358 definitions for vec_float, vec_float2, vec_floato,
23359 vec_floate built-ins.
23360 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
23361 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
23363 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
23364 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
23365 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
23366 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
23367 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
23368 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
23369 vec_floato): Add builtin defines.
23370 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
23371 Update the built-in documentation file for the new built-in
23374 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23376 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
23377 (mthumb): Mark as the negative of -marm.
23379 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23381 * doc/invoke.texi (ARM Options, -mcpu): Document supported
23383 (ARM Options, -mtune): Document that this accepts the same
23384 extension options as -mcpu.
23385 (ARM Options, -mfpu): Document addition of -mfpu=auto.
23387 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23389 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
23390 permitted extensions.
23392 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23394 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
23395 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
23396 (armv8-m.main): Add option +nodsp.
23397 * config/arm/arm-cpu-cdata.h: Regenerated.
23399 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23401 * config/arm/t-fuchsia: New file.
23402 * config.gcc (arm*-*-fuchsia*): Use it.
23404 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23406 * config/arm/t-symbian: Rewrite for new option infrastructure.
23408 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23410 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
23411 (MULTILIB_REQUIRED): Likewise.
23413 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23415 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
23416 (MULTILIB_RESUE): Likewise.
23417 (MULTILIB_MATCHES): Likewise.
23418 (MULTLIB_REQUIRED): Likewise.
23420 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23422 * config/arm/t-rtems: Rewrite for new option framework.
23424 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23426 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
23427 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
23428 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
23429 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
23430 * config/arm/t-multilib: ... here.
23431 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
23432 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
23433 armv7-a and armv8*-a when A-profile libraries have not been built.
23434 * config/arm/t-rmprofile: Rewrite.
23436 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23438 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
23439 with a backslash. Remove the backslash after substituting unescaped
23441 * doc/fragments.texi (MULTILIB_REUSE): Document it.
23443 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23445 * config.gcc: (arm*-*-*): When building a-profile libraries, force
23446 the driver to pass through the default setting of -mfloat-abi.
23447 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
23449 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
23450 (all_feat_combs): New rule.
23451 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
23453 * config/arm/t-aprofile: Rewrite.
23455 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23457 * config/arm/arm.h (FPUTYPE_AUTO): Define.
23458 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
23459 fpu is not specified by the user/command-line.
23460 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
23461 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
23462 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
23463 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
23464 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
23465 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
23467 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23469 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
23470 * config/arm/t-arm-elf: Rewritten.
23472 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23474 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
23475 have some floating-point instructions.
23476 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
23477 (TARGET_MAYBE_HARD_FLOAT): New macro.
23478 * config/arm/arm-builtins.c (arm_init_builtins): Use
23479 TARGET_MAYBE_HARD_FLOAT.
23480 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
23482 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23484 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
23485 (configargs.h): Include it.
23486 (arm_print_hint_for_fpu_option): New function.
23487 (arm_parse_fpu_option): New function.
23488 (candidate_extension): New class.
23489 (arm_canon_for_multilib): New function.
23490 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
23491 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
23492 (ARCH_CANONICAL_SPECS): New macro.
23493 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
23495 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23497 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
23498 are set after handling multilib fragments. Set target_cpu_default2
23501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23503 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
23505 (arm*-*-*): Set target_cpu_default2 to a quoted string.
23506 * config/arm/parsecpu.awk (check_cpu): Validate any extension
23508 (check_arch): Likewise.
23509 * config/arm/arm.c (arm_configure_build_target): Handle
23510 TARGET_CPU_DEFAULT being a string constant. Scan any feature
23511 options in the default.
23513 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23515 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
23516 when an option is an alias of another.
23517 * config/arm/parsecpu.awk (optalias): New parser token.
23518 (gen_comm_data): Mark non-alias options as such. Emit entries
23519 for extension aliases.
23520 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
23521 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
23522 (armv6kz, armv6zk, armv6t2): Likewise.
23523 (armv7): Make vfpv3-d16 an alias.
23524 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
23526 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
23527 Sort in canonical order.
23528 (armv8-a): Sort in canonical order.
23529 (armv8.1-a, armv8.2-a): Likewise.
23530 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
23532 (cortex-a9): Sort in canonical order.
23533 * config/arm/arm.c (selftests.h): Include it.
23534 (arm_test_cpu_arch_data): New function.
23535 (arm_run_self_tests): New function.
23536 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
23537 (targetm): Move declaration to the end of the file.
23538 * arm-cpu-cdata.h: Regenerated.
23540 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23542 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
23543 call to target_mode_check describing the type of option passed.
23544 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
23545 (arm_target_thumb_only): Use arm_parse_arch_option_name or
23546 arm_parse_cpu_option_name to match parameters against list of
23548 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
23549 arm_arch_core_flags data structure.
23550 * config/arm/arm-cpu_cdata.h: Regenerated.
23552 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23554 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
23556 (arm_print_hint_for_cpu_option): Likewise.
23557 (arm_print_hint_for_arch_option): Likewise.
23558 (arm_parse_cpu_option_name): Likewise.
23559 (arm_parse_arch_option_name): Likewise.
23560 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
23561 of entries in the all_fpus list.
23562 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
23563 (arm_parse_cpu_option_name): Declare.
23564 (arm_parse_arch_option_name): Declare.
23565 (arm_parse_option_features): Declare.
23566 (arm_intialize_isa): Declare.
23567 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
23569 (gen_comm_data): ... here. Make definitions non-static.
23570 * config/arm/arm-cpu-data.h: Regenerated.
23571 * config/arm/arm-cpu-cdata.h: Regenerated.
23573 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23575 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
23576 (cpu_arch_extension): New structure.
23577 (cpu_arch_option, arch_option, cpu_option): New structures.
23578 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
23579 architecture types.
23580 (gen_data): Generate new format data tables.
23581 * config/arm/arm.c (cpu_tune): New structure.
23582 (cpu_option, processors): Delete.
23583 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
23584 (arm_print_hint_for_cpu_option): ... this and ...
23585 (arm_print_hint_for_arch_option): ... this.
23586 (arm_parse_arch_cpu_name): Delete. Replace with ...
23587 (arm_parse_cpu_option_name): ... this and ...
23588 (arm_parse_arch_option_name): ... this.
23589 (arm_unrecognized_feature): Change type of target parameter to
23591 (arm_parse_arch_cpu_features): Delete. Replace with ...
23592 (arm_parse_option_features): ... this.
23593 (arm_configure_build_target): Rework to use new configuration data
23595 (arm_print_tune_info): Rework for new configuration data tables.
23596 * config/arm/arm-cpu-data.h: Regenerated.
23597 * config/arm/arm-cpu.h: Regenerated.
23599 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23601 * Makefile.in (OBJS): Move sbitmap.o from here ...
23602 (OBJS-libcommon): ... to here.
23604 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23606 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
23607 (ISA_ALL_CRYPTO): New macro.
23608 (ISA_ALL_SIMD): New macro
23609 (ISA_ALL_FP): New macro.
23610 * config/arm/arm.c (fpu_bitlist): Update initializer.
23611 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
23613 (arm9e): Add fpu. Add option for nofp
23614 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
23615 (arm926ej-s, arm1026ej-s): Likewise.
23616 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
23617 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
23618 neon-fp16, neon-vfpv4, nofp and nosimd.
23619 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
23620 (cortex-a8): Add fpu. Add option for nofp.
23621 (cortex-a9): Add fpu. Add options for nosimd and nofp.
23622 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
23623 (cortex-r4f): Add fpu.
23624 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
23625 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
23627 (cortex-r8): Likewise.
23628 (cortex-m4): Add fpu. Add option for nofp.
23629 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
23630 (cortex-a17.cortex-a7): Likewise.
23631 (cortex-a32): Add fpu. Add options for crypto and nofp.
23632 (cortex-a35, cortex-a53): Likewise.
23633 (cortex-a57): Add fpu. Add option for crypto.
23634 (cortex-a72, cortex-a73): Likewise.
23635 (exynos-m1): Likewise.
23636 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
23637 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
23638 (cortex-m33): Add fpu. Add option for nofp.
23639 * config/arm/arm-cpu-cdata.h: Regenerated
23640 * config/arm/arm-cpu-data.h: Regenerated.
23642 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23644 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
23645 (armv5te, armv5tej): Likewise.
23646 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
23647 (armv7): Add options fp and vfpv3-d16.
23648 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
23649 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
23651 (armv7ve): Likewise.
23652 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
23653 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
23654 (armv8-a): Add nocrypto option.
23655 (armv8.1-a, armv8.2-a): Likewise.
23656 (armv8-m.main): add options fp, fp.dp and nofp.
23658 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23660 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
23662 (armv8-a+crc): Delete.
23663 (armv8.1-a): Add options simd, crypto and nofp.
23664 (armv8.2-a): Add options fp16, simd, crypto and nofp.
23665 (armv8.2-a+fp16): Delete.
23666 (armv8-m.main): Add option dsp.
23667 (armv8-m.main+dsp): Delete.
23668 (cortex-a8): Add fpu. Add nofp option.
23669 (cortex-a9): Add fpu. Add nofp and nosimd options.
23670 * config/arm/parsecpu.awk (gen_data): Generate option tables and
23671 link to main cpu and architecture data structures.
23672 (gen_comm_data): Only put isa attributes from the main architecture
23674 (option): New statement for architecture and CPU entries.
23675 * arm.c (struct cpu_option): New structure.
23676 (struct processors): Add entry for options.
23677 (arm_unrecognized_feature): New function.
23678 (arm_parse_arch_cpu_name): Ignore any characters after the first
23680 (arm_parse_arch_cpu_feature): New function.
23681 (arm_configure_build_target): Separate out any CPU and architecture
23682 features and parse separately. Don't error out if -mfpu=auto is
23683 used with only an architecture string.
23684 (arm_print_asm_arch_directives): New function.
23685 (arm_file_start): Call it.
23686 * config/arm/arm-cpu-cdata.h: Regenerated.
23687 * config/arm/arm-cpu-data.h: Likewise.
23688 * config/arm/arm-tables.opt: Likewise.
23690 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23692 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
23693 assembler when it is not -mfpu=auto.
23695 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23697 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
23698 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
23699 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
23700 (ASM_CPU_SPEC): Rewrite.
23701 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
23702 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
23703 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
23704 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
23705 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
23706 copied string is NUL-terminated. Also strip any characters prefixed
23708 (arm_rewrite_selected_arch): New function.
23709 (arm_rewrite_march): New function.
23711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23713 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
23714 (x_arm_cpu_string, x_arm_tune_string): Likewise.
23715 (march, mcpu, mtune): Convert to string-based options.
23716 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
23717 (arm_parse_arch_cpu_name): New function.
23718 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
23719 identify selected architecture or CPU.
23720 (arm_option_save): New function.
23721 (TARGET_OPTION_SAVE): Redefine.
23722 (arm_option_restore): Restore string options.
23723 (arm_option_print): Print string options.
23725 2017-06-16 Martin Sebor <msebor@redhat.com>
23727 PR tree-optimization/80933
23728 PR tree-optimization/80934
23729 * builtins.c (fold_builtin_3): Do not handle bcmp here.
23730 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
23731 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
23732 (gimple_fold_builtin): Call them.
23734 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23736 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
23737 as unlikely; update profile.
23739 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23741 * predict.c (force_edge_cold): Handle declaring edges impossible
23744 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23746 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
23748 (try_unroll_loop_completely): Fix reporting.
23750 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23752 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
23754 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
23757 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
23758 if given a non-constant argument for an intrinsic which requires a
23761 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23763 * profile.c (compare_freqs): New function.
23764 (branch_prob): Sort edge list.
23765 (find_spanning_tree): Assume that the list is priority sorted.
23767 2017-06-16 Richard Biener <rguenther@suse.de>
23769 PR tree-optimization/81090
23770 * passes.def (pass_record_bounds): Remove.
23771 * tree-pass.h (make_pass_record_bounds): Likewise.
23772 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
23773 make_pass_record_bounds): Likewise.
23774 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
23775 not free niter estimates at the beginning but at the end.
23776 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
23778 2017-06-16 Richard Biener <rguenther@suse.de>
23780 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
23781 initializer to workaround ICE in host GCC 4.8.
23783 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23785 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
23787 (clone_inlined_nodes): Update.
23789 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23791 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
23792 prefetch settings, and enable prefetching by default at -O3.
23794 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23796 * config/aarch64/aarch64.c (aarch64_override_options_internal):
23797 Set flag_prefetch_loop_arrays according to tuning data.
23799 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23801 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
23802 New tune structure.
23803 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
23804 [Unrelated to main purpose of the patch] Place the pointer field last
23805 to enable type checking errors when tune structure are wrongly merged.
23806 * config/aarch64/aarch64.c (generic_prefetch_tune,)
23807 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
23808 (thunderx2t99_prefetch_tune): New tune constants.
23809 (tune_params *_tunings): Update all tunings (no functional change).
23810 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
23811 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
23812 from tunings structures.
23814 2017-06-16 Jakub Jelinek <jakub@redhat.com>
23817 * ubsan.c (instrument_null): Add T argument, use it instead
23818 of computing it based on IS_LHS.
23819 (instrument_object_size): Likewise.
23820 (pass_ubsan::execute): Adjust instrument_null and
23821 instrument_object_size callers to pass gimple_get_lhs or
23822 gimple_assign_rhs1 result to it. Use instrument_null instead of
23823 calling get_base_address and instrument_mem_ref. Handle
23824 aggregate call arguments for object-size sanitization.
23826 2017-06-16 Yury Gribov <tetra2005@gmail.com>
23828 PR tree-optimization/81089
23829 * tree-vrp.c (is_masked_range_test): Validate operands of
23832 2017-06-15 Martin Sebor <msebor@redhat.com>
23835 * dumpfile.c (dump_register): Avoid calling memset to initialize
23836 a class with a default ctor.
23837 * gcc.c (struct compiler): Remove const qualification.
23838 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
23839 * hash-table.h: Ditto.
23840 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
23842 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
23843 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
23845 * params.h (struct param_info): Make struct members non-const.
23846 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
23847 with default initialization.
23848 * vec.h (vec_copy_construct, vec_default_construct): New helper
23850 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
23851 with vec_copy_construct.
23852 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
23853 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
23854 * doc/invoke.texi (-Wclass-memaccess): Document.
23856 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23858 * emit-rtl.h (is_leaf): Update comment about local
23859 register allocator.
23861 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23864 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
23865 for a variable to have a section before checking if the section has a
23867 Set section to.persistent if persistent attribute is set.
23868 Warn if .persistent attribute is used on an automatic variable.
23870 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
23872 PR rtl-optimization/80474
23873 * reorg.c (update_block): Do not ignore instructions in a delay slot.
23875 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
23877 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
23880 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
23882 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
23883 (casesi): Emit bounds checking as RTL.
23884 (casesi_internal_mips16_<mode>): Remove bounds checking.
23886 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
23888 * config/xtensa/xtensa.c (xtensa_option_override): Append
23889 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
23890 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
23891 xtensa_doloop_hooks): Define unconditionally.
23892 (xtensa_reorg_loops): Only call reorg_loops in the presence of
23894 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
23895 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
23896 for it in xtensa_option_override.
23897 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
23898 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
23900 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
23902 * doc/cppopts.texi: Document '-' special value to -MF.
23904 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
23906 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
23907 (cortex_a53_fconst): Likewise.
23908 (cortex_a53_fpmul): Likewise.
23909 (cortex_a53_f_load_64): Likewise.
23910 (cortex_a53_f_load_many): Likewise.
23911 (cortex_a53_advsimd_alu): Likewise.
23912 (cortex_a53_advsimd_alu_q): Likewise.
23913 (cortex_a53_advsimd_mul): Likewise.
23914 (cortex_a53_advsimd_mul_q): Likewise.
23915 (fpmac bypass): Add new bypass for fpmac-fpmac case.
23916 Add missing fmul, r2f_cvt and fconst cases.
23918 2017-06-14 Richard Biener <rguenther@suse.de>
23920 PR middle-end/81088
23921 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
23923 (fold_binary_loc): When associating do not treat pre-existing
23924 TREE_OVERFLOW on literal constants as a reason to allow
23925 TREE_OVERFLOW on associated literal constants.
23927 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23929 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
23930 (MASK_FEATURES): New macro.
23931 * config/sparc/sparc.c (sparc_option_override): Remove the special
23932 handling of -mfpu and generalize it to all MASK_FEATURES switches.
23934 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23936 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
23937 a division of 0 if non-call exceptions are enabled.
23939 2017-06-14 Andrew Pinski <apinski@cavium.com>
23940 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23943 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
23944 Improve vector initialization code gen for only variable case.
23946 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23948 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
23950 2017-06-14 Richard Biener <rguenther@suse.de>
23952 PR tree-optimization/81083
23953 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
23956 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23958 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
23959 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
23960 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
23961 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
23962 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
23963 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
23965 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23967 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
23968 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
23970 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23972 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
23974 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23976 * config/rs6000/t-rtems: Don't handle SPE.
23978 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23980 * config/rs6000/t-linux: Don't handle SPE.
23982 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23984 * config/rs6000/eabispe.h: Delete file.
23986 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23988 * config/rs6000/t-spe: Delete file.
23990 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23992 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
23993 (rs6000_legitimate_offset_address_p): Return false for anything in
23994 V2SImode or V2SFmode.
23996 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23998 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
23999 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24000 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24002 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24003 (rs6000_legitimate_offset_address_p): Ditto.
24004 (rs6000_emit_move): Ditto.
24005 (rs6000_init_builtins): Remove V4HI_type_node.
24007 2017-06-13 Martin Liska <mliska@suse.cz>
24010 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24011 (gate_asan): Likewise.
24012 * asan.h (asan_no_sanitize_address_p): Remove the function.
24013 (sanitize_flags_p): New function.
24014 * builtins.def: Fix coding style.
24015 * common.opt: Use renamed enum value.
24016 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24017 * doc/extend.texi: Document no_sanitize attribute.
24018 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24019 to SANITIZE_UNDEFINED_NONDEFAULT.
24020 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24021 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24022 Use sanitize_flags_p.
24023 * gimplify.c (gimplify_function_tree): Likewise.
24024 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24025 * opts.c (parse_no_sanitize_attribute): New function.
24026 (common_handle_option): Use renamed enum value.
24027 * opts.h (parse_no_sanitize_attribute): Declare.
24028 * tree.c (sanitize_flags_p): New function.
24029 * tree.h: Declared here.
24030 * tsan.c: Use sanitize_flags_p.
24031 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24032 (instrument_mem_ref): Likewise.
24033 (instrument_bool_enum_load): Likewise.
24034 (do_ubsan_in_current_function): Remove the function.
24035 (pass_ubsan::execute): Use sanitize_flags_p.
24036 * ubsan.h: Remove do_ubsan_in_current_function
24037 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24038 (dump_function_to_file): Use it here.
24040 2017-06-13 Martin Jambor <mjambor@suse.cz>
24042 PR tree-optimization/80803
24043 PR tree-optimization/81063
24044 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24045 (propagate_subaccesses_across_link): Enqueue subtree whenever
24046 necessary instead of relying on the caller.
24048 2017-06-13 Martin Jambor <mjambor@suse.cz>
24050 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24051 that have a first_link.
24052 (sort_and_splice_var_accesses): Do not check first_link before
24054 (subtree_mark_written_and_enqueue): Likewise.
24055 (propagate_all_subaccesses): Likewise and do not stop at first
24056 parent with a first_link.
24058 2017-06-13 Martin Jambor <mjambor@suse.cz>
24060 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24063 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24065 * match.pd: New pattern.
24067 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24069 * tree-vrp.c (is_masked_range_test): New function.
24070 (register_edge_assert_for): Determine ranges for
24073 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24075 PR tree-optimization/67328
24076 * fold-const.c (maskable_range_p): New function.
24077 (build_range_check): Generate bittests if possible.
24079 2017-06-13 Martin Liska <mliska@suse.cz>
24081 * gimple-pretty-print.c (dump_probability): Add new argument.
24082 (dump_edge_probability): Dump both probability and count.
24083 (dump_gimple_label): Likewise.
24084 (dump_gimple_bb_header): Likewise.
24086 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24089 * config/avr/avr-devices.c: Fix indentation.
24090 * config/avr/gen-avr-mmcu-specs.c: Dito.
24092 2017-06-13 Richard Biener <rguenther@suse.de>
24094 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24095 instead get vector type from stmt_info.
24096 (vectorizable_reduction): Adjust. Remove dead code.
24098 2017-06-13 Richard Biener <rguenther@suse.de>
24100 PR middle-end/81065
24101 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24102 case of C * (x * C2 + C3).
24103 (fold_addr_of_array_ref_difference): Properly fold index difference.
24105 2017-06-12 David S. Miller <davem@davemloft.net>
24108 * config/sparc/sparc.md (return expander): Emit frame blockage if
24109 function uses alloca.
24111 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24113 * combine.c (make_field_assignment): Check len rather than the mode
24114 precision when calling force_to_mode.
24116 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24118 Support multilibs and devices that see flash in RAM address range.
24121 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24122 (avr_mcu_t) <flash_pm_offset>: New field.
24123 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24124 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24125 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24126 (AVR_TINY_PM_OFFSET): Remove macro.
24127 * config/avr/avr.opt (-mshort-calls): New option.
24128 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24129 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24130 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24131 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24132 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24133 instead of avr_arch->have_jmp_call.
24134 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24135 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24136 avr_arch->flash_pm_offset to define.
24137 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24138 new field flash_pm_offset. Add entry for avrxmega3.
24139 (avr_texinfo): Add entry for avrxmega3.
24140 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24141 attiny212, attiny214,
24142 attiny412, attiny414, attiny416, attiny417,
24143 attiny814, attiny816, attiny817,
24144 attiny1614, attiny1616, attiny1617,
24145 attiny3214, attiny3216, attiny3217.
24146 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24147 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24148 (avr_print_operand_address) [AVR_TINY]: Same.
24149 (avr_asm_init_sections) <readonly_data_section>: Only patch
24150 callback if avr_arch->flash_pm_offset = 0.
24151 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24152 for rodata if avr_arch->flash_pm_offset != 0.
24153 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24154 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24155 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24156 (m_options): Append opt_rcall.
24157 (m_dirnames): Append dir_rcall.
24158 * config/avr/t-multilib: Regenerate.
24160 * configure.ac [target=avr]: Check whether avrxmega3 default
24161 linker description file works as needed.
24162 * configure: Regenerate.
24163 * doc/avr-mmcu.texi: Regenerate.
24164 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24165 <__AVR_ARCH__>: Document avrxmega3 and 103.
24166 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24167 <__AVR_SHORT_CALLS__>: Document it.
24168 <__AVR_PM_BASE_ADDRESS__>: Document it.
24169 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24170 (AVR Variable Attributes) <progmem>: Document this is
24171 not needed for avrxmega3.
24172 (AVR Named Address Spaces) <__flash>: Dito.
24174 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24176 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24178 2017-06-12 Doug Rupp <rupp@adacore.com>
24180 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24181 Append vxworks-stdint.h to the tm_file list.
24182 * config/vxworks-stdint.h: New file.
24184 2017-06-12 Martin Liska <mliska@suse.cz>
24186 PR tree-optimization/81041
24187 * tree-profile.c (gimple_gen_ic_func_profiler):
24188 Create an extra BB in profile-generate
24189 (gimple_gen_time_profiler): Likewise.
24191 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24193 PR tree-optimization/81003
24194 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24195 (update_range_test): Use it instead of force_gimple_operand_gsi.
24197 2017-06-12 Richard Biener <rguenther@suse.de>
24199 PR tree-optimization/81053
24200 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24201 with backedge value not defined in loop. Simplify def stmt
24204 2017-06-11 Tom de Vries <tom@codesourcery.com>
24207 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24209 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24210 nvptx_cannot_force_const_mem.
24212 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24214 * opts.c (finish_options): Move test for flag_split_stack after
24215 it has been initialized.
24217 2017-06-11 Jason Merrill <jason@redhat.com>
24219 * tree.h (id_equal): New.
24220 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
24221 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
24222 instead of strcmp of IDENTIFIER_POINTER.
24224 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24226 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
24227 (mark_all_inlined_calls_cdtor): Fix formating.
24228 (inline_transform): Rescale profile before inlining.
24230 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24232 * cgraph.h (cgraph_edge::clone): Update prototype.
24233 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
24234 (cgraph_node::create_clone): Update.
24235 (cgraph_node::create_version_clone): Update.
24236 * tree-inline.c (copy_bb): Update.
24237 (expand_call_inline): Update.
24239 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24241 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
24242 factored out from ...
24243 (rs6000_emit_prologue): ... here.
24245 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24247 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
24248 factored out from ...
24249 (rs6000_emit_prologue): ... here.
24251 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24253 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
24255 (handle_missing_profiles): Fix computation of tp_first_run.
24256 (counts_to_freqs): Do not touch freqs when count is 0.
24258 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24260 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
24263 2017-06-10 Tom de Vries <tom@codesourcery.com>
24265 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24266 attributes): Document signal effective target.
24268 2017-06-10 Tom de Vries <tom@codesourcery.com>
24270 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24271 Document effective target stack_size.
24273 2017-06-09 David Malcolm <dmalcolm@redhat.com>
24275 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
24276 to the edit_context if they can be auto-applied.
24278 2017-06-9 Ian Lance Taylor <iant@golang.org>
24280 * opts.c (finish_options): If -fsplit-stack, disable implicit
24281 -forder-blocks-and-partition.
24282 * doc/invoke.texi (Optimize Options): Document that when using
24283 -fsplit-stack -forder-blocks-and-partition is not implicitly
24286 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24288 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
24289 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
24290 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
24291 * builtins.def (abort, trap, unreachable): Declare cold.
24292 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
24293 * tree-core.h (ECF_COLD): New.
24294 * tree.c (set_call_expr_flags): Handle ECF_COLD.
24295 (build_common_builtin_nodes): Mark unreachable and abort as cold.
24297 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24299 * predict.c (unlikely_executed_stmt_p): Cleanup.
24301 2017-06-09 Richard Biener <rguenther@suse.de>
24303 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
24304 model if the ref is always written to.
24306 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24308 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
24310 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24312 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
24315 2017-06-09 Tom de Vries <tom@codesourcery.com>
24318 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
24319 "target cannot support label values" when encountering LABEL_REF.
24321 2017-06-09 Martin Liska <mliska@suse.cz>
24323 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
24324 (gimple_gen_ic_func_profiler): Emit direct comparison
24325 of __gcov_indirect_call_callee with NULL.
24326 (gimple_gen_time_profiler): Change probability from
24327 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
24329 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24331 * profile.c (edge_gcov_counts): Turn to pointer.
24332 (compute_branch_probabilities, compute_branch_probabilities): Update.
24333 (branch_prob): Do not clear edge_gcov_count.
24334 * profile.h (edge_gcov_counts): Turn to pointer.
24335 (edge_gcov_count): Update.
24337 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24339 * gimple.h (gimple_check_failed): Mark cold.
24341 2017-06-09 Richard Biener <rguenther@suse.de>
24343 PR tree-optimization/66623
24344 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
24345 refactor check_reduction into two parts, properly computing
24346 whether we have to check reduction validity for outer loop
24349 2017-06-09 Richard Biener <rguenther@suse.de>
24351 PR tree-optimization/79483
24352 * graphite-scop-detection.c (order): New global.
24353 (get_order): Compute bb to order mapping that satisfies code
24354 generation constraints.
24355 (cmp_pbbs): New helper.
24356 (build_scops): Start domwalk at entry block, sort generated
24359 2017-06-09 Richard Biener <rguenther@suse.de>
24361 PR middle-end/81007
24362 * ipa-polymorphic-call.c
24363 (ipa_polymorphic_call_context::restrict_to_inner_class):
24364 Skip FIELD_DECLs with error_mark_node type.
24365 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
24368 2017-06-09 Martin Liska <mliska@suse.cz>
24370 * predict.c (struct branch_predictor): New struct.
24371 (test_prediction_value_range): New test.
24372 (predict_c_tests): New function.
24373 * selftest-run-tests.c (selftest::run_tests): Run the function.
24374 * selftest.h: Declare new tests.
24376 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
24379 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
24380 gen_add3_insn did not fail.
24381 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
24382 r0, construct that number in a temporary reg and add that reg to r0.
24383 If asked to put the result in r0 as well, fail.
24385 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
24387 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24388 for early expansion of vec_eqv.
24390 2017-06-08 Jakub Jelinek <jakub@redhat.com>
24392 PR middle-end/81005
24393 * ubsan.c (instrument_null): Avoid pointless code temporary.
24394 (pass_ubsan::execute): Instrument aggregate arguments of calls.
24396 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
24400 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
24403 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
24406 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24408 * predict.c (unlikely_executed_edge_p): Move ahead.
24409 (probably_never_executed_edge_p): Use it.
24411 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24413 PR middle-end/79988
24414 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
24415 gimple_call_builtin_p call.
24417 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24419 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
24420 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
24421 rtl_check_failed_type2, rtl_check_failed_code1,
24422 rtl_check_failed_code2, rtl_check_failed_code_mode,
24423 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
24424 rtvec_check_failed_bounds, rtl_check_failed_flag,
24425 _fatal_insn_not_found, _fatal_insn): Likewise.
24426 * tree.h (tree_contains_struct_check_failed,
24427 tree_check_failed, tree_not_check_failed,
24428 tree_class_check_failed, tree_range_check_failed,
24429 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
24430 tree_vec_elt_check_failed, phi_node_elt_check_failed,
24431 tree_operand_check_failed, omp_clause_check_failed,
24432 omp_clause_operand_check_failed, omp_clause_range_check_failed):
24435 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24437 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
24438 flag_branch_probabilities.
24439 * ipa-inline.c (edge_badness): Likewise.
24440 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
24441 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
24442 * predict.c (maybe_hot_frequency_p): Likewise.
24443 (probably_never_executed): Likewise.
24444 * sched-ebb.c (schedule_ebbs): Likewise.
24445 * sched-rgn.c (find_single_block_region): Likewise.
24446 * tracer.c (tail_duplicate): Likewise.
24448 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24450 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
24451 longer requires x_flag_profile_use.
24453 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24455 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
24456 instead of flag_reorder_blocks_and_partition.
24457 * dbxout.c (dbxout_function_end): Likewise.
24458 * dwarf2out.c (gen_subprogram_die): Likewise.
24459 * haifa-sched.c (sched_create_recovery_edges): Likewise.
24460 * hw-doloop.c (reorg_loops): Likewise.
24461 * varasm.c (assemble_start_function,
24462 assemble_end_function): Likewise.
24463 (decide_function_section): Do not check for
24464 flag_reorder_blocks_and_partition.
24466 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24468 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24470 (chkp_get_hard_register_fake_addr_expr): Ditto.
24471 (chkp_build_addr_expr): Add check for hard reg case.
24472 (chkp_parse_array_and_component_ref): Ditto.
24473 (chkp_find_bounds_1): Ditto.
24474 (chkp_process_stmt): Don't generate bounds store for
24477 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24479 * predict.c (maybe_hot_bb_p): Do not check profile status.
24480 (maybe_hot_edge_p): Likewise.
24481 (probably_never_executed): Check for zero counts even if profile
24483 (unlikely_executed_edge_p): New function.
24484 (unlikely_executed_stmt_p): New function.
24485 (unlikely_executed_bb_p): New function.
24486 (set_even_probabilities): Use unlikely predicates.
24487 (combine_predictions_for_bb): Likewise.
24488 (predict_paths_for_bb): Likewise.
24489 (predict_paths_leading_to_edge): Likewise.
24490 (determine_unlikely_bbs): New function.
24491 (estimate_bb_frequencies): Use it.
24492 (compute_function_frequency): Use zero counts even if profile is
24494 * profile-count.h: Fix typo.
24496 2017-08-08 Julia Koval <julia.koval@intel.com>
24498 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
24499 _mm512_mask_cvtsepi16_storeu_epi8,
24500 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
24501 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
24502 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
24503 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
24504 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
24505 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
24506 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
24507 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
24508 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
24509 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
24510 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
24511 __builtin_ia32_pmovuswb256mem_mask,
24512 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
24513 __builtin_ia32_pmovwb512mem_mask): New builtins.
24515 2017-08-08 Julia Koval <julia.koval@intel.com>
24517 PR target/73350,80862
24518 * config/i386/subst.md (round): Fix round pattern.
24519 * config/i386/i386.c (ix86_erase_embedded_rounding):
24520 Fix erasing rounding for the fixed pattern.
24522 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24524 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
24526 2017-06-08 Martin Liska <mliska@suse.cz>
24528 PR gcov-profile/80911
24529 * gcov.c (block_info::block_info): New constructor.
24531 2017-06-07 Carl Love <cel@us.ibm.com>
24533 * config/rs6000/rs6000-c: The return type of the following
24534 built-in functions was implemented as int not long long. Fix sign
24535 of return value for the unsigned version of vec_mulo and vec_mule.
24536 vector unsigned long long vec_bperm (vector unsigned long long,
24537 vector unsigned char)
24538 vector signed long long vec_mule (vector signed int,
24540 vector unsigned long long vec_mule (vector unsigned int,
24541 vector unsigned int)
24542 vector signed long long vec_mulo (vector signed int,
24544 vector unsigned long long vec_mulo (vector unsigned int,
24545 vector unsigned int)
24546 * doc/extend.texi: Fix the documentation for the built-in
24549 2017-06-07 Carl Love <cel@us.ibm.com>
24552 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
24555 2017-06-07 Carl Love <cel@us.ibm.com>
24557 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
24558 support, Generate doublehv for signed int/float for BE case only.
24560 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
24562 * doc/invoke.texi (mcx16): Rewrite.
24564 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24566 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
24567 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
24568 *mov<mode>_softfloat, and an anonymous splitter): Use
24569 nonimmediate_operand instead of rs6000_nonimmediate_operand.
24571 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24573 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
24575 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24576 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
24577 (rs6000_debug_reg_global): Adjust.
24578 (rs6000_init_hard_regno_mode_ok): Adjust.
24579 (rs6000_dbx_register_number): Adjust.
24580 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
24581 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24582 Remove SPE_ACC and SPEFSCR.
24583 (REG_ALLOC_ORDER): Ditto.
24584 (FRAME_POINTER_REGNUM): Change to 111.
24585 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
24586 (REG_CLASS_NAMES): Ditto.
24587 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
24588 (REGISTER_NAMES): Ditto.
24589 (ADDITIONAL_REG_NAMES): Ditto.
24590 (rs6000_reg_names): Ditto.
24591 * config/rs6000/rs6000.md: Renumber some register number
24594 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24596 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
24598 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24599 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
24601 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
24602 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24603 Delete the SPE high registers.
24604 (REG_ALLOC_ORDER): Ditto.
24605 (enum reg_class): Remove SPE_HIGH_REGS.
24606 (REG_CLASS_NAMES): Ditto.
24607 (REG_CLASS_CONTENTS): Delete the SPE high registers.
24608 (REGISTER_NAMES): Ditto.
24609 (rs6000_reg_names): Ditto.
24610 * doc/tm.texi.in: Remove SPE as example.
24611 * doc/tm.texi: Regenerate.
24613 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24615 * config/rs6000/8540.md (ppc8540_brinc): Delete.
24616 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
24617 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
24618 * config/rs6000/rs6000.md (type): Remove "brinc".
24620 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24622 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
24623 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
24624 * config/rs6000/linuxspe.h: Delete file.
24625 * config/rs6000/rs6000.md: Don't include spe.md.
24626 * config/rs6000/spe.h: Delete file.
24627 * config/rs6000/spe.md: Delete file.
24628 * config/rs6000/t-rs6000: Remove spe.md.
24630 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24632 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
24633 (reg_or_none500mem_operand): Delete.
24634 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
24635 instead of reg_or_none500mem_operand.
24637 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24639 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24640 handling of SPE flags.
24641 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
24643 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24645 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
24647 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
24648 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
24649 paired_divv2sf3): Similar.
24650 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
24651 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
24652 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
24654 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
24655 Rename the paired_* instruction patterns.
24656 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
24658 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
24659 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
24660 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
24661 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
24662 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
24663 PAIRED_VECTOR_MODE.
24664 (struct machine_function): Delete field spe_insn_chain_scanned_p.
24665 (spe_func_has_64bit_regs_p): Delete.
24666 (spe_expand_predicate_builtin): Delete.
24667 (spe_expand_evsel_builtin): Delete.
24668 (TARGET_DWARF_REGISTER_SPAN): Do not define.
24669 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
24670 (invalid_e500_subreg): Delete.
24671 (rs6000_legitimize_address): Always force_reg op2 as well, for
24672 paired single memory accesses.
24673 (rs6000_member_type_forces_blk): Delete.
24674 (rs6000_spe_function_arg): Delete.
24675 (rs6000_expand_unop_builtin): Delete SPE handling.
24676 (rs6000_expand_binop_builtin): Ditto.
24677 (spe_expand_stv_builtin): Delete.
24678 (bdesc_2arg_spe): Delete.
24679 (spe_expand_builtin): Delete.
24680 (spe_expand_predicate_builtin): Delete.
24681 (spe_expand_evsel_builtin): Delete.
24682 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
24683 (spe_init_builtins): Delete.
24684 (spe_func_has_64bit_regs_p): Delete.
24685 (savres_routine_name): Delete "info" parameter. Adjust callers.
24686 (rs6000_emit_stack_reset): Ditto.
24687 (rs6000_dwarf_register_span): Delete.
24688 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
24689 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
24690 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
24692 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
24694 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
24695 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
24696 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
24697 mulv2sf3, divv2sf3): Delete expanders.
24699 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24701 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
24703 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24705 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
24706 * config/rs6000/rs6000.c: Ditto.
24708 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24710 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
24711 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
24712 comparison_operator.
24714 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24716 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
24717 * config/rs6000/rs6000.opt: Ditto.
24718 * config/rs6000/t-rtems: Ditto.
24720 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24722 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
24723 TARGET_E500_SINGLE by 0, simplify.
24724 * config/rs6000/rs6000.c: Ditto.
24725 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
24726 (spe_build_register_parallel): Delete.
24727 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
24728 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
24729 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
24730 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
24731 (E500_CONVERT): Delete.
24732 * config/rs6000/spe.md: Remove many patterns and all define_constants.
24734 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24736 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
24737 * config/rs6000/dfp.md: Ditto.
24738 (negdd2, *negdd2_fpr): Merge.
24739 (absdd2, *absdd2_fpr): Merge.
24740 (negtd2, *negtd2_fpr): Merge.
24741 (abstd2, *abstd2_fpr): Merge.
24742 * config/rs6000/e500.h: Delete file.
24743 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
24744 TARGET_FPRS by 1 and simplify.
24745 * config/rs6000/rs6000-c.c: Ditto.
24746 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
24747 TARGET_DF_SPE by 0.
24748 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
24750 * config/rs6000/rs6000.md: Ditto.
24751 (floatdidf2, *floatdidf2_fpr): Merge.
24752 (move_from_CR_gt_bit): Delete.
24753 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
24754 (E500_CR_IOR_COMPARE): Delete.
24755 (All patterns that require !TARGET_FPRS): Delete.
24756 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
24758 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24760 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
24762 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24764 * graphds.c (add_edge): Intitialize edge's attached data.
24765 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
24766 pointer parameter. Call pointed function on each edge during
24767 graph traversing. Skip traversing the edge when the function
24769 (graphds_dfs, graphds_scc): Ditto.
24770 (for_each_edge): New parameter. Pass the new parameter to callback
24772 * graphds.h (skip_edge_callback): New function pointer type.
24773 (graphds_dfs, graphds_scc): New function pointer parameter.
24774 (graphds_edge_callback, for_each_edge): New parameter.
24776 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24778 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
24779 out code checking if runtime alias check is possible to below ...
24780 Call the new function.
24781 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
24782 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
24784 2017-06-07 Marek Polacek <polacek@redhat.com>
24787 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
24788 TYPE_OVERFLOW_WRAPS check.
24790 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24792 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
24793 if versioning is required.
24794 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
24795 peeling with the check for versioning.
24797 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24799 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
24800 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
24801 Set true to new parameter if new ssa variable is defined.
24802 (vect_gen_vector_loop_niters): Refactor. Set range information
24803 for the new vector loop bound variable.
24804 (vect_do_peeling): Ditto.
24806 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24808 * tree-affine.c (ssa.h): Include header file.
24809 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
24810 has wrapping overflow behavior.
24812 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24814 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
24816 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24818 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
24819 (tree_to_aff_combination): ... here.
24821 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24823 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
24824 reg_pressure model function.
24825 (ivopts_global_cost_for_size): Delete.
24826 (determine_set_costs, iv_ca_recount_cost): Call new model function
24827 ivopts_estimate_reg_pressure.
24829 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24831 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
24832 expensive than udiv. Remove floating point cases from mod.
24834 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24836 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
24837 Increase idiv cost.
24839 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24841 * config/aarch64/aarch64.md
24842 (copysignsf3): Fix mask generation.
24844 2017-06-07 Jakub Jelinek <jakub@redhat.com>
24846 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
24848 (class dump_manager): Add register_dumps method.
24849 * dumpfile.c: Include langhooks.h.
24850 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
24851 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
24852 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
24853 (dump_manager::dump_register): Start with 512 entries instead of 32.
24854 (dump_manager::register_dumps): New method.
24855 * toplev.c (general_init): Instead of invoking register_dumps
24856 langhook, invoke register_dumps method on the dump manager.
24857 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
24860 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
24862 * doc/md.texi: Clarify the restrictions on a define_insn condition.
24863 Say that # requires an associated define_split to exist, and that
24864 the define_split must be suitable for use after register allocation.
24866 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24868 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
24869 (compute_outgoing_frequencies): Also initialize zero counts.
24870 (find_many_sub_basic_blocks): Do not produce uninitialized profile
24871 around loops; preserve more of profile when nothing changes.
24873 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
24875 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
24877 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
24878 * config/arm/arm-cpu-cdata.h: Regenerate.
24879 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
24880 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
24881 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
24882 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
24883 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
24885 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
24887 * config/arm/t-rmprofile: Likewise.
24888 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
24890 2017-06-06 David S. Miller <davem@davemloft.net>
24893 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
24894 blockage if function uses alloca.
24896 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24898 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
24899 New "uid" fields to hold pretty-print IDs of group and ref.
24900 Memory references are now identified as <group_id>:<ref_id>
24901 instead of using [random] addresses.
24902 (dump_mem_details): Simplify, no functional change.
24903 (dump_mem_ref): Simplify and make output more concise.
24904 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
24905 (find_or_create_group): Initialize group uid.
24906 (record_ref): Initialize ref uid. Improve debug output.
24907 (prune_group_by_reuse, should_issue_prefetch_p,)
24908 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
24909 (mark_nontemporal_store, determine_loop_nest_reuse):
24910 Improve debug output.
24912 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24914 * dbgcnt.def (prefetch): New debug counter.
24915 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
24916 (schedule_prefetches): Stop issueing prefetches if debug counter
24919 2017-06-06 Tom de Vries <tom@codesourcery.com>
24921 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
24922 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
24924 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24926 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
24927 Use aarch64_reg_or_zero predicate for operand 4.
24928 (aarch64_compare_and_swap<mode> define_insn_and_split):
24929 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
24930 (aarch64_store_exclusive<mode>): Likewise for operand 2.
24932 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24934 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
24935 (arm_compute_save_core_reg_mask): This.
24936 (thumb1_compute_save_reg_mask): Rename into ...
24937 (thumb1_compute_save_core_reg_mask): This.
24938 (arm_compute_save_reg0_reg12_mask): Adapt comment.
24939 (arm_compute_frame_layout): Likewise.
24941 2017-06-06 Richard Biener <rguenther@suse.de>
24943 PR tree-optimization/80974
24944 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
24945 keep or clear leaders SSA info.
24947 2017-06-06 Tom de Vries <tom@codesourcery.com>
24949 * config/nvptx/nvptx.c (split_mode_p): New function.
24950 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
24952 2017-06-06 Tom de Vries <tom@codesourcery.com>
24954 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
24956 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24959 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
24962 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24964 * shrink-wrap.c (handle_simple_exit): Update profile.
24965 (try_shrink_wrapping): Upate profile.
24967 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24969 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
24970 (tree_guess_outgoing_edge_probabilities): New.
24971 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
24972 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
24974 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24976 * ipa-split.c (split_function): Initialize return bb profile.
24978 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24980 * profile.c (compute_branch_probabilities): Also initialize
24981 EXIT_BLOCK profile.
24983 2017-06-06 Richard Biener <rguenther@suse.de>
24985 PR tree-optimization/80928
24986 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
24987 (vect_analyze_loop_operations): Properly guard analysis for
24989 (vect_transform_loop): Likewise.
24990 (vect_analyze_loop_2): Also reset SLP type on PHIs.
24991 (vect_model_induction_cost): Do not cost for pure SLP.
24992 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
24993 of induction in inner loop vectorization.
24994 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
24995 (vect_get_and_check_slp_defs): Handle vect_induction_def.
24996 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
24998 (vect_analyze_slp_cost_1): Cost induction.
24999 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25000 (vect_get_slp_vect_defs): Likewise.
25001 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25002 (vect_transform_stmt): Handle SLP reductions.
25003 * tree-vectorizer.h (vectorizable_induction): Adjust.
25005 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25007 * config/rs6000/rs6000.c (make_resolver_func): Update
25008 init_lowered_empty_function call.
25010 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25012 * doc/invoke.texi: Document the -fprofile-abs-path option.
25013 * common.opt (fprofile-abs-path): New option.
25014 * gcov-io.h (gcov_write_filename): Declare.
25015 * gcov-io.c (gcov_write_filename): New function.
25016 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25017 * profile.c (output_location): Likewise.
25019 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25021 * shring-wrap.c: Revert accidental commit.
25023 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25025 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25027 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25029 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25031 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25032 profile in callgraph edge.
25033 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25034 (apply_scale): Likewise.
25035 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25036 Also scale profile when inlining function with zero profile.
25037 (initialize_cfun): Update exit block profile even when it is zero.
25038 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25039 when profile is read.
25041 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25043 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25044 (CLONE_*): New constants to define the processors we can generate
25045 code for with the target_clone attribute.
25046 (rs6000_clone_map): New array to identify which clone processors
25047 the current program is running on.
25048 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25049 target_clone attribute.
25050 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25051 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25052 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25053 (cpu_expand_builtin): Add support for target_clone attribute.
25054 (rs6000_valid_attribute_p): Allow "default" attribute.
25055 (get_decl_name): New debug function to simplify printing the
25056 current function name in debugging statements.
25057 (rs6000_clone_priority): New functions to support the target_clone
25058 attribute, and be able to generate code to switch between ISA 2.05
25059 through ISA 3.0 (power6 through power9).
25060 (rs6000_compare_version_priority): Likewise.
25061 (rs6000_get_function_versions_dispatcher): Likewise.
25062 (make_resolver_func): Likewise.
25063 (add_condition_to_bb): Likewise.
25064 (dispatch_function_versions): Likewise.
25065 (rs6000_generate_version_dispatcher_body): Likewise.
25066 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25067 (fusion_gpr_load_p): Fix a spacing issue.
25068 * doc/extend.texi (Common Function Attributes): Document that the
25069 PowerPC supports the target_clone attribute.
25071 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25073 * config/arm/arm.h: explain F symbol found in description of ARM
25074 register allocation in its legend.
25076 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25078 * config/mips/frame-header-opt.c: Include profile-count.h.
25079 * config/riscv/riscv.c: Include profile-count.h
25081 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25083 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25085 (sm_set_flag_if_changed): Add bbs field.
25086 (execute_sm_if_changed_flag_set): Pass BBS.
25087 (execute_sm): Update.
25089 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25091 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25094 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25096 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25097 (peephole2): New peephole2 to emit the above.
25098 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25100 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25102 * config/aarch64/aarch64.c (define_peephole2 above
25103 *sub_<shift>_<mode>): New peephole.
25105 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25107 * config/i386/i386.c (make_resolver_func): Update.
25108 * Makefile.in: Add profile-count.h and profile-count.o
25109 * auto-profile.c (afdo_indirect_call): Update to new API.
25110 (afdo_set_bb_count): Update.
25111 (afdo_propagate_edge): Update.
25112 (afdo_propagate_circuit): Update.
25113 (afdo_calculate_branch_prob): Update.
25114 (afdo_annotate_cfg): Update.
25115 * basic-block.h: Include profile-count.h
25116 (struct edge_def): Turn count to profile_count.
25117 (struct basic_block_def): Likewie.
25118 (REG_BR_PROB_BASE): Move to profile-count.h
25119 (RDIV): Move to profile-count.h
25120 * bb-reorder.c (max_entry_count): Turn to profile_count.
25121 (find_traces): Update.
25122 (rotate_loop):Update.
25123 (connect_traces):Update.
25124 (sanitize_hot_paths):Update.
25125 * bt-load.c (migrate_btr_defs): Update.
25126 * cfg.c (RDIV): Remove.
25127 (init_flow): Use alloc_block.
25128 (alloc_block): Uninitialize count.
25129 (unchecked_make_edge): Uninitialize count.
25130 (check_bb_profile): Update.
25131 (dump_edge_info): Update.
25132 (dump_bb_info): Update.
25133 (update_bb_profile_for_threading): Update.
25134 (scale_bbs_frequencies_int): Update.
25135 (scale_bbs_frequencies_gcov_type): Update.
25136 (scale_bbs_frequencies_profile_count): New.
25137 * cfg.h (update_bb_profile_for_threading): Update.
25138 (scale_bbs_frequencies_profile_count): Declare.
25139 * cfgbuild.c (compute_outgoing_frequencies): Update.
25140 (find_many_sub_basic_blocks): Update.
25141 * cfgcleanup.c (try_forward_edges): Update.
25142 (try_crossjump_to_edge): Update.
25143 * cfgexpand.c (expand_gimple_tailcall): Update.
25144 (construct_exit_block): Update.
25145 * cfghooks.c (verify_flow_info): Update.
25146 (dump_bb_for_graph): Update.
25147 (split_edge): Update.
25148 (make_forwarder_block): Update.
25149 (duplicate_block): Update.
25150 (account_profile_record): Update.
25151 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25152 (get_estimated_loop_iterations): Update.
25153 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25154 (single_likely_exit): Update.
25155 * cfgloopmanip.c (scale_loop_profile): Update.
25157 (set_zero_probability): Update.
25158 (lv_adjust_loop_entry_edge): Update.
25159 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25160 (purge_dead_edges): Update.
25161 (rtl_account_profile_record): Update.
25162 * cgraph.c (cgraph_node::create): Uninitialize count.
25163 (symbol_table::create_edge): Uninitialize count.
25164 (cgraph_update_edges_for_call_stmt_node): Update.
25165 (cgraph_edge::dump_edge_flags): Update.
25166 (cgraph_node::dump): Update.
25167 (cgraph_edge::maybe_hot_p): Update.
25168 * cgraph.h: Include profile-count.h
25169 (create_clone), create_edge, create_indirect_edge): Update.
25170 (cgraph_node): Turn count to profile_count.
25171 (cgraph_edge0: Likewise.
25172 (make_speculative, clone): Update.
25173 (create_edge): Update.
25174 (init_lowered_empty_function): Update.
25175 * cgraphclones.c (cgraph_edge::clone): Update.
25176 (duplicate_thunk_for_node): Update.
25177 (cgraph_node::create_clone): Update.
25178 * cgraphunit.c (cgraph_node::analyze): Update.
25179 (cgraph_node::expand_thunk): Update.
25180 * final.c (dump_basic_block_info): Update.
25181 * gimple-streamer-in.c (input_bb): Update.
25182 * gimple-streamer-out.c (output_bb): Update.
25183 * graphite.c (print_global_statistics): Update.
25184 (print_graphite_scop_statistics): Update.
25185 * hsa-brig.c: Include basic-block.h.
25186 * hsa-dump.c: Include basic-block.h.
25187 * hsa-gen.c (T sum_slice): Update.
25188 (convert_switch_statements):Update.
25189 * hsa-regalloc.c: Include basic-block.h.
25190 * ipa-chkp.c (chkp_produce_thunks): Update.
25191 * ipa-cp.c (struct caller_statistics): Update.
25192 (init_caller_stats): Update.
25193 (gather_caller_stats): Update.
25194 (ipcp_cloning_candidate_p): Update.
25195 (good_cloning_opportunity_p): Update.
25196 (get_info_about_necessary_edges): Update.
25197 (dump_profile_updates): Update.
25198 (update_profiling_info): Update.
25199 (update_specialized_profile): Update.
25200 (perhaps_add_new_callers): Update.
25201 (decide_about_value): Update.
25202 (ipa_cp_c_finalize): Update.
25203 * ipa-devirt.c (struct odr_type_warn_count): Update.
25204 (struct decl_warn_count): Update.
25205 (struct final_warning_record): Update.
25206 (possible_polymorphic_call_targets): Update.
25207 (ipa_devirt): Update.
25208 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25209 * ipa-icf.c (sem_function::merge): Update.
25210 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
25211 * ipa-inline.c (compute_uninlined_call_time): Update.
25212 (compute_inlined_call_time): Update.
25213 (want_inline_small_function_p): Update.
25214 (want_inline_self_recursive_call_p): Update.
25215 (edge_badness): Update.
25216 (lookup_recursive_calls): Update.
25217 (recursive_inlining): Update.
25218 (inline_small_functions): Update.
25219 (dump_overall_stats): Update.
25220 (dump_inline_stats): Update.
25221 * ipa-profile.c (ipa_profile_generate_summary): Update.
25222 (ipa_propagate_frequency): Update.
25223 (ipa_profile): Update.
25224 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25225 * ipa-utils.c (ipa_merge_profiles): Update.
25226 * loop-doloop.c (doloop_modify): Update.
25227 * loop-unroll.c (report_unroll): Update.
25228 (unroll_loop_runtime_iterations): Update.
25229 * lto-cgraph.c (lto_output_edge): Update.
25230 (lto_output_node): Update.
25231 (input_node): Update.
25232 (input_edge): Update.
25233 (merge_profile_summaries): Update.
25234 * lto-streamer-in.c (input_cfg): Update.
25235 * lto-streamer-out.c (output_cfg): Update.
25236 * mcf.c (create_fixup_graph): Update.
25237 (adjust_cfg_counts): Update.
25238 (sum_edge_counts): Update.
25239 * modulo-sched.c (sms_schedule): Update.
25240 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
25241 * predict.c (maybe_hot_count_p): Update.
25242 (probably_never_executed): Update.
25243 (dump_prediction): Update.
25244 (combine_predictions_for_bb): Update.
25245 (propagate_freq): Update.
25246 (handle_missing_profiles): Update.
25247 (counts_to_freqs): Update.
25248 (rebuild_frequencies): Update.
25249 (force_edge_cold): Update.
25250 * predict.h: Include profile-count.h
25251 (maybe_hot_count_p, counts_to_freqs): UPdate.
25252 * print-rtl-function.c: Do not include cfg.h
25253 * print-rtl.c: Include basic-block.h
25254 * profile-count.c: New file.
25255 * profile-count.h: New file.
25256 * profile.c (is_edge_inconsistent): Update.
25257 (correct_negative_edge_counts): Update.
25258 (is_inconsistent): Update.
25259 (set_bb_counts): Update.
25260 (read_profile_edge_counts): Update.
25261 (compute_frequency_overlap): Update.
25262 (compute_branch_probabilities): Update; Initialize and deinitialize
25264 (branch_prob): Update.
25265 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
25266 (edge_gcov_count): New.
25267 (bb_gcov_count): New.
25268 * shrink-wrap.c (try_shrink_wrapping): Update.
25269 * tracer.c (better_p): Update.
25270 * trans-mem.c (expand_transaction): Update.
25271 (ipa_tm_insert_irr_call): Update.
25272 (ipa_tm_insert_gettmclone_call): Update.
25273 * tree-call-cdce.c: Update.
25274 * tree-cfg.c (gimple_duplicate_sese_region): Update.
25275 (gimple_duplicate_sese_tail): Update.
25276 (gimple_account_profile_record): Update.
25277 (execute_fixup_cfg): Update.
25278 * tree-inline.c (copy_bb): Update.
25279 (copy_edges_for_bb): Update.
25280 (initialize_cfun): Update.
25281 (freqs_to_counts): Update.
25282 (copy_cfg_body): Update.
25283 (expand_call_inline): Update.
25284 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25285 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
25286 (try_unroll_loop_completely): Update.
25287 (try_peel_loop): Update.
25288 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25289 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
25290 * tree-ssa-loop-split.c (connect_loops): Update.
25291 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
25292 * tree-ssa-reassoc.c (branch_fixup): Update.
25293 * tree-ssa-tail-merge.c (replace_block_by): Update.
25294 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
25295 (compute_path_counts): Update.
25296 (update_profile): Update.
25297 (recompute_probabilities): Update.
25298 (update_joiner_offpath_counts): Update.
25299 (estimated_freqs_path): Update.
25300 (freqs_to_counts_path): Update.
25301 (clear_counts_path): Update.
25302 (ssa_fix_duplicate_block_edges): Update.
25303 (duplicate_thread_path): Update.
25304 * tree-switch-conversion.c (case_bit_test_cmp): Update.
25305 (struct switch_conv_info): Update.
25306 * tree-tailcall.c (decrease_profile): Update.
25307 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25308 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25309 * value-prof.c (check_counter): Update.
25310 (gimple_divmod_fixed_value): Update.
25311 (gimple_mod_pow2): Update.
25312 (gimple_mod_subtract): Update.
25313 (gimple_ic_transform): Update.
25314 (gimple_stringop_fixed_value): Update.
25315 * value-prof.h (gimple_ic): Update.
25317 2017-06-02 Carl Love <cel@us.ibm.com>
25319 * config/rs6000/rs6000-c: Add support for built-in functions
25320 vector double vec_doublee (vector signed int);
25321 vector double vec_doublee (vector unsigned int);
25322 vector double vec_doublee (vector float);
25323 vector double vec_doubleh (vector signed int);
25324 vector double vec_doubleh (vector unsigned int);
25325 vector double vec_doubleh (vector float);
25326 vector double vec_doublel (vector signed int);
25327 vector double vec_doublel (vector unsigned int);
25328 vector double vec_doublel (vector float);
25329 vector double vec_doubleo (vector signed int);
25330 vector double vec_doubleo (vector unsigned int);
25331 vector double vec_doubleo (vector float);.
25332 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
25333 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
25335 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
25336 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
25337 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
25339 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
25340 vec_doublel, vec_doubleh.
25341 * doc/extend.texi: Update the built-in documentation file for the
25342 new built-in functions.
25344 2017-06-02 David Malcolm <dmalcolm@redhat.com>
25347 * ipa-inline-analysis.c (free_growth_caches): Set
25348 edge_removal_hook_holder to NULL after removing it.
25350 2017-06-02 Sudakshina Das <sudi.das@arm.com>
25352 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
25353 comparision with zero.
25355 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25356 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25357 for early expansion of vec_min and vec_max builtins.
25358 (builtin_function_type): Add min/max unsigned variants to those
25359 identified as having unsigned arguments.
25361 2017-06-02 Olivier Hainque <hainque@adacore.com>
25363 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
25365 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25367 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
25368 Use VALL_F16 iterator rather than VALL.
25370 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25372 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
25373 Emit CBNZ inside loop when doing a strong exchange and comparing
25374 against zero. Generate the CC flags after the loop.
25376 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
25378 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
25379 (dl_section_ref): New.
25380 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
25381 On AIX, append an expression to subtract the size of the
25382 section length to dl_section_ref.
25384 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25386 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25387 for early expansion of vector absolute builtins.
25389 2017-06-02 Richard Biener <rguenther@suse.de>
25391 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
25392 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
25394 2017-06-02 Richard Biener <rguenther@suse.de>
25396 PR tree-optimization/80948
25397 * tree-tailcall.c (find_tail_calls): Track stmts to move in
25398 stmt order as well.
25400 2017-06-02 Richard Biener <rguenther@suse.de>
25402 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
25404 * tree-vect-stmts.c (process_use): Do not mark backedge defs
25405 for inductions as relevant.
25407 2017-06-02 Richard Biener <rguenther@suse.de>
25409 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
25410 (vectorizable_induction): ... this. Remove dead code.
25412 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
25414 * builtins. (expand_builtin_alloca): Remove second parameter and
25415 infer its value from the first parameter instead.
25416 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
25418 2017-06-02 Jakub Jelinek <jakub@redhat.com>
25420 PR rtl-optimization/80903
25421 * loop-doloop.c (add_test): Unshare sequence.
25423 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
25425 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
25427 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
25429 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
25431 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
25432 xlogue_layout::get_instance, logue_layout::xlogue_layout,
25433 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
25434 (xlogue_layout::get_stub_rtx): Make static.
25435 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
25436 (xlogue_layout::compute_stub_managed_regs): Rename to...
25437 (xlogue_layout::count_stub_managed_regs): ...this.
25438 (xlogue_layout::is_stub_managed_reg): New function.
25439 (xlogue_layout::m_stub_names): Rename to...
25440 (xlogue_layout::s_stub_names): ...this, make static.
25441 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
25442 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
25443 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
25444 xlogue_layout::s_stub_names): Instantiate statics.
25445 (stub_managed_regs): Remove.
25446 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
25447 (disable_call_ms2sysv_xlogues): Rename to...
25448 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
25449 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
25451 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
25452 change after reload_completed.
25453 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
25455 (ix86_expand_prologue): Likewise.
25456 (ix86_expand_epilogue): Likewise.
25457 (ix86_expand_split_stack_prologue): Likewise.
25458 (ix86_compute_frame_layout): Remove frame parameter ...
25459 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
25460 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
25462 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
25463 (ix86_frame): Move from here ...
25464 * config/i386/i386.h (ix86_frame): ... to here.
25465 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
25466 complete ix86_frame data structure instead. Remove some_ld_name.
25468 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25470 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
25471 symbols that hold a DECL_VALUE_EXPR.
25473 2017-06-01 Martin Jambor <mjambor@suse.cz>
25475 PR tree-optimization/80898
25476 * tree-sra.c (process_subtree_disqualification): Removed.
25477 (disqualify_candidate): Do not acll
25478 process_subtree_disqualification.
25479 (subtree_mark_written_and_enqueue): New function.
25480 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
25481 RHS has been disqualified and re-queue LHS if necessary. Apart
25482 from that, ignore disqualified RHS.
25484 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25486 * config/s390/s390.c (s390_emit_epilogue): Disable early return
25487 address fetch for z10 or later.
25489 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25491 * config/arc/arc.md (tst_movb): Add guard when splitting.
25493 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25495 * config/arc/arc.c (arc_can_eliminate): Test against
25496 arc_frame_pointer_needed.
25498 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25500 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
25501 to prevent store reordering.
25502 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
25503 (type): Add block type.
25504 (stack_tie): Define special instruction to be used in
25507 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25509 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
25510 constraint. It is not valid for the pattern.
25511 (noncommutative_binary_comparison): Likewise.
25513 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25515 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
25518 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25520 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
25521 be used by the reg-alloc.
25523 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25525 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
25526 reg-alloc when having mul64 or mul32x16 instructions.
25527 (mulsidi3): Likewise.
25528 (umulsidi3): Likewise.
25529 (mulsi32x16): New pattern.
25530 (mulsi64): Likewise.
25531 (mulsidi64): Likewise.
25532 (umulsidi64): Likewise.
25533 (MUL32x16_REG): Define.
25534 (mul64_600): Use MUL32x16_REG.
25535 (mac64_600): Likewise.
25536 (umul64_600): Likewise.
25537 (umac64_600): Likewise.
25539 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25541 * config/arc/arc.md (mulsi3_700): Make it commutative.
25543 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
25545 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
25547 (*sign_extendsidi2_insn): Likewise for movstosw.
25549 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25551 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
25552 the type of the input discriminant value. Convert the
25553 discriminant value of signedness vary.
25555 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
25557 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
25558 Add to -Wall section.
25560 2017-06-01 Richard Biener <rguenther@suse.de>
25562 PR middle-end/66313
25563 * fold-const.c (fold_plusminus_mult_expr): If the factored
25564 factor may be zero use a wrapping type for the inner operation.
25565 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
25566 and handle moved defs.
25567 (process_assignment): Properly guard the unary op case. Return a
25568 tri-state indicating that moving the stmt before the call may allow
25569 to continue. Pass through to_move.
25570 (find_tail_calls): Handle moving unrelated defs before
25573 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
25576 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
25577 splitter result in the canonical way.
25579 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
25581 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
25582 also for 32bit target. Update insn attributes.
25583 (zero-extendsidi2 splitter): Allow all registers for operand 1.
25585 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
25587 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
25588 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
25589 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
25590 (_mm_maskz_min_ss): New intrinsics.
25592 2017-05-31 Martin Liska <mliska@suse.cz>
25594 * tree-vect-loop.c (vect_create_epilog_for_reduction):
25595 Change comment style to one we normally use.
25596 (vectorizable_reduction): Likewise.
25597 (vectorizable_induction): Likewise.
25598 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
25599 (vectorizable_call): Likewise.
25600 (vectorizable_simd_clone_call): Likewise.
25601 (vectorizable_conversion): Likewise.
25602 (vectorizable_assignment): Likewise.
25603 (vectorizable_shift): Likewise.
25604 (vectorizable_operation): Likewise.
25605 (vectorizable_store): Likewise.
25606 (vectorizable_load): Likewise.
25607 * tree-vectorizer.h: Likewise.
25609 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
25611 * passes.c (emergency_dump_function): New.
25612 * tree-pass.h (emergency_dump_function): Declare.
25613 * plugin.c (plugins_internal_error_function): Remove.
25614 * plugin.h (plugins_internal_error_function): Remove declaration.
25615 * toplev.c (internal_error_function): New static function. Use it...
25616 (general_init): ...here.
25618 2017-05-31 Graham Markall <graham.markall@embecosm.com>
25620 * config/arc/arc.c (arc_print_operand): Handle constant operands.
25621 (arc_rtx_costs): Add costs for new patterns.
25622 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
25623 * config/arc/predicates.md: Add _1_2_3_operand predicate.
25625 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25627 * tree-ssa-strlen.c (get_next_strinfo): New function.
25628 (get_stridx_plus_constant): Use it.
25629 (zero_length_string): Likewise.
25630 (adjust_related_strinfos): Likewise.
25631 (adjust_last_stmt): Likewise.
25633 2017-05-31 Richard Biener <rguenther@suse.de>
25636 * config/i386/i386.c (ix86_expand_builtin): Remove assert
25637 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
25639 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25641 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
25642 loop_vinfo argument and use of dependence distance vectors.
25643 Check instead whether the two references differ only in their
25644 initial value and assume that they have the same alignment if the
25645 difference is a multiple of the vector alignment.
25646 (vect_analyze_data_refs_alignment): Update call accordingly.
25648 2017-05-31 Martin Liska <mliska@suse.cz>
25651 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
25653 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25655 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
25656 (create_intersect_range_checks): Move from ...
25657 * tree-data-ref.c (create_intersect_range_checks_index)
25658 (create_intersect_range_checks): ... to here.
25659 (create_runtime_alias_checks): New function factored from ...
25660 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
25661 here. Call above function.
25662 * tree-data-ref.h (create_runtime_alias_checks): New function.
25664 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25666 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
25667 segment length for dr_b and compute it in wide_int.
25669 2017-05-31 Richard Biener <rguenther@suse.de>
25671 PR tree-optimization/80906
25672 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
25673 and pass through iv_map.
25674 (copy_bb_and_scalar_dependences): Adjust.
25675 (translate_pending_phi_nodes): Likewise.
25676 (copy_loop_close_phi_args): Handle code-generating IVs instead
25679 2017-05-30 David Malcolm <dmalcolm@redhat.com>
25681 * diagnostic-color.c (color_dict): Add "type-diff".
25682 (parse_gcc_colors): Update comment.
25683 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
25684 -fdiagnostics-show-template-tree and -fno-elide-type.
25685 (GCC_COLORS): Add type-diff to example.
25687 (-fdiagnostics-show-template-tree): New.
25688 (-fno-elide-type): New.
25689 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
25690 the pp_format_decoder callback. Call any m_format_postprocessor's
25692 (pretty_printer::pretty_printer): Initialize
25693 m_format_postprocessor.
25694 (pretty_printer::~pretty_printer): Delete any
25695 m_format_postprocessor.
25696 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
25697 (class format_postprocessor): New class.
25698 (struct pretty_printer::format_decoder): Document the new parameters.
25699 (struct pretty_printer::m_format_postprocessor): New field.
25700 * tree-diagnostic.c (default_tree_printer): Update for new
25701 bool and const char ** params.
25702 * tree-diagnostic.h (default_tree_printer): Likewise.
25704 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
25706 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
25707 (lwa_operand): Delete rs6000_gen_cell_microcode test.
25708 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25709 rs6000_gen_cell_microcode code.
25710 (rs6000_final_prescan_insn): Delete.
25711 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
25712 "warn-cell-microcode" entries.
25713 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
25714 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
25715 throughout. Change cc_reg_not_micro_cr0_operand to
25716 cc_reg_not_cr0_operand throughout.
25717 (*extendhi<mode>2_noload): Delete.
25718 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
25719 (mwarn-cell-microcode): Delete.
25720 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
25721 -mgen-cell-microcode and -mwarn-cell-microcode.
25723 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
25726 * config/i386/constraints.md (Yd): New constraint.
25728 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
25729 and (?Yd, r) alternatives. Update insn attributes.
25730 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
25731 and (?*Yd, r) alternatives. Update insn attributes.
25732 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
25734 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
25736 * gimplify.c (gimplify_modify_expr): Don't create a
25737 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
25740 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
25742 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
25744 2017-05-30 Richard Biener <rguenther@suse.de>
25746 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
25747 and reduc_def fields.
25748 (STMT_VINFO_REDUC_TYPE): New define.
25749 (STMT_VINFO_REDUC_DEF): Likewise.
25750 (vect_force_simple_reduction): Adjust prototype.
25751 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
25752 (vect_is_simple_reduction): Remove check_reduction argument.
25753 (vect_force_simple_reduction): Adjust and set
25754 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25755 (vectorizable_reduction): Do not re-do reduction analysis
25756 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25757 * tree-parloops.c (gather_scalar_reductions): Adjust.
25759 2017-05-30 Richard Biener <rguenther@suse.de>
25761 PR middle-end/80901
25762 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
25765 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
25767 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
25768 Introduce unknown_misalignment parameter and remove vf.
25769 (vect_peeling_hash_get_lowest_cost):
25770 Pass unknown_misalignment parameter.
25771 (vect_enhance_data_refs_alignment):
25772 Fix unsupportable data ref treatment.
25774 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25776 * tree-vect-data-refs.c (vect_get_data_access_cost):
25777 Workaround for SLP handling.
25778 (vect_enhance_data_refs_alignment):
25779 Compute costs for doing no peeling at all, compare to the best
25780 peeling costs so far and avoid peeling if cheaper.
25782 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25784 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
25785 Return peeling info and set costs to zero for unlimited cost
25787 (vect_enhance_data_refs_alignment): Also inspect all datarefs
25788 with unknown misalignment. Compute and costs for unknown
25789 misalignment, compare them to the costs for known misalignment
25790 and choose the cheapest for peeling.
25792 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25794 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
25795 (vect_get_peeling_costs_all_drs): Create function.
25796 (vect_peeling_hash_get_lowest_cost):
25797 Use vect_get_peeling_costs_all_drs.
25798 (vect_peeling_supportable): Create function.
25799 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
25801 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25803 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
25804 DR_HAS_NEGATIVE_STEP.
25805 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
25806 (vect_enhance_data_refs_alignment): Use.
25807 (vect_duplicate_ssa_name_ptr_info): Use.
25808 * tree-vectorizer.h (dr_misalignment): Use.
25809 (known_alignment_for_access_p): Use.
25811 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25814 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
25816 (has_section_name): New function.
25818 2017-05-30 Martin Liska <mliska@suse.cz>
25821 * auto-profile.c (get_function_decl_from_block): Fix
25824 2017-05-30 Richard Biener <rguenther@suse.de>
25826 PR middle-end/80876
25827 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
25829 2017-05-30 Martin Liska <mliska@suse.cz>
25831 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
25832 * dumpfile.h (struct dump_file_info): Remove ctors.
25834 2017-05-30 Martin Liska <mliska@suse.cz>
25836 * predict.def: Fix GNU coding style.
25838 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25840 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25841 Mark 'to' argument with ATTRIBUTE_UNUSED.
25843 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25845 * config/xtensa/xtensa.c (xtensa_emit_call): Use
25846 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
25847 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
25850 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
25852 * doc/install.texi (Options specification): Restore entry of
25853 --enable-sjlj-exceptions.
25855 2017-05-27 Michael Eager <eager@eagercon.com>
25858 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
25860 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
25862 * config/microblaze/microblaze.h
25863 (FIXED_REGISTERS): Update in macro.
25864 (CALL_USED_REGISTERS): Update in macro.
25866 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
25868 * doc/install.texi: Add links to macOS binary distributions.
25870 2017-05-27 Jakub Jelinek <jakub@redhat.com>
25874 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25876 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25878 2017-05-26 Martin Liska <mliska@suse.cz>
25880 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
25882 2017-05-26 Martin Liska <mliska@suse.cz>
25884 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
25885 always leading ';; '.
25886 (dump_bb_info): Likewise.
25887 (brief_dump_cfg): Likewise.
25888 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
25889 * dumpfile.c: Remove usage of TDF_VERBOSE.
25890 * dumpfile.h (enum dump_kind): Likewise.
25891 (dump_gimple_bb_header): Do not use TDF_COMMENT.
25892 * print-tree.c (debug_verbose): Remove.
25893 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
25894 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
25895 * tree-diagnostic.c (default_tree_printer): Replace
25896 TDF_DIAGNOSTIC with TDF_SLIM.
25898 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25900 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
25901 in parameter loop, rather than loop_vinfo.
25902 (create_intersect_range_checks): Ditto.
25903 (vect_create_cond_for_alias_checks): Update call to above functions.
25905 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25907 PR tree-optimization/80815
25908 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
25909 for merging runtime alias checks. Handle negative DR_STEPs.
25911 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25913 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
25915 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
25916 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
25917 out code pruning runtime alias checks.
25918 * tree-data-ref.c (prune_runtime_alias_test_list): New function
25919 factored out from above.
25920 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25922 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25924 (prune_runtime_alias_test_list): New decalaration.
25926 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25928 * tree-vect-data-refs.c (compare_tree): Rename and move ...
25929 * tree-data-ref.c (data_ref_compare_tree): ... to here.
25930 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
25931 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
25932 (operator==, comp_dr_with_seg_len_pair): Ditto.
25933 (vect_prune_runtime_alias_test_list): Ditto.
25935 2017-05-26 Martin Liska <mliska@suse.cz>
25938 * params.def: Bound partial-inlining-entry-probability param.
25940 2017-05-26 Marek Polacek <polacek@redhat.com>
25943 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
25946 2017-05-26 Richard Biener <rguenther@suse.de>
25948 PR tree-optimization/80842
25949 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
25952 2017-05-26 Richard Biener <rguenther@suse.de>
25954 PR tree-optimization/80844
25955 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
25957 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
25959 * doc/md.texi (Machine Constraints): Update x86 family
25960 machine constraints section to match 'config/i386/constraints.md'.
25962 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
25964 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
25966 2017-05-25 Nathan Sidwell <nathan@acm.org>
25968 * doc/invoke.texi (--enable-languages): Update documentation.
25970 2017-05-25 Martin Liska <mliska@suse.cz>
25972 * dumpfile.c: Add TDF_FOLDING.
25973 * dumpfile.h (enum dump_kind): Likewise.
25974 * genmatch.c (dt_simplify::gen_1): Use it.
25976 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25978 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
25980 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25982 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25983 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
25985 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25987 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
25988 * match.pd (X == C): Rewrite it here.
25989 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
25990 with_certain_nonzero_bits2): New predicates.
25991 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
25993 2017-05-24 Nathan Sidwell <nathan@acm.org>
25995 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
25998 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26001 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26003 * config/powerpcspe: New port. Files are copied from the rs6000
26004 port, with "rs6000" in filenames replaced by "powerpcspe".
26006 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26008 PR rtl-optimization/80754
26009 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26011 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26013 * config/sparc/sparc.md (length): Return the correct value for -mflat
26014 sibcalls to match output_sibcall.
26016 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26020 * config/rs6000/rs6000.c (struct machine_function): Add new field
26022 (rs6000_get_separate_components): Init that field, use it.
26023 (rs6000_components_for_bb): Use the field.
26025 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26027 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26029 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26031 PR middle-end/80823
26032 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26034 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26037 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26038 address against address_operand predicate.
26039 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26041 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26043 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26044 parameters passed indirectly.
26046 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26048 * config/i386/i386.md (*movdi_internal): Remove SSE4
26049 alternative 18 (?r, *v). Update insn attributes.
26050 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26051 Update insn attributes.
26052 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26053 Update insn attributes.
26054 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26055 alternative 1 (r, v). Remove isa attribute.
26056 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26057 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26058 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26060 2017-05-23 Tom de Vries <tom@codesourcery.com>
26062 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26065 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26067 * cgraphunit.c (symbol_table::process_new_functions): Update.
26068 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26069 (inline_generate_summary): Rename to ...
26070 (ipa_fn_summary_generate): ... this one.
26071 (inline_read_summary): Rename to ...
26072 (ipa_fn_summary_read): ... this one.
26073 (inline_write_summary): Rename to ...
26074 (ipa_fn_summary_write): ... this one.
26075 (inline_free_summary): Rename to ...
26076 (ipa_free_fn_summary): ... this one.
26077 (pass_data_local_fn_summary, pass_local_fn_summary,
26078 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26079 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26080 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26081 make_pass_ipa_fn_summary): New.
26082 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26083 inline_write_summary, inline_free_summary): Remove.
26084 (ipa_free_fn_summary) : New.
26085 * ipa-inline.c (ipa_inline): Update.
26086 (pass_ipa_inline): Do not generate summaries.
26087 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26089 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26090 and add pass_ipa_fn_summary.
26091 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26093 (make_pass_inline_parameters): Remove.
26095 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26097 * omp-low.c (struct omp_context): Remove "default_kind" member.
26100 * omp-offload.c (execute_oacc_device_lower): Remove the
26101 parallelism dimensions function attributes for unparallelized
26102 OpenACC kernels constructs.
26104 2017-05-23 Martin Liska <mliska@suse.cz>
26106 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26108 (cgraph_edge::make_speculative): Likewise.
26109 (cgraph_edge::resolve_speculation): Likewise.
26110 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26111 (cgraph_node::dump): Likewise.
26112 * cgraph.h: Likewise.
26113 * cgraphunit.c (analyze_functions): Likewise.
26114 (symbol_table::compile): Likewise.
26115 * ipa-cp.c (print_all_lattices): Likewise.
26116 (determine_versionability): Likewise.
26117 (initialize_node_lattices): Likewise.
26118 (ipcp_verify_propagated_values): Likewise.
26119 (estimate_local_effects): Likewise.
26120 (update_profiling_info): Likewise.
26121 (create_specialized_node): Likewise.
26122 (perhaps_add_new_callers): Likewise.
26123 (decide_about_value): Likewise.
26124 (decide_whether_version_node): Likewise.
26125 (identify_dead_nodes): Likewise.
26126 (ipcp_store_bits_results): Likewise.
26127 * ipa-devirt.c (dump_targets): Likewise.
26128 (ipa_devirt): Likewise.
26129 * ipa-icf.c (sem_item::dump): Likewise.
26130 (sem_function::equals): Likewise.
26131 (sem_variable::equals): Likewise.
26132 (sem_item_optimizer::read_section): Likewise.
26133 (sem_item_optimizer::execute): Likewise.
26134 (congruence_class::dump): Likewise.
26135 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26136 (dump_inline_summary): Likewise.
26137 (estimate_node_size_and_time): Likewise.
26138 (inline_analyze_function): Likewise.
26139 * ipa-inline-transform.c (inline_call): Likewise.
26140 * ipa-inline.c (report_inline_failed_reason): Likewise.
26141 (want_early_inline_function_p): Likewise.
26142 (edge_badness): Likewise.
26143 (update_edge_key): Likewise.
26144 (inline_small_functions): Likewise.
26145 * ipa-profile.c (ipa_profile): Likewise.
26146 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26147 (ipa_make_edge_direct_to_target): Likewise.
26148 (remove_described_reference): Likewise.
26149 (ipa_impossible_devirt_target): Likewise.
26150 (propagate_controlled_uses): Likewise.
26151 (ipa_print_node_params): Likewise.
26152 (ipcp_transform_function): Likewise.
26153 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26154 (propagate_pure_const): Likewise.
26155 * ipa-reference.c (generate_summary): Likewise.
26156 (read_write_all_from_decl): Likewise.
26157 (propagate): Likewise.
26158 (ipa_reference_read_optimization_summary): Likewise.
26159 * ipa-utils.c (ipa_merge_profiles): Likewise.
26160 * ipa.c (walk_polymorphic_call_targets): Likewise.
26161 (symbol_table::remove_unreachable_nodes): Likewise.
26162 (ipa_single_use): Likewise.
26163 * passes.c (execute_todo): Likewise.
26164 * predict.c (drop_profile): Likewise.
26165 * symtab.c (symtab_node::get_dump_name): New function.
26166 (symtab_node::dump_name): Likewise.
26167 (symtab_node::dump_asm_name): Likewise.
26168 (symtab_node::dump_references): Likewise.
26169 (symtab_node::dump_referring): Likewise.
26170 (symtab_node::dump_base): Likewise.
26171 (symtab_node::debug_symtab): Likewise.
26172 * tree-sra.c (convert_callers_for_node): Likewise.
26173 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26174 * value-prof.c (init_node_map): Likewise.
26176 2017-05-23 Martin Liska <mliska@suse.cz>
26178 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26179 and symtab_node::debug_symtab to symbol_table::debug.
26180 * cgraphunit.c (analyze_functions): Use the renamed function.
26181 (symbol_table::compile): Likewise.
26182 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26183 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26184 * passes.c (execute_todo): Likewise.
26185 * symtab.c (symbol_table::dump): New function.
26186 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26188 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26190 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26191 that nonconst implies exec.
26193 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26195 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26196 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26197 (inline_edge_summary_vec): Turn into ...
26198 (ipa_call_summaries): ... this one.
26199 (redirect_to_unreachable, edge_set_predicate,
26200 evaluate_properties_for_edge, inline_summary_alloc,
26201 reset_ipa_call_summary, reset_inline_summary,
26202 inline_summary_t::duplicate): Update.
26203 (inline_edge_duplication_hook): Turn to ...
26204 (ipa_call_summary_t::duplicate): ... this one.
26205 (inline_edge_removal_hook): Turn to ...
26206 (ipa_call_summary_t::remove): ... this one.
26207 (dump_inline_edge_summary): Turn to ...
26208 (dump_ipa_call_summary): ... this one.
26209 (estimate_function_body_sizes): Update.
26210 (inline_update_callee_summaries): Update.
26211 (remap_edge_change_prob): Update.
26212 (remap_edge_summaries): Update.
26213 (inline_merge_summary): Update.
26214 (do_estimate_edge_time): Update.
26215 (inline_generate_summary): Update.
26216 (inline_read_section): Update.
26217 (inline_read_summary): Update.
26218 (inline_free_summary): Update.
26219 * ipa-inline.c (can_inline_edge_p): Update.
26220 (compute_inlined_call_time): Update.
26221 (want_inline_small_function_p): Update.
26222 (edge_badness): Update.
26223 (early_inliner): Update.
26224 * ipa-inline.h (inline_edge_summary): Turn to ...
26225 (ipa_call_summary): ... this one.
26226 (ipa_call_summary_t): New class.
26227 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26228 (ipa_call_summaries): New.
26229 (inline_edge_summary): Remove.
26230 (estimate_edge_growth): Update.
26231 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26232 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26233 * ipa-split.c (execute_split_functions): Update.
26234 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26236 2017-05-23 Tom de Vries <tom@codesourcery.com>
26238 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26239 attributes): Document rdrand effective target.
26241 2017-05-23 Tom de Vries <tom@codesourcery.com>
26243 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26244 attributes): Sort alphabetically.
26246 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
26248 * config/avr/genmultilib.awk: Use gsub instead of gensub.
26250 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26253 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
26254 V2DF/V2DI splat into two separate patterns, one that handles
26255 registers, and the other that only handles memory. Drop support
26256 for splatting from a GPR on ISA 2.07 and then splitting the
26257 splat into direct move and splat.
26258 (vsx_splat_<mode>_reg): Likewise.
26259 (vsx_splat_<mode>_mem): Likewise.
26261 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
26263 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
26265 2017-05-22 Jakub Jelinek <jakub@redhat.com>
26267 PR middle-end/80809
26268 * omp-low.c (finish_taskreg_remap): New function.
26269 (finish_taskreg_scan): If unit size of ctx->record_type
26270 is non-constant, unshare the size expression and replace
26271 decls in it with possible outer var refs.
26273 PR middle-end/80809
26274 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
26275 GOVD_SHARED rather than GOVD_PRIVATE with it.
26276 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
26277 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
26279 PR middle-end/80853
26280 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
26281 as last argument to build_outer_var_ref for pointer bases of array
26282 section reductions.
26284 2017-05-19 Martin Sebor <msebor@redhat.com>
26286 * print-tree.c (print_node): Print DECL_READ_P flag.
26288 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26290 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
26291 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
26292 * cgraph.c: Likewise.
26293 * cgraphunit.c: Likewise.
26294 * gengtype.c: Likewise.
26295 * ipa-cp.c: Likewise.
26296 * ipa-devirt.c: Likewise.
26297 * ipa-icf.c: Likewise.
26298 * ipa-predicate.c: Likewise.
26299 * ipa-profile.c: Likewise.
26300 * ipa-prop.c: Likewise.
26301 * ipa-split.c: Likewise.
26303 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
26304 edge_predicate_pool, dump_inline_hints,
26305 inline_summary::account_size_time, redirect_to_unreachable,
26306 edge_set_predicate, set_hint_predicate,
26307 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
26308 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
26309 inline_summary_t::remove, remap_hint_predicate_after_duplication,
26310 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
26311 ipa_call_summary_t::remove, initialize_growth_caches,
26312 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
26313 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
26314 mark_modified, unmodified_parm_1, unmodified_parm,
26315 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
26316 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
26317 compute_bb_predicates, will_be_nonconstant_expr_predicate,
26318 will_be_nonconstant_predicate, record_modified_bb_info,
26319 get_minimal_bb, record_modified, param_change_prob,
26320 phi_result_unknown_predicate, predicate_for_phi_result,
26321 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
26322 estimate_function_body_sizes, compute_inline_parameters,
26323 compute_inline_parameters_for_curren, pass_data_inline_parameters,
26324 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
26325 inline_update_callee_summaries, remap_edge_change_prob,
26326 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
26327 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
26328 inline_analyze_function, inline_summary_t::insert,
26329 inline_generate_summary, read_ipa_call_summary, inline_read_section,
26330 inline_read_summary, write_ipa_call_summary, inline_write_summary,
26331 inline_free_summary): Move to ipa-fnsummary.h
26332 (predicate_t): Remove.
26333 * ipa-fnsummary.c: New file.
26334 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
26335 (enum inline_hints_vals, inline_hints, agg_position_info,
26336 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
26337 inline_summaries, ipa_call_summary, ipa_call_summary_t,
26338 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
26339 dump_inline_summary, dump_inline_hints, inline_generate_summary,
26340 inline_read_summary, inline_write_summary, inline_free_summary,
26341 inline_analyze_function, initialize_inline_failed,
26342 inline_merge_summary, inline_update_overall_summary,
26343 compute_inline_parameters): Move to ipa-fnsummary.h
26344 * ipa-fnsummary.h: New file.
26345 * ipa-inline-transform.h: Include ipa-inline.h.
26346 * ipa-inline.c: LIkewise.
26348 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26350 * ipa-inline.c (edge_badness): Use inlined_time instead of
26351 inline_summaries->get.
26353 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26355 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
26357 2017-05-22 Nathan Sidwell <nathan@acm.org>
26359 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
26360 (fdump-lang): Document 'raw' option.
26361 * dumpfile.h (TDI_tu): Delete.
26362 * dumpfile.c (dump_files): Remove translation-unit.
26363 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
26365 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
26367 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
26368 command option from $(AWK) call.
26369 * config/avr/genmultilib.awk: Simplify and rewrite so that it
26370 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
26371 [FORMAT]: Remove handling of variable.
26372 * config/avr/t-multilib: Regenerate.
26374 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26376 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
26378 (dump_inline_summary): Do not print self_time.
26379 (estimate_function_body_sizes): Do not set self_time.
26380 (compute_inline_parameters): Likewise.
26381 (inline_read_section, inline_write_summary): Do not stream self_time.
26382 * ipa-inline.h (inline_summary): Drop self_time.
26384 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26386 * ipa-inline-analysis.c (account_size_time): Rename to ...
26387 (inline_summary::account_size_time): ... this one.
26388 (reset_ipa_call_summary): Turn to ...
26389 (ipa_call_summary::reset): ... this one.
26390 (reset_inline_summary): Turn to ...
26391 (inline_summary::reset): ... this one.
26392 (inline_summary_t::remove): Update.
26393 (inline_summary_t::duplicate): Update.
26394 (ipa_call_summary_t::remove): Update.
26395 (dump_inline_summary): Update.
26396 (estimate_function_body_sizes): Update.
26397 (compute_inline_parameters): Update.
26398 (estimate_node_size_and_time): Update.
26399 (inline_merge_summary): Update.
26400 (inline_update_overall_summary): Update.
26401 (inline_read_section): Update.
26402 (inline_write_summary): Update.
26403 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
26404 add account_size_time and reset member functions.
26405 (ipa_call_summary): Add reset function.
26406 * ipa-predicate.h (predicate::operator &): Constify.
26408 2017-05-22 Richard Biener <rguenther@suse.de>
26410 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
26412 2017-05-19 Jason Merrill <jason@redhat.com>
26414 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
26416 2017-05-19 Marek Polacek <polacek@redhat.com>
26419 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
26420 TYPE_OVERFLOW_WRAPS checks.
26422 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
26424 * tree-core.h (enum omp_clause_default_kind): Add
26425 "OMP_CLAUSE_DEFAULT_PRESENT".
26426 * tree-pretty-print.c (dump_omp_clause): Handle it.
26427 * gimplify.c (enum gimplify_omp_var_data): Add
26428 "GOVD_MAP_FORCE_PRESENT".
26429 (gimplify_adjust_omp_clauses_1): Map it to
26430 "GOMP_MAP_FORCE_PRESENT".
26431 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
26433 * gimplify.c (oacc_default_clause): Clarify.
26435 2017-05-19 Nathan Sidwell <nathan@acm.org>
26437 LANG_HOOK_REGISTER_DUMPS
26438 * toplev.c (general_init): Call register dump lang hook.
26439 * doc/invoke.texi: Document -fdump-lang option family.
26440 * dumpfile.c (dump_files): Remove class dump here.
26441 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
26442 * dumpfile.h (tree_dump_index): Remove TDI_class.
26443 * langhooks-def.h (lhd_register_dumps): Declare.
26444 (LANG_HOOKS_REGISTER_DUMPS): Define.
26445 (LANG_HOOKS_INITIALIZER): Add it.
26446 * langhooks.c (lhd_register_dumps): Define.
26447 * langhooks.h (struct lang_hooks): Add register_dumps.
26449 2017-05-19 Nathan Sidwell <nathan@acm.org>
26451 * context.h (context::set_passes): New.
26452 * context.c (context::context): Do not create pass manager.
26453 * toplev.c (general_init): Create pass manager here.
26455 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
26457 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
26458 use this splitter if two add or or instructions would also work for
26459 the constant we want to generate.
26461 2017-05-19 Richard Biener <rguenther@suse.de>
26464 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
26465 predicate evaluation.
26467 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26469 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
26471 * ipa-inline.c (want_inline_small_function_p): Do not cast to
26474 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26476 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26477 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26478 (inline_edge_summary_vec): Turn into ...
26479 (ipa_call_summaries): ... this one.
26480 (redirect_to_unreachable, edge_set_predicate,
26481 evaluate_properties_for_edge, inline_summary_alloc,
26482 reset_ipa_call_summary, reset_inline_summary,
26483 inline_summary_t::duplicate): Update.
26484 (inline_edge_duplication_hook): Turn to ...
26485 (ipa_call_summary_t::duplicate): ... this one.
26486 (inline_edge_removal_hook): Turn to ...
26487 (ipa_call_summary_t::remove): ... this one.
26488 (dump_inline_edge_summary): Turn to ...
26489 (dump_ipa_call_summary): ... this one.
26490 (estimate_function_body_sizes): Update.
26491 (inline_update_callee_summaries): Update.
26492 (remap_edge_change_prob): Update.
26493 (remap_edge_summaries): Update.
26494 (inline_merge_summary): Update.
26495 (do_estimate_edge_time): Update.
26496 (inline_generate_summary): Update.
26497 (inline_read_section): Update.
26498 (inline_read_summary): Update.
26499 (inline_free_summary): Update.
26500 * ipa-inline.c (can_inline_edge_p): Update.
26501 (compute_inlined_call_time): Update.
26502 (want_inline_small_function_p): Update.
26503 (edge_badness): Update.
26504 (early_inliner): Update.
26505 * ipa-inline.h (inline_edge_summary): Turn to ...
26506 (ipa_call_summary): ... this one.
26507 (ipa_call_summary_t): New class.
26508 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26509 (ipa_call_summaries): New.
26510 (inline_edge_summary): Remove.
26511 (estimate_edge_growth): Update.
26512 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26513 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26514 * ipa-split.c (execute_split_functions): Update.
26515 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26517 2017-05-19 Richard Biener <rguenther@suse.de>
26519 PR middle-end/80764
26520 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
26522 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
26524 * config/rs6000/rs6000.c (struct machine_function): Add field
26525 fpr_is_wrapped_separately.
26526 (rs6000_get_separate_components): Use 64 components. Handle the
26527 new FPR components.
26528 (rs6000_components_for_bb): Handle the FPR components.
26529 (rs6000_emit_prologue_components): Handle the FPR components.
26530 (rs6000_emit_epilogue_components): Handle the FPR components.
26531 (rs6000_set_handled_components): Handle the FPR components.
26532 (rs6000_emit_prologue): Don't output prologue code for those FPRs
26533 that are already separately shrink-wrapped.
26534 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
26535 that are already separately shrink-wrapped.
26537 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26540 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
26543 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
26544 (define_peephole2 for Altivec d-form load): Add peepholes to catch
26545 cases where the register allocator uses a move and an offsettable
26546 memory operation to/from a FPR register on ISA 2.06/2.07.
26547 (define_peephole2 for Altivec d-form store): Likewise.
26549 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
26552 * config/i386/mmx.md (*mov<mode>_internal): Enable
26553 alternatives 11, 12, 13 and 14 also for 32bit targets.
26554 Remove alternatives 15, 16, 17 and 18.
26555 * config/i386/sse.md (vec_concatv2di): Change
26556 alternative (!x, *y) to (x, ?!*Yn).
26558 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
26560 * dumpfile.h (enum dump_kind): Remove stray comma.
26562 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26564 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
26565 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
26566 predicate::num_conditions
26567 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
26568 (CHANGED): turn into predicate::changed.
26569 (agg_position_info): Move to ipa-predicate.h
26570 (add_condition, predicate::add_clause, predicate::operator &=,
26571 predicate::or_with, predicate::evaluate, predicate::probability,
26572 dump_condition, dump_clause, predicate::dump,
26573 predicate::remap_after_duplication, predicate::remap_after_inlining,
26574 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
26575 (evaluate_conditions_for_known_args): Update.
26576 (set_cond_stmt_execution_predicate): Update.
26577 * ipa-inline.h: Include ipa-predicate.h
26578 (condition, inline_param_summary, conditions, agg_position_info,
26579 predicate): Move to ipa-predicate.h
26580 * ipa-predicate.c: New file.
26581 * ipa-predicate.h: New file.
26583 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
26585 * final.c (leaf_function_p): Check we are not in a sequence.
26587 2017-05-18 Martin Liska <mliska@suse.cz>
26589 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
26590 * dumpfile.c (dump_register): Use new enum dump_kind.
26591 (get_dump_file_name): Likewise.
26592 (dump_enable_all): Likewise.
26593 (dump_switch_p_1): Likewise.
26594 (enable_rtl_dump_file): Remove usage of TDF_RTL.
26595 * dumpfile.h (enum dump_kind): New enum type.
26596 (struct dump_file_info): Create constructor and
26597 format fields and comments.
26598 * passes.c (pass_manager::register_one_dump_file):
26600 * statistics.c (statistics_early_init): Likewise.
26601 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
26602 TDF_TREE with TDF_SLIM.
26603 (gather_memory_references_ref): Likewise.
26605 2017-05-18 Martin Liska <mliska@suse.cz>
26607 * vec.h (struct vnull): Use it.
26609 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26611 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
26612 (true_predicate, false_predicate, true_predicate_p,
26613 false_predicate_p): Remove.
26614 (single_cond_predicate, not_inlined_predicate): Turn to member function
26616 (add_condition): Update.
26617 (add_clause): Turn to...
26618 (predicate::add_clause): ... this one; update; allow passing NULL
26620 (and_predicates): Turn to ...
26621 (predicate::operator &=): ... this one.
26622 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
26623 (or_predicates): Turn to ...
26624 (predicate::or_with): ... this one.
26625 (evaluate_predicate): Turn to ...
26626 (predicate::evaluate): ... this one.
26627 (predicate_probability): Turn to ...
26628 (predicate::probability): ... this one.
26629 (dump_condition): Update.
26630 (dump_predicate): Turn to ...
26631 (predicate::dump): ... this one.
26632 (account_size_time): Update.
26633 (edge_set_predicate): Update.
26634 (set_hint_predicate): UPdate.
26635 (evaluate_conditions_for_known_args): Update.
26636 (evaluate_properties_for_edge): Update.
26637 (remap_predicate_after_duplication): Turn to...
26638 (predicate::remap_after_duplication): ... this one.
26639 (remap_hint_predicate_after_duplication): Update.
26640 (inline_summary_t::duplicate): UPdate.
26641 (dump_inline_edge_summary): Update.
26642 (dump_inline_summary): Update.
26643 (set_cond_stmt_execution_predicate): Update.
26644 (set_switch_stmt_execution_predicate): Update.
26645 (compute_bb_predicates): Update.
26646 (will_be_nonconstant_expr_predicate): Update.
26647 (will_be_nonconstant_predicate): Update.
26648 (phi_result_unknown_predicate): Update.
26649 (predicate_for_phi_result): Update.
26650 (array_index_predicate): Update.
26651 (estimate_function_body_sizes): Update.
26652 (estimate_node_size_and_time): Update.
26653 (estimate_ipcp_clone_size_and_time): Update.
26654 (remap_predicate): Rename to ...
26655 (predicate::remap_after_inlining): ... this one.
26656 (remap_hint_predicate): Update.
26657 (inline_merge_summary): Update.
26658 (inline_update_overall_summary): Update.
26659 (estimate_size_after_inlining): Update.
26660 (read_predicate): Rename to ...
26661 (predicate::stream_in): ... this one.
26662 (read_inline_edge_summary): Update.
26663 (write_predicate): Rename to ...
26664 (predicate::stream_out): ... this one.
26665 (write_inline_edge_summary): Update.
26666 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
26667 (clause_t): Turn to uint32_t
26668 (predicate): Turn to class; implement constructor and operators
26670 (size_time_entry): Update.
26671 (inline_summary): Update.
26672 (inline_edge_summary): Update.
26674 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
26676 * fold-const.c (fold_binary_loc): Move transformation...
26677 * match.pd (C - X CMP X): ... here.
26679 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
26681 * config/sparc/sparc.c (sparc_option_override): Set function
26682 alignment for -mcpu=niagara7 to 64 to match the I$ line.
26683 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
26685 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
26687 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
26689 2017-05-18 Marek Polacek <polacek@redhat.com>
26692 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
26693 (pass_ubsan::execute): Call gimple_assign_single_p instead of
26694 gimple_assign_load_p.
26696 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
26698 PR middle-end/80692
26699 * real.c (do_compare): Give decimal_do_compare preference over
26700 comparing just the signs.
26702 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
26704 * doc/md.texi (Canonicalization of Instructions): Describe the
26705 canonical form of instructions that inherently set a condition
26708 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
26710 PR middle-end/80775
26711 * tree-cfg.c: Move deletion of unreachable case statements to after
26712 the merging of consecutive case labels.
26714 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
26716 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
26717 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
26718 restoring of callee-saved registers.
26720 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
26722 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
26723 * config/visium/visium.c (single_set_and_flags): Likewise.
26724 * config/visium/visium.md (Substitutions): Likewise.
26726 2017-05-17 Martin Liska <mliska@suse.cz>
26728 * cfg.c: Introduce dump_flags_t type and
26729 use it instead of int type.
26731 * cfghooks.c: Likewise.
26732 * cfghooks.h (struct cfg_hooks): Likewise.
26733 * cfgrtl.c: Likewise.
26734 * cfgrtl.h: Likewise.
26735 * cgraph.c (cgraph_node::get_body): Likewise.
26736 * coretypes.h: Likewise.
26737 * domwalk.c: Likewise.
26738 * domwalk.h: Likewise.
26739 * dumpfile.c (struct dump_option_value_info): Likewise.
26740 (dump_enable_all): Likewise.
26741 (dump_switch_p_1): Likewise.
26742 (opt_info_switch_p): Likewise.
26743 * dumpfile.h (enum tree_dump_index): Likewise.
26744 (struct dump_file_info): Likewise.
26745 * genemit.c: Likewise.
26746 * generic-match-head.c: Likewise.
26747 * gengtype.c (open_base_files): Likewise.
26748 * gimple-pretty-print.c: Likewise.
26749 * gimple-pretty-print.h: Likewise.
26750 * graph.c (print_graph_cfg): Likewise.
26751 * graphite-scop-detection.c (dot_all_sese): Likewise.
26752 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
26753 * loop-unroll.c (report_unroll): Likewise.
26754 * passes.c (pass_manager::register_one_dump_file): Likewise.
26755 * print-tree.c: Likewise.
26756 * statistics.c: Likewise.
26757 * tree-cfg.c: Likewise.
26758 * tree-cfg.h: Likewise.
26759 * tree-dfa.c: Likewise.
26760 * tree-dfa.h: Likewise.
26761 * tree-dump.c (dump_function): Likewise.
26762 * tree-dump.h (struct dump_info): Likewise.
26763 * tree-pretty-print.c: Likewise.
26764 * tree-pretty-print.h: Likewise.
26765 * tree-ssa-live.c: Likewise.
26766 * tree-ssa-live.h: Likewise.
26767 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
26768 * tree-vect-loop.c: Likewise.
26769 * tree-vect-slp.c: Likewise.
26771 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
26772 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26774 PR tree-optimization/80457
26775 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
26776 of all arguments to a statement as scalar_to_vec operations.
26777 (vectorizable_call): Adjust call to vect_model_simple_cost for
26779 (vectorizable_conversion): Likewise.
26780 (vectorizable_assignment): Likewise.
26781 (vectorizable_shift): Likewise.
26782 (vectorizable_operation): Likewise.
26783 (vectorizable_comparison): Likewise.
26784 (vect_is_simple_cond): Record the def types for operands.
26785 (vectorizable_condition): Likewise, call vect_model_simple_cost.
26786 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
26787 for statement argument count.
26789 2017-05-16 Carl Love <cel@us.ibm.com>
26791 * config/rs6000/rs6000-c: Add support for built-in functions
26792 vector unsigned long long vec_bperm (vector unsigned long long,
26793 vector unsigned char)
26794 vector signed long long vec_mule (vector signed int,
26796 vector unsigned long long vec_mule (vector unsigned int,
26797 vector unsigned int)
26798 vector signed long long vec_mulo (vector signed int,
26800 vector unsigned long long vec_mulo (vector unsigned int,
26801 vector unsigned int)
26802 vector signed char vec_sldw (vector signed char,
26803 vector signed char,
26805 vector unsigned char vec_sldw (vector unsigned char,
26806 vector unsigned char,
26808 vector signed short vec_sldw (vector signed short,
26809 vector signed short,
26811 vector unsigned short vec_sldw (vector unsigned short,
26812 vector unsigned short,
26814 vector signed int vec_sldw (vector signed int,
26817 vector unsigned int vec_sldw (vector unsigned int,
26818 vector unsigned int,
26820 vector signed long long vec_sldw (vector signed long long,
26821 vector signed long long,
26823 vector unsigned long long vec_sldw (vector unsigned long long,
26824 vector unsigned long long,
26826 * config/rs6000/rs6000-c: Add support for built-in functions
26827 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
26828 * config/rs6000/altivec.h: Add defintion for vec_sldw.
26829 * doc/extend.texi: Update the built-in documentation for the
26830 new built-in functions.
26832 2017-05-16 Marek Polacek <polacek@redhat.com>
26836 * tree.c (save_expr): Don't fold the expression.
26838 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
26840 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
26841 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
26842 and (?*y,m). Update insn attributes.
26844 2017-05-16 Martin Liska <mliska@suse.cz>
26846 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
26847 flags argument of print_gimple_stmt, print_gimple_expr,
26848 print_generic_stmt and print_generic_expr.
26849 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
26850 * coretypes.h: Likewise.
26851 * except.c (dump_eh_tree): Likewise.
26852 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
26853 * gimple-pretty-print.h: Likewise.
26854 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
26855 (backprop::push_to_worklist): Likewise.
26856 (backprop::pop_from_worklist): Likewise.
26857 (backprop::process_use): Likewise.
26858 (backprop::intersect_uses): Likewise.
26859 (note_replacement): Likewise.
26860 * gimple-ssa-store-merging.c
26861 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
26862 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
26863 (pass_store_merging::execute): Likewise.
26864 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
26865 (ssa_base_cand_dump_callback): Likewise.
26866 (dump_incr_vec): Likewise.
26867 (replace_refs): Likewise.
26868 (replace_mult_candidate): Likewise.
26869 (create_add_on_incoming_edge): Likewise.
26870 (create_phi_basis): Likewise.
26871 (insert_initializers): Likewise.
26872 (all_phi_incrs_profitable): Likewise.
26873 (introduce_cast_before_cand): Likewise.
26874 (replace_one_candidate): Likewise.
26875 * gimplify.c (gimplify_expr): Likewise.
26876 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
26877 (set_rename): Likewise.
26878 (rename_uses): Likewise.
26879 (copy_loop_phi_nodes): Likewise.
26880 (add_close_phis_to_merge_points): Likewise.
26881 (copy_loop_close_phi_args): Likewise.
26882 (copy_cond_phi_args): Likewise.
26883 (graphite_copy_stmts_from_block): Likewise.
26884 (translate_pending_phi_nodes): Likewise.
26885 * graphite-poly.c (print_pdr): Likewise.
26886 (dump_gbb_cases): Likewise.
26887 (dump_gbb_conditions): Likewise.
26888 (print_scop_params): Likewise.
26889 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
26890 (build_cross_bb_scalars_use): Likewise.
26891 (gather_bbs::before_dom_children): Likewise.
26892 * hsa-dump.c (dump_hsa_immed): Likewise.
26893 * ipa-cp.c (print_ipcp_constant_value): Likewise.
26894 (get_replacement_map): Likewise.
26895 * ipa-inline-analysis.c (dump_condition): Likewise.
26896 (estimate_function_body_sizes): Likewise.
26897 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
26898 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
26899 * ipa-prop.c (ipa_dump_param): Likewise.
26900 (ipa_print_node_jump_functions_for_edge): Likewise.
26901 (ipa_modify_call_arguments): Likewise.
26902 (ipa_modify_expr): Likewise.
26903 (ipa_dump_param_adjustments): Likewise.
26904 (ipa_dump_agg_replacement_values): Likewise.
26905 (ipcp_modif_dom_walker::before_dom_children): Likewise.
26906 * ipa-pure-const.c (check_stmt): Likewise.
26907 (pass_nothrow::execute): Likewise.
26908 * ipa-split.c (execute_split_functions): Likewise.
26909 * omp-offload.c (dump_oacc_loop_part): Likewise.
26910 (dump_oacc_loop): Likewise.
26911 * trans-mem.c (tm_log_emit): Likewise.
26912 (tm_memopt_accumulate_memops): Likewise.
26913 (dump_tm_memopt_set): Likewise.
26914 (dump_tm_memopt_transform): Likewise.
26915 * tree-cfg.c (gimple_verify_flow_info): Likewise.
26916 (print_loop): Likewise.
26917 * tree-chkp-opt.c (chkp_print_addr): Likewise.
26918 (chkp_gather_checks_info): Likewise.
26919 (chkp_get_check_result): Likewise.
26920 (chkp_remove_check_if_pass): Likewise.
26921 (chkp_use_outer_bounds_if_possible): Likewise.
26922 (chkp_reduce_bounds_lifetime): Likewise.
26923 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
26924 (chkp_mark_completed_bounds): Likewise.
26925 (chkp_register_incomplete_bounds): Likewise.
26926 (chkp_mark_invalid_bounds): Likewise.
26927 (chkp_maybe_copy_and_register_bounds): Likewise.
26928 (chkp_build_returned_bound): Likewise.
26929 (chkp_get_bound_for_parm): Likewise.
26930 (chkp_build_bndldx): Likewise.
26931 (chkp_get_bounds_by_definition): Likewise.
26932 (chkp_generate_extern_var_bounds): Likewise.
26933 (chkp_get_bounds_for_decl_addr): Likewise.
26934 * tree-chrec.c (chrec_apply): Likewise.
26935 * tree-data-ref.c (dump_data_reference): Likewise.
26936 (dump_subscript): Likewise.
26937 (dump_data_dependence_relation): Likewise.
26938 (analyze_overlapping_iterations): Likewise.
26939 * tree-inline.c (expand_call_inline): Likewise.
26940 (tree_function_versioning): Likewise.
26941 * tree-into-ssa.c (dump_defs_stack): Likewise.
26942 (dump_currdefs): Likewise.
26943 (dump_names_replaced_by): Likewise.
26944 (dump_update_ssa): Likewise.
26945 (update_ssa): Likewise.
26946 * tree-object-size.c (pass_object_sizes::execute): Likewise.
26947 * tree-parloops.c (build_new_reduction): Likewise.
26948 (try_create_reduction_list): Likewise.
26949 (ref_conflicts_with_region): Likewise.
26950 (oacc_entry_exit_ok_1): Likewise.
26951 (oacc_entry_exit_single_gang): Likewise.
26952 * tree-pretty-print.h: Likewise.
26953 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
26954 (get_scalar_evolution): Likewise.
26955 (add_to_evolution): Likewise.
26956 (get_loop_exit_condition): Likewise.
26957 (analyze_evolution_in_loop): Likewise.
26958 (analyze_initial_condition): Likewise.
26959 (analyze_scalar_evolution): Likewise.
26960 (instantiate_scev): Likewise.
26961 (number_of_latch_executions): Likewise.
26962 (gather_chrec_stats): Likewise.
26963 (final_value_replacement_loop): Likewise.
26964 (scev_const_prop): Likewise.
26965 * tree-sra.c (dump_access): Likewise.
26966 (disqualify_candidate): Likewise.
26967 (create_access): Likewise.
26968 (reject): Likewise.
26969 (maybe_add_sra_candidate): Likewise.
26970 (create_access_replacement): Likewise.
26971 (analyze_access_subtree): Likewise.
26972 (analyze_all_variable_accesses): Likewise.
26973 (sra_modify_assign): Likewise.
26974 (initialize_constant_pool_replacements): Likewise.
26975 (find_param_candidates): Likewise.
26976 (decide_one_param_reduction): Likewise.
26977 (replace_removed_params_ssa_names): Likewise.
26978 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
26979 * tree-ssa-copy.c (dump_copy_of): Likewise.
26980 (copy_prop_visit_cond_stmt): Likewise.
26981 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
26982 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
26983 (record_equivalences_from_stmt): Likewise.
26984 * tree-ssa-dse.c (compute_trims): Likewise.
26985 (delete_dead_call): Likewise.
26986 (delete_dead_assignment): Likewise.
26987 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
26988 (forward_propagate_into_cond): Likewise.
26989 (pass_forwprop::execute): Likewise.
26990 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
26991 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
26993 (move_computations_worker): Likewise.
26994 (execute_sm): Likewise.
26995 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
26996 (remove_exits_and_undefined_stmts): Likewise.
26997 (remove_redundant_iv_tests): Likewise.
26998 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
26999 (adjust_iv_update_pos): Likewise.
27000 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27001 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27002 (value_replacement): Likewise.
27003 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27004 * tree-ssa-pre.c (print_pre_expr): Likewise.
27005 (get_representative_for): Likewise.
27006 (create_expression_by_pieces): Likewise.
27007 (insert_into_preds_of_block): Likewise.
27008 (eliminate_insert): Likewise.
27009 (eliminate_dom_walker::before_dom_children): Likewise.
27010 (eliminate): Likewise.
27011 (remove_dead_inserted_code): Likewise.
27012 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27013 * tree-ssa-reassoc.c (get_rank): Likewise.
27014 (eliminate_duplicate_pair): Likewise.
27015 (eliminate_plus_minus_pair): Likewise.
27016 (eliminate_not_pairs): Likewise.
27017 (undistribute_ops_list): Likewise.
27018 (eliminate_redundant_comparison): Likewise.
27019 (update_range_test): Likewise.
27020 (optimize_range_tests_var_bound): Likewise.
27021 (optimize_vec_cond_expr): Likewise.
27022 (rewrite_expr_tree): Likewise.
27023 (rewrite_expr_tree_parallel): Likewise.
27024 (linearize_expr): Likewise.
27025 (break_up_subtract): Likewise.
27026 (linearize_expr_tree): Likewise.
27027 (attempt_builtin_powi): Likewise.
27028 (attempt_builtin_copysign): Likewise.
27029 (transform_stmt_to_copy): Likewise.
27030 (transform_stmt_to_multiply): Likewise.
27031 (dump_ops_vector): Likewise.
27032 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27033 (print_scc): Likewise.
27034 (set_ssa_val_to): Likewise.
27035 (visit_reference_op_store): Likewise.
27036 (visit_use): Likewise.
27037 (sccvn_dom_walker::before_dom_children): Likewise.
27038 (run_scc_vn): Likewise.
27039 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27041 (expr_hash_elt::print): Likewise.
27042 (const_and_copies::pop_to_marker): Likewise.
27043 (const_and_copies::record_const_or_copy_raw): Likewise.
27044 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27045 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27046 (dump_predicates): Likewise.
27047 (find_uninit_use): Likewise.
27048 (warn_uninitialized_phi): Likewise.
27049 (pass_late_warn_uninitialized::execute): Likewise.
27050 * tree-ssa.c (verify_vssa): Likewise.
27051 (verify_ssa): Likewise.
27052 (maybe_optimize_var): Likewise.
27053 * tree-vrp.c (dump_value_range): Likewise.
27054 (dump_all_value_ranges): Likewise.
27055 (dump_asserts_for): Likewise.
27056 (register_edge_assert_for_2): Likewise.
27057 (vrp_visit_cond_stmt): Likewise.
27058 (vrp_visit_switch_stmt): Likewise.
27059 (vrp_visit_stmt): Likewise.
27060 (vrp_visit_phi_node): Likewise.
27061 (simplify_cond_using_ranges_1): Likewise.
27062 (fold_predicate_in): Likewise.
27063 (evrp_dom_walker::before_dom_children): Likewise.
27064 (evrp_dom_walker::push_value_range): Likewise.
27065 (evrp_dom_walker::pop_value_range): Likewise.
27066 (execute_early_vrp): Likewise.
27068 2017-05-16 Richard Biener <rguenther@suse.de>
27070 * dwarf2out.c (loc_list_from_tree_1): Do not create
27071 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27073 2017-05-16 Richard Biener <rguenther@suse.de>
27075 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27077 (note_variable_value_in_expr): If we resolved the decl ref
27078 do not push to the stack.
27080 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27082 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27083 operations in fast-math mode.
27084 (vaddq_f16): Likewise.
27085 (vmul_f16): Likewise.
27086 (vmulq_f16): Likewise.
27087 (vsub_f16): Likewise.
27088 (vsubq_f16): Likewise.
27089 * config/arm/neon.md (add<mode>3): New.
27091 (fma:<VH:mode>3): New. Also remove outdated comment.
27094 2017-05-16 Martin Liska <mliska@suse.cz>
27098 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27099 (odr_types_equivalent_p): Likewise.
27101 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27103 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27105 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27108 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27109 non-interunit SSE move alternatives with '?'.
27110 (zero-extendsidi peephole2): New peephole to skip intermediate
27111 general register in SSE zero-extend sequence.
27113 2017-05-15 Jeff Law <law@redhat.com>
27115 * reorg.c (relax_delay_slots): Create a new variable to hold
27116 the temporary target rather than clobbering TARGET_LABEL.
27118 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27119 missing argument to extract_bit_field call.
27120 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27122 2017-05-15 Martin Liska <mliska@suse.cz>
27125 * gcc.c (process_command): Do not allow empty argument of -o option.
27127 2017-05-15 Renlin Li <renlin.li@arm.com>
27129 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27130 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27131 * config/aarch64/constraints.md (Usf): Add long call check.
27132 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27133 (call_value): Likewise.
27134 (sibcall): Likewise.
27135 (sibcall_value): Likewise.
27137 (call_value_insn): New.
27138 (sibcall_insn): Update rtx pattern.
27139 (sibcall_value_insn): Likewise.
27140 (call_internal): Remove.
27141 (call_value_internal): Likewise.
27142 (sibcall_internal): Likewise.
27143 (sibcall_value_internal): Likewise.
27144 (call_reg): Likewise.
27145 (call_symbol): Likewise.
27146 (call_value_reg): Likewise.
27147 (call_value_symbol): Likewise.
27149 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27152 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27154 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27156 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27157 compatible with CCGOCmode and with CCZmode.
27159 2017-05-14 Martin Sebor <msebor@redhat.com>
27161 PR middle-end/77671
27162 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27163 (gimple_fold_builtin_snprintf): Same.
27164 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27165 (gimple_fold_builtin_snprintf): Same.
27166 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27167 of character types.
27168 (is_call_safe): New function.
27169 (try_substitute_return_value): Call it.
27170 (try_simplify_call): New function.
27171 (pass_sprintf_length::handle_gimple_call): Call it.
27173 2017-05-14 Martin Sebor <msebor@redhat.com>
27175 PR middle-end/80669
27176 * builtins.c (expand_builtin_stpncpy): Simplify.
27178 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27180 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27181 * config/i386/i386.h
27182 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27183 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27184 (struct machine_function): Add new members call_ms2sysv,
27185 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27186 (struct machine_frame_state): New fields sp_realigned and
27187 sp_realigned_offset.
27188 * config/i386/i386.c
27189 (enum xlogue_stub): New enum.
27190 (enum xlogue_stub_sets): New enum.
27191 (class xlogue_layout): New class.
27192 (struct ix86_frame): New fields stack_realign_allocate_offset,
27193 stack_realign_offset and outlined_save_offset. Modify comments to
27194 detail stack layout when using out-of-line stubs.
27195 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27196 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27197 -mcall-ms2sysv-xlogues.
27198 (stub_managed_regs): New static variable.
27199 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27200 registers managed by out-of-line stub.
27201 (disable_call_ms2sysv_xlogues): New function.
27202 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27203 m->call_ms2sysv when appropriate and compute frame layout for
27205 (sp_valid_at, fp_valid_at): New inline functions.
27206 (choose_basereg): New function.
27207 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27209 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27210 Use align parameter of choose_baseaddr to generated aligned SSE movs
27212 (pro_epilogue_adjust_stack): Modify to track
27213 machine_frame_state::sp_realigned.
27214 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
27215 (ix86_nsaved_sseregs): Likewise.
27216 (ix86_emit_save_regs): Likewise.
27217 (ix86_emit_save_regs_using_mov): Likewise.
27218 (ix86_emit_save_sse_regs_using_mov): Likewise.
27219 (get_scratch_register_on_entry): Likewise.
27220 (gen_frame_set): New function.
27221 (gen_frame_load): Likewise.
27222 (gen_frame_store): Likewise.
27223 (emit_outlined_ms2sysv_save): Likewise.
27224 (emit_outlined_ms2sysv_restore): Likewise.
27225 (ix86_expand_prologue): Modify stack re-alignment code and call
27226 emit_outlined_ms2sysv_save when appropriate.
27227 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
27228 parameter rtx_insn *insn, which allows the function to be used to only
27229 generate the notes.
27230 (ix86_expand_epilogue): Modify validity checks of frame and stack
27231 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
27232 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
27233 * config/i386/predicates.md
27234 (save_multiple): New predicate.
27235 (restore_multiple): Likewise.
27236 * config/i386/sse.md
27237 (save_multiple<mode>): New pattern.
27238 (save_multiple_realign<mode>): Likewise.
27239 (restore_multiple<mode>): Likewise.
27240 (restore_multiple_and_return<mode>): Likewise.
27241 (restore_multiple_leave_return<mode>): Likewise.
27242 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
27244 2017-05-14 Julia Koval <julia.koval@intel.com>
27246 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
27247 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
27248 (__builtin_ia32_xsetbv): New builtins.
27249 * config/i386/i386.c (ix86_expand_special_args_builtin):
27251 (ix86_expand_builtin): Special expand for new intrinsics.
27252 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
27253 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
27254 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
27256 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27258 * cfganal.c (inverted_post_order_compute): Change argument type
27260 * cfganal.h (inverted_post_order_compute): Adjust prototype.
27261 * df-core.c (rest_of_handle_df_initialize): Adjust.
27262 (rest_of_handle_df_finish): Likewise.
27263 (df_analyze_1): Likewise.
27264 (df_analyze): Likewise.
27265 (loop_inverted_post_order_compute): Change argument to be a vec *.
27266 (df_analyze_loop): Adjust.
27267 (df_get_n_blocks): Likewise.
27268 (df_get_postorder): Likewise.
27269 * df.h (struct df_d): Change field to be a vec.
27270 * lcm.c (compute_laterin): Adjust.
27271 (compute_available): Likewise.
27272 * lra-lives.c (lra_create_live_ranges_1): Likewise.
27273 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
27274 * tree-ssa-pre.c (compute_antic): Likewise.
27276 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27278 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
27279 (depth_first_search::depth_first_search): Change structure init
27280 function to this constructor.
27281 (depth_first_search::add_bb): Rename function to this member.
27282 (depth_first_search::execute): Likewise.
27283 (flow_dfs_compute_reverse_finish): Adjust.
27285 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27287 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
27288 (longest_simple_path): Likewise.
27289 * shrink-wrap.c (spread_components): Likewise.
27290 (disqualify_problematic_components): Likewise.
27291 (emit_common_heads_for_components): Likewise.
27292 (emit_common_tails_for_components): Likewise.
27293 (insert_prologue_epilogue_for_components): Likewise.
27295 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27297 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
27300 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27302 * df-core.c (df_set_blocks): Start using auto_bitmap.
27303 (df_compact_blocks): Likewise.
27304 * df-problems.c (df_rd_confluence_n): Likewise.
27305 * df-scan.c (df_insn_rescan_all): Likewise.
27306 (df_process_deferred_rescans): Likewise.
27307 (df_update_entry_block_defs): Likewise.
27308 (df_update_exit_block_uses): Likewise.
27309 (df_entry_block_bitmap_verify): Likewise.
27310 (df_exit_block_bitmap_verify): Likewise.
27311 (df_scan_verify): Likewise.
27312 * lra-constraints.c (lra_constraints): Likewise.
27313 (undo_optional_reloads): Likewise.
27314 (lra_undo_inheritance): Likewise.
27315 * lra-remat.c (calculate_gen_cands): Likewise.
27316 (do_remat): Likewise.
27317 * lra-spills.c (assign_spill_hard_regs): Likewise.
27318 (spill_pseudos): Likewise.
27319 * tree-ssa-pre.c (bitmap_set_and): Likewise.
27320 (bitmap_set_subtract_values): Likewise.
27322 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27324 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
27325 management with auto_bitmap.
27326 (fix_inter_tick): Likewise.
27327 (fix_recovery_deps): Likewise.
27328 * ira.c (add_store_equivs): Likewise.
27329 (find_moveable_pseudos): Likewise.
27330 (split_live_ranges_for_shrink_wrap): Likewise.
27331 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
27332 (rtx_reuse_manager::seen_def_p): Likewise.
27333 (rtx_reuse_manager::set_seen_def): Likewise.
27334 * print-rtl.h (class rtx_reuse_manager): Likewise.
27336 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27338 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
27340 (migrate_btr_def): Likewise.
27341 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
27342 * df-core.c (loop_post_order_compute): Likewise.
27343 (loop_inverted_post_order_compute): Likewise.
27344 * hsa-common.h: Likewise.
27345 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
27346 * init-regs.c (initialize_uninitialized_regs): Likewise.
27347 * ipa-inline.c (resolve_noninline_speculation): Likewise.
27348 (inline_small_functions): Likewise.
27349 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
27350 * ira.c (combine_and_move_insns): Likewise.
27351 (build_insn_chain): Likewise.
27352 * loop-invariant.c (find_invariants): Likewise.
27353 * lower-subreg.c (propagate_pseudo_copies): Likewise.
27354 * predict.c (tree_predict_by_opcode): Likewise.
27355 (predict_paths_leading_to): Likewise.
27356 (predict_paths_leading_to_edge): Likewise.
27357 (estimate_loops_at_level): Likewise.
27358 (estimate_loops): Likewise.
27359 * shrink-wrap.c (try_shrink_wrapping): Likewise.
27360 (spread_components): Likewise.
27361 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
27362 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
27363 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
27364 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
27365 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
27366 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
27367 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
27368 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
27369 (mark_threaded_blocks): Likewise.
27370 (thread_through_all_blocks): Likewise.
27371 * tree-ssa.c (verify_ssa): Likewise.
27372 (execute_update_addresses_taken): Likewise.
27373 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
27375 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27377 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
27379 (post_order_compute): Likewise.
27380 (inverted_post_order_compute): Likewise.
27381 (pre_and_rev_post_order_compute_fn): Likewise.
27383 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27385 * genrecog.c (int_set::int_set): Explicitly construct our
27386 auto_vec base class.
27387 * vec.h (auto_vec::auto_vec): New constructor.
27389 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27391 * bitmap.h (class auto_bitmap): New constructor taking
27392 bitmap_obstack * argument.
27394 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27396 * bitmap.h (class auto_bitmap): Change type of m_bits to
27397 bitmap_head, and adjust ctor / dtor and member operators.
27399 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
27401 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
27402 when returned register mode doesn't match original mode.
27404 2017-05-12 Jeff Law <law@redhat.com>
27405 Jakub Jelinek <jakub@redhat.com>
27407 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
27408 we look for cc setter after the compare-elim changes.
27409 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
27410 within the vector to match what compare-elim now expects.
27411 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
27412 (xorsi3_flags, one_cmplsi2_flags): Likewise.
27414 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
27415 after the compare-elim changes.
27416 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
27417 the vector to match what compare-elim now expects.
27418 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
27419 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
27420 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
27421 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
27422 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
27424 * config/visium/visium.c (single_set_and_flags): Fix where
27425 we look for cc setter after the compare-elim changes.
27426 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
27427 with the vector to match what compare-elim now expects.
27428 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
27429 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
27430 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
27431 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
27432 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
27433 (neg<mode>2_insn_set_overflow): Likewise.
27435 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
27437 PR middle-end/79794
27438 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
27439 maybe_expand_insn call, set ops[0].target. If still set after call,
27440 set alt_rtl. Add extra arg to recursive calls.
27441 (extract_bit_field): Add alt_rtl argument. Pass to
27443 * expmed.h (extract_bit_field): Fix prototype.
27444 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
27445 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
27446 to extract_bit_field_calls.
27447 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
27448 Pass alt_rtl to extract_bit_field calls.
27449 * calls.c (store_unaligned_arguments_into_psuedos)
27450 load_register_parameters): Pass extra NULL to extract_bit_field calls.
27451 * optabs.c (maybe_legitimize_operand): Clear op->target when call
27453 * optabs.h (struct expand_operand): Add target bitfield.
27455 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27457 * compare-elim.c (try_eliminate_compare): Canonicalize
27458 operation with embedded compare to
27459 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
27460 (set (reg) (operation)].
27462 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
27464 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27467 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
27468 cost of adding a carry flag for ADC instruction.
27469 [case MINUS]: Ignore the cost of subtracting a carry flag
27470 for SBB instruction.
27472 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
27474 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
27476 * config/rs6000/bmiintrin.h: New file.
27477 * config/rs6000/bmi2intrin.h: New file.
27478 * config/rs6000/x86intrin.h: New file.
27480 2017-05-12 Jeff Law <law@redhat.com>
27482 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
27485 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
27487 PR middle-end/80707
27488 * tree-cfg.c: Remove cfg edges of unreachable case statements.
27490 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27492 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27493 early expansion of vector divide builtins.
27494 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
27495 builtins identified as having unsigned arguments.
27497 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27499 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
27500 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
27501 expansion of vector logical operations (and, andc, or, xor,
27504 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27506 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
27507 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
27509 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27511 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27512 early GIMPLE expansion of vector multiplies.
27514 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27516 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
27517 TARGET_HAVE_MOVT conditional.
27518 (movt splitter): Likewise.
27520 2017-05-12 Richard Biener <rguenther@suse.de>
27522 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
27523 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27524 Fold all stmts not inplace.
27526 2017-05-12 Richard Biener <rguenther@suse.de>
27528 PR tree-optimization/80713
27529 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
27530 inserted_exprs bit for not removed stmts.
27532 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
27534 PR middle-end/69921
27535 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
27536 parallelized" attribute for parallelized OpenACC kernels.
27537 * omp-offload.c (execute_oacc_device_lower): Use it.
27539 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
27540 Set "oacc kernels" attribute.
27541 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
27542 parameter. Adjust all users.
27543 (oacc_fn_attrib_kernels_p): Remove function.
27544 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
27545 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
27546 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
27547 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
27548 assert "oacc kernels" attribute is set.
27550 2017-05-11 Carl Love <cel@us.ibm.com>
27552 * config/rs6000/rs6000-c: Add support for built-in functions
27553 vector unsigned char vec_popcnt (vector signed char)
27554 vector unsigned char vec_popcnt (vector unsigned char)
27555 vector unsigned short vec_popcnt (vector signed short)
27556 vector unsigned short vec_popcnt (vector unsigned short)
27557 vector unsigned int vec_popcnt (vector signed int)
27558 vector unsigned int vec_popcnt (vector unsigned int)
27559 vector unsigned long long vec_popcnt (vector signed long long)
27560 vector unsigned long long vec_popcnt (vector unsigned long long)
27561 vector signed long long vec_slo (vector signed long long,
27562 vector signed char)
27563 vector signed long long vec_slo (vector signed long long,
27564 vector unsigned char)
27565 vector unsigned long long vec_slo (vector unsigned long long,
27566 vector signed char)
27567 vector unsigned long long vec_slo (vector unsigned long long,
27568 vector unsigned char)
27569 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
27570 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
27571 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
27572 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
27573 * doc/extend.texi: Update the built-in documentation file for the
27574 new built-in functions.
27576 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27578 * attribs.h (sorted_attr_string): Move machine independent
27579 functions for target clone support from the i386 port to common
27580 code. Rename ix86_function_versions to common_function_versions.
27581 Rename make_name to make_unique_name.
27582 (common_function_versions): Likewise.
27583 (make_unique_name): Likewise.
27584 (make_dispatcher_decl): Likewise.
27585 (is_function_default_version): Likewise.
27586 * attribs.c (attr_strcmp): Likewise.
27587 (sorted_attr_string): Likewise.
27588 (common_function_versions): Likewise.
27589 (make_unique_name): Likewise.
27590 (make_dispatcher_decl): Likewise.
27591 (is_function_default_version): Likewise.
27592 * config/i386/i386.c (attr_strcmp): Likewise.
27593 (sorted_attr_string): Likewise.
27594 (ix86_function_versions): Likewise.
27595 (make_name): Likewise.
27596 (make_dispatcher_decl): Likewise.
27597 (is_function_default_version): Likewise.
27598 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
27600 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27603 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
27604 Account for direct move costs for vec_construct of integer
27607 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
27610 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
27611 (UNSPEC_STX_ATOMIC): Ditto.
27612 (loaddi_via_sse): New insn.
27613 (storedi_via_sse): Ditto.
27614 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
27615 Update corresponding peephole2 patterns.
27616 (atomic_storedi_fpu): Ditto.
27618 2017-05-11 Julia Koval <julia.koval@intel.com>
27620 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
27621 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
27623 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
27624 (__builtin_ia32_rsqrt14ss_mask): New builtins.
27625 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
27627 2017-05-11 Nathan Sidwell <nathan@acm.org>
27629 * graphite-poly.c: Include dumpfile.h.
27631 * dumpfle.h (dump_function): Declare here ...
27632 * tree-dump.h (dump_function): ... not here.
27633 * dumpfile.c: #include tree-cfg.h.
27634 (dump_function): Move here from ...
27635 * tree-dump.c (dump_function): ... here.
27636 * gimplify.c: #include splay-tree.h, not tree-dump.h.
27637 * graphite-poly.c: Don't include tree-dump.h.
27638 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
27639 * print-tree.c: Likewise.
27640 * stor-layout.c: Likewise.
27641 * tree-nested.c: Likewise.
27643 * dumpfile.c (dump_start): Use TDF_FLAGS.
27644 (dump_enable_all): Fix TDF_KIND check thinko.
27646 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
27648 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27649 array entries to represent two legal parameterizations of the
27650 overloaded __builtin_cmpb function, as represented by the
27651 P6_OV_BUILTIN_CMPB constant.
27652 (altivec_resolve_overloaded_builtin): Add special case handling
27653 for the __builtin_cmpb function, as represented by the
27654 P6_OV_BUILTIN_CMPB constant.
27655 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
27656 (BU_P6_64BIT_2): New macro.
27657 (BU_P6_OVERLOAD_2): New macro
27658 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
27659 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
27660 (CMPB): Add overload support to represent both 32-bit and 64-bit
27661 compare-bytes function.
27662 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
27663 support for TARGET_CMPB.
27664 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
27665 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27666 documentation of the __builtin_cmpb overloaded built-in function.
27668 2017-05-11 Richard Biener <rguenther@suse.de>
27670 PR tree-optimization/80705
27671 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
27672 bases are not vectorizable.
27674 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27676 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
27677 when counting register pressure.
27679 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27681 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
27682 (iv_ca_more_deps): Renamed to ...
27683 (iv_ca_compare_deps): ... this.
27684 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
27686 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27688 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
27690 (determine_group_iv_costs): ... here.
27691 (find_inv_vars_cb): Record inv var if it's not recorded before.
27693 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27695 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
27696 (get_shiftadd_cost): Ditto.
27698 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27700 * tree-ssa-address.c: Include header file.
27701 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
27703 (add_to_parts): Refactor.
27704 (addr_to_parts): New parameter. Update use of move_hint_to_base.
27705 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
27708 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27710 PR tree-optimization/53090
27711 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
27713 (extract_cond_operands): Detect condition with IV on both sides
27714 and return COMP_IV_EXPR_2.
27715 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
27716 (rewrite_use_compare): Simplify by removing call to function
27717 extract_cond_operands.
27719 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27721 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
27722 (extract_cond_operands): Detect condition comparing against non-
27723 invariant bound and return appropriate enum value.
27724 (find_interesting_uses_cond): Update use of extract_cond_operands.
27725 Handle its return value accordingly.
27726 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
27728 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27730 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
27731 nonlinear iv_use computation in loop invariant sensitive way.
27733 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27735 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
27736 (find_iv_candidates): Call relate_compare_use_with_all_cands.
27738 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27740 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
27741 (dump_cand): Support iv_cand.inv_exprs.
27742 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
27744 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
27747 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27749 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
27751 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
27752 as local function. Include necessary header files.
27753 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
27755 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27757 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
27759 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27761 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
27762 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
27763 RSHIFT_EXPR and BIT_NOT_EXPR.
27765 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27767 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
27768 (adjust_setup_cost): New parameter supporting round up adjustment.
27769 (struct address_cost_data): Delete.
27770 (force_expr_to_var_cost): Don't bound cost with spill_cost.
27771 (split_address_cost, ptr_difference_cost): Delete.
27772 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
27773 (struct ainc_cost_data): New struct.
27774 (get_address_cost_ainc): New function.
27775 (get_address_cost, get_computation_cost): Reimplement.
27776 (determine_group_iv_cost_address): Record inv_expr for all uses of
27778 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
27779 (iv_ca_has_deps): Reimplemented to ...
27780 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
27782 (iv_ca_extend): Call iv_ca_more_deps.
27784 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27786 * tree-ssa-address.c (struct mem_address): Move to header file.
27787 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
27788 * tree-ssa-address.h (struct mem_address): Move from C file.
27789 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
27791 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27793 * tree-affine.h (aff_combination_type): New interface.
27794 (aff_combination_zero_p): Remove static.
27795 (aff_combination_const_p): New interface.
27796 (aff_combination_singleton_var_p): New interfaces.
27798 2017-05-11 Richard Biener <rguenther@suse.de>
27800 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27801 Skip unreachable blocks and destinations.
27802 (eliminate): Move stmt removal and fixup ...
27803 (fini_eliminate): ... here. Skip inserted exprs.
27804 (pass_pre::execute): Move fini_pre after fini_eliminate.
27805 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
27806 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
27807 PRE to get rid of dead code that has invalid SSA form and
27808 split critical edges again.
27810 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27812 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27814 2017-05-11 Richard Biener <rguenther@suse.de>
27816 * passes.c (execute_function_todo): Verify loops if they are
27817 said to be up-to-date.
27818 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
27819 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
27821 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
27824 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
27825 handle calling assemble_external ourself.
27828 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
27829 modes with zero size. Enhance comment.
27831 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27833 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
27834 built-ins for vec_xl and vec_xst with short and char pointer
27837 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27839 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
27840 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
27841 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
27842 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
27843 (_mm_maskz_min_round_ss): New intrinsics.
27844 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
27845 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
27846 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
27847 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
27848 (__builtin_ia32_minss_mask_round): New builtins.
27849 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27850 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27851 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
27853 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
27854 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
27856 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
27859 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27861 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
27862 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
27863 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
27864 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
27865 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
27866 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
27867 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
27868 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27869 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27870 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
27871 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
27872 (__builtin_ia32_mulss_mask_round): New builtins.
27873 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27874 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27875 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
27877 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
27878 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27880 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27883 2017-05-10 Julia Koval <julia.koval@intel.com>
27885 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
27886 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
27887 (_mm256_setr_m128i): New intrinsics.
27889 2017-05-10 Julia Koval <julia.koval@intel.com>
27891 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
27892 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
27893 (_mm_maskz_rcp14_ss): New intrinsics.
27894 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
27895 (__builtin_ia32_rcp14ss_mask): New builtins.
27896 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
27898 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
27900 PR tree-optimization/51513
27901 * tree-cfg.c (gimple_seq_unreachable_p): New function.
27902 (assert_unreachable_fallthru_edge_p): Use it.
27903 (group_case_labels_stmt): Likewise.
27904 * tree-cfg.h: Prototype it.
27905 * stmt.c: Include cfghooks.h and tree-cfg.h.
27906 (emit_case_dispatch_table) <gap_label>: New local variable.
27907 Use it to fill dispatch table gaps.
27908 Test for default_label before updating probabilities.
27909 (expand_case) <default_label>: Remove unneeded initialization.
27910 Test for unreachable default case statement and remove its edge.
27911 Set default_label accordingly.
27912 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
27914 2017-05-10 Carl Love <cel@us.ibm.com>
27916 * config/rs6000/rs6000-c: Add support for built-in functions
27917 vector signed char vec_neg (vector signed char)
27918 vector signed short int vec_neg (vector short int)
27919 vector signed int vec_neg (vector signed int)
27920 vector signed long long vec_neg (vector signed long long)
27921 vector float vec_neg (vector float)
27922 vector double vec_neg (vector double)
27923 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
27925 * config/rs6000/altivec.h: Add define for vec_neg
27926 * doc/extend.texi: Update the built-in documentation for the
27927 new built-in functions.
27929 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27931 PR tree-optimization/77644
27932 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
27934 2017-05-10 Nathan Sidwell <nathan@acm.org>
27936 * dumpfile.h (TDI_lang_all): New.
27937 (TDF_KIND): New. Renumber others
27938 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
27940 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
27942 (get_dump_file_name): Adjust suffix generation.
27943 (dump_enable_all): Use TDF_KIND.
27944 * doc/invoke.texi (-fdump-lang-all): Document.
27946 * dumpfile.h: Tabify.
27948 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
27951 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
27952 Move member access before delete.
27954 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
27956 * tree-inline.c (expand_call_inline): Split block at stmt
27959 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
27962 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
27963 are now unused after splitting mov{sf,sd}_hardfloat.
27964 (f32_lr2): Likewise.
27965 (f32_lm): Likewise.
27966 (f32_lm2): Likewise.
27967 (f32_li): Likewise.
27968 (f32_li2): Likewise.
27969 (f32_lv): Likewise.
27970 (f32_sr): Likewise.
27971 (f32_sr2): Likewise.
27972 (f32_sm): Likewise.
27973 (f32_sm2): Likewise.
27974 (f32_si): Likewise.
27975 (f32_si2): Likewise.
27976 (f32_sv): Likewise.
27977 (f32_dm): Likewise.
27978 (f32_vsx): Likewise.
27979 (f32_av): Likewise.
27980 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
27981 For movsf, order stores so the VSX stores occur before the GPR
27982 store which encourages the register allocator to use a traditional
27983 FPR instead of a GPR. For movsd, order the stores so that the GPR
27984 store comes before the VSX stores to allow the power6 to work.
27985 This is due to the power6 not having a 32-bit integer store
27986 instruction from a FPR.
27987 (movsf_hardfloat): Likewise.
27988 (movsd_hardfloat): Likewise.
27990 2017-05-09 Martin Sebor <msebor@redhat.com>
27992 PR translation/80280
27993 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
27996 PR translation/80280
27997 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
27998 data member added in r247778.
27999 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28001 2017-05-09 Nathan Sidwell <nathan@acm.org>
28003 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28005 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28008 2017-05-09 Marek Polacek <polacek@redhat.com>
28010 * doc/invoke.texi: Fix typo.
28012 2017-05-09 Richard Biener <rguenther@suse.de>
28014 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28015 (vrp_val_is_min): Likewise.
28016 (set_value_range_to_value): Likewise.
28017 (set_value_range_to_nonnegative): Likewise.
28018 (gimple_assign_nonzero_p): Likewise.
28019 (gimple_stmt_nonzero_p): Likewise.
28020 (vrp_int_const_binop): Likewise. Remove unreachable case.
28021 (adjust_range_with_scev): Adjust comments.
28022 (compare_range_with_value): Likewise.
28023 (extract_range_from_phi_node): Likewise.
28024 (test_for_singularity): Likewise.
28026 2017-05-09 Richard Biener <rguenther@suse.de>
28028 * tree-vrp.c (get_single_symbol): Add assert that we don't
28029 get overflowed constants as invariant part.
28030 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28031 checks. Use wi::cmp instead of recursing for integer constants.
28032 (compare_values): Just ignore whether we assumed undefined
28033 overflow instead of failing the compare.
28034 (extract_range_for_var_from_comparison_expr): Add comment before the
28035 TREE_NO_WARNING sets.
28036 (test_for_singularity): Likewise.
28037 (extract_range_from_comparison): Do not disable optimization
28038 when we assumed undefined overflow.
28039 (extract_range_basic): Remove init of unused var.
28041 2017-05-09 Richard Biener <rguenther@suse.de>
28043 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28044 (extract_range_from_multiplicative_op_1): Adjust.
28045 (extract_range_from_binary_expr_1): Use int_const_binop.
28047 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28050 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28051 rs6000_store_data_bypass_p in seven define_bypass directives and
28052 in several comments.
28053 * config/rs6000/rs6000-protos.h: Add prototype for
28054 rs6000_store_data_bypass_p function.
28055 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28056 function implements slightly different (rs6000-specific) semantics
28057 than store_data_bypass_p, returning false rather than aborting
28058 with assertion error when arguments do not satisfy the
28059 requirements of store data bypass.
28060 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28061 rs6000_store_data_bypass_p.
28063 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28065 * config/xtensa/xtensa-protos.h
28066 (xtensa_initial_elimination_offset): New declaration.
28067 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28068 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28069 macro definition, add case for FRAME_POINTER_REGNUM when
28070 FRAME_GROWS_DOWNWARD.
28071 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28072 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28073 xtensa_initial_elimination_offset.
28075 2017-05-08 Nathan Sidwell <nathan@acm.org>
28077 * doc/invoke.texi: Alphabetize -fdump options.
28079 2017-05-08 Martin Sebor <msebor@redhat.com>
28081 PR translation/80280
28082 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28084 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28086 * target.def (compute_frame_layout): New optional target hook.
28087 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28088 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28089 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28091 * reload1.c (verify_initial_elim_offsets): Likewise.
28092 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28093 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28094 (arm_get_frame_offsets): Split up into this ...
28095 (arm_compute_frame_layout): ... and this function.
28097 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28099 * config/aarch64/constraints.md (Usa): New constraint.
28100 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28102 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28104 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28105 with_multilib_list after it has been checked.
28107 2017-05-08 Richard Biener <rguenther@suse.de>
28109 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28110 (bitmap_set_subtract_values): Likewise.
28112 2017-05-08 Richard Biener <rguenther@suse.de>
28114 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28115 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28117 (gimple_stmt_nonzero_warnv_p): Rename to ...
28118 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28120 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28121 (extract_range_basic): Adjust, do not disable propagation on
28122 strict overflow sensitive simplification.
28123 (vrp_visit_cond_stmt): Likewise.
28125 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28127 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28128 body size unconditionally.
28130 2017-05-07 Jeff Law <law@redhat.com>
28133 2017-05-06 Jeff Law <law@redhat.com>
28134 PR tree-optimization/78496
28135 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28138 PR tree-optimization/78496
28139 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28140 (simplify_stmt_using_ranges): Call it.
28141 (vrp_dom_walker::before_dom_children): Extract equivalences
28142 from an ASSERT_EXPR with an equality comparison against a
28145 2017-05-06 Jeff Law <law@redhat.com>
28147 PR tree-optimization/78496
28148 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28151 PR tree-optimization/78496
28152 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28153 (simplify_stmt_using_ranges): Call it.
28154 (vrp_dom_walker::before_dom_children): Extract equivalences
28155 from an ASSERT_EXPR with an equality comparison against a
28158 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28160 * lra-constraints.c (lra_copy_reg_equiv): New function.
28161 (split_reg): Use it to copy equivalence information from the
28162 original register to the spill register.
28164 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28166 PR rtl-optimization/75964
28167 * simplify-rtx.c (simplify_const_relational_operation): Remove
28168 invalid handling of comparisons of integer ABS.
28170 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28172 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28173 initialize to zero.
28174 (init_regs): Remove declaration.
28175 (function_arg_advance_32): Initialize error_p as boolean variable.
28177 2017-05-05 Nathan Sidwell <nathan@acm.org>
28179 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28180 lines. Use for (;;).
28182 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28184 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28185 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28186 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28187 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28188 VF=2 that require versioning.
28190 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28192 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28195 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28197 * diagnostic.h (diagnostic_override_option_index): Convert from
28198 macro to inline function.
28200 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28202 * diagnostic.c (last_module_changed_p): New function.
28203 (set_last_module): New function.
28204 (diagnostic_report_current_module): Convert macro usage to
28205 the above functions.
28206 * diagnostic.h (diagnostic_context::last_module): Strengthen
28207 from const line_map * to const line_map_ordinary *.
28208 (diagnostic_last_module_changed): Delete macro.
28209 (diagnostic_set_last_module): Delete macro.
28211 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28213 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
28214 with diagnostic_report_diagnostic.
28215 (diagnostic_n_impl_richloc): Likewise.
28216 * diagnostic.h (report_diagnostic): Delete macro.
28217 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
28218 with diagnostic_report_diagnostic.
28219 * substring-locations.c (format_warning_va): Likewise.
28221 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28223 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
28224 save/restor of format_spec. Move option-printing code to...
28225 (print_option_information): ...this new function, and
28226 reimplement by simply printing to the pretty_printer,
28227 rather than appending to the format string.
28229 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28231 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
28232 handling logic into...
28233 (update_effective_level_from_pragmas): ...this new function.
28235 2017-05-04 Andrew Waterman <andrew@sifive.com>
28237 * config/riscv/riscv.opt (mstrict-align): New option.
28238 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
28239 (SLOW_UNALIGNED_ACCESS): Define.
28240 (riscv_slow_unaligned_access): Declare.
28241 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
28243 (riscv_slow_unaligned_access): New variable.
28244 (rocket_tune_info): Set slow_unaligned_access to true.
28245 (optimize_size_tune_info): Set slow_unaligned_access to false.
28246 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
28247 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
28248 (riscv_option_override): Set riscv_slow_unaligned_access.
28249 * doc/invoke.texi: Add -mstrict-align to RISC-V.
28251 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
28253 * config/riscv/riscv.md: Unify indentation.
28255 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28260 * config/rs6000/rs6000.md (u code attribute): Add FIX and
28262 (extendsi<mode>2): Add support for doing sign extension via
28263 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
28264 don't have ISA 3.0 instructions.
28265 (extendsi<mode>2 splitter): Likewise.
28266 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
28267 generate the normal insns since SImode can now go in vector
28268 registers. Disallow the special UNSPECs needed for previous
28269 machines to hide SImode being used. Add new insns
28270 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
28271 (fix_trunc<mode>si2_stfiwx): Likewise.
28272 (fix_trunc<mode>si2_internal): Likewise.
28273 (fixuns_trunc<mode>si2): Likewise.
28274 (fixuns_trunc<mode>si2_stfiwx): Likewise.
28275 (fctiw<u>z_<mode>_smallint): Likewise.
28276 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
28277 of floating point to 32-bit integer from doing a direct move to
28278 the GPR registers to do a store.
28279 (fctiwz_<mode>): Break long line.
28281 2017-05-05 Bin Cheng <bin.cheng@arm.com>
28283 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
28284 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
28285 (addr_list, addr_offset_valid_p): New.
28286 (split_address_groups): Check offset validity with above function.
28287 (gt-tree-ssa-loop-ivopts.h): Include header file.
28289 2017-05-05 Nathan Sidwell <nathan@acm.org>
28291 * config.gcc (arm*-*-*): Add missing 'fi'.
28293 2017-05-05 Steve Ellcey <sellcey@cavium.com>
28295 * doc/invoke.texi (-fopt-info): Explicitly say order of options
28296 included in -fopt-info does not matter.
28297 * doc/optinfo.texi (-fopt-info): Fix description of default
28298 behavour. Explicitly say order of options included in -fopt-info
28301 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28303 * config.gcc: Allow combinations of aprofile and rmprofile values for
28304 --with-multilib-list.
28305 * config/arm/t-multilib: New file.
28306 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
28307 variables. Remove setting of ISA and floating-point ABI in
28308 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
28309 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
28310 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
28311 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
28313 * config/arm/t-rmprofile: Likewise except for the matches changes.
28314 * doc/install.texi (--with-multilib-list): Document the combination of
28315 aprofile and rmprofile values and warn about pitfalls in doing that.
28317 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28319 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
28320 (movdi_aarch64): Likewise.
28322 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28324 PR tree-optimization/80632
28325 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
28327 (build_arrays): Initialize it for virtual phis.
28328 (fix_phi_nodes): Use it for virtual phis.
28330 PR tree-optimization/80558
28331 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
28332 [x, y] op z into [x op, y op z] for op & or | if conditions
28335 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
28336 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28339 * config/arm/arm.md (use_literal_pool): Remove.
28340 (64-bit immediate split): No longer takes cost into consideration
28341 if arm_disable_literal_pool is enabled.
28342 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
28343 used when arm_disable_literal_pool is enabled.
28344 (arm_max_const_double_inline_cost): Remove use of
28345 arm_disable_literal_pool.
28346 (push_minipool_fix): Add assert.
28347 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
28348 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
28349 (no_literal_pool_sf_immediate): New.
28351 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28353 PR tree-optimization/80613
28354 * tree-ssa-dce.c (propagate_necessity): Remove cases for
28355 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
28357 2017-05-05 Richard Biener <rguenther@suse.de>
28359 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
28361 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
28363 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
28364 of this flag from insn conditions due to removal from r247495.
28366 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28368 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
28370 (arm_early_store_addr_dep_ptr): Likewise.
28371 * config/arm/aarch-common-protos.h
28372 (arm_early_load_addr_dep_ptr): Add prototype.
28373 (arm_early_store_addr_dep_ptr): Likewise.
28374 * config/arm/cortex-a53.md: Add new bypasses.
28376 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28378 * tree.c (next_type_uid): Change type to unsigned.
28379 (type_hash_canon): Decrement back next_type_uid if
28380 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
28381 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
28384 2017-05-04 Martin Sebor <msebor@redhat.com>
28386 * builtins.c: Fix a trivial typo in a comment.
28388 PR middle-end/79234
28389 * builtins.c (check_sizes): Adjust to handle reading past the end.
28390 Avoid printing excessive upper bound of ranges. Use %E to print
28391 tree nodes instead of converting them to %wu.
28392 (expand_builtin_memchr): New function.
28393 (compute_dest_size): Rename...
28394 (compute_objsize): ...to this.
28395 (expand_builtin_memcpy): Adjust.
28396 (expand_builtin_mempcpy): Adjust.
28397 (expand_builtin_strcat): Adjust.
28398 (expand_builtin_strcpy): Adjust.
28399 (check_strncat_sizes): Adjust.
28400 (expand_builtin_strncat): Adjust.
28401 (expand_builtin_strncpy): Adjust and simplify.
28402 (expand_builtin_memset): Adjust.
28403 (expand_builtin_bzero): Adjust.
28404 (expand_builtin_memcmp): Adjust.
28405 (expand_builtin): Handle memcmp.
28406 (maybe_emit_chk_warning): Check strncat just once.
28408 2017-05-04 Martin Sebor <msebor@redhat.com>
28410 PR preprocessor/79214
28411 PR middle-end/79222
28412 PR middle-end/79223
28413 * builtins.c (check_sizes): Add inlining context and issue
28414 warnings even when -Wno-system-headers is set.
28415 (check_strncat_sizes): Same.
28416 (expand_builtin_strncat): Same.
28417 (expand_builtin_memmove): New function.
28418 (expand_builtin_stpncpy): Same.
28419 (expand_builtin): Handle memmove and stpncpy.
28421 2017-05-04 Bin Cheng <bin.cheng@arm.com>
28423 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
28424 which is not used any more.
28426 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28428 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
28430 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28432 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
28433 (cortexa53_tunings): Likewise.
28434 (cortexa57_tunings): Likewise.
28435 (cortexa72_tunings): Likewise.
28436 (cortexa73_tunings): Likewise.
28438 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28440 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
28441 Set loop alignment to 8.
28443 2017-05-04 Martin Sebor <msebor@redhat.com>
28445 PR translation/80280
28446 * builtins.c (expand_builtin_object_size): Add missing quoting to
28447 %D and like directives.
28448 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
28449 (hsa_type_for_tree_type): Same.
28450 (verify_function_arguments): Same.
28451 * symtab.c (symbol_table::change_decl_assembler_name): Same.
28452 * varasm.c (get_section): Same.
28455 2017-05-04 Martin Sebor <msebor@redhat.com>
28457 PR translation/80280
28458 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
28460 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28462 * config/aarch64/aarch64.c (generic_addrcost_table):
28463 Change HI/TI mode setting.
28465 2017-05-04 Martin Jambor <mjambor@suse.cz>
28467 PR tree-optimization/80622
28468 * tree-sra.c (comes_initialized_p): New function.
28469 (build_accesses_from_assign): Only set write lazily when
28470 comes_initialized_p is false.
28471 (analyze_access_subtree): Use comes_initialized_p.
28472 (propagate_subaccesses_across_link): Assert !comes_initialized_p
28473 instead of testing for PARM_DECL.
28475 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28477 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
28478 constraint on operand 0 to allow more general addressing modes.
28479 Adjust output template.
28480 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
28482 * config/aarch64/aarch64-protos.h
28483 (aarch64_address_valid_for_prefetch_p): Declare prototype.
28484 * config/aarch64/constraints.md (Dp): New address constraint.
28485 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
28488 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
28490 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
28491 update use of estimate_ipcp_clone_size_and_time.
28492 (estimate_local_effects): Update use of
28493 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
28494 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
28495 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
28496 Return nonspecialized time.
28498 2017-05-04 Richard Biener <rguenther@suse.de>
28500 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
28501 for the last VUSE which def dominates the PHI. Directly call
28503 (get_continuation_for_phi_1): Remove.
28505 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
28507 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
28508 to explain the use of truncating division. Cap the number of
28509 iterations to the maximum given by nb_iterations_upper_bound,
28512 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28514 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
28515 * configure: Regenerate.
28516 * config.in: Regenerate.
28517 * config/i386/driver-mingw32.c: new file.
28518 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
28519 * config.host: Link driver-mingw32.o on MinGW host.
28520 * doc/install.texi: Document new --enable-mingw-wildcard configure
28523 2017-05-04 Marek Polacek <polacek@redhat.com>
28525 PR tree-optimization/80612
28526 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
28528 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28529 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
28531 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
28532 (movt splitter): Likewise.
28533 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
28534 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
28535 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
28536 block for Thumb-1 with MOVT.
28537 (thumb2_legitimate_address_p): Move code block ...
28538 (can_avoid_literal_pool_for_label_p): ... into this new function.
28539 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
28541 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
28542 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
28543 "M-profile targets with the MOVT instruction".
28545 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28547 * config/arm/arm-builtins.c (arm_init_builtins): Rename
28548 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
28549 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
28551 2017-05-04 Martin Liska <mliska@suse.cz>
28553 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
28554 variable cond_code.
28556 2017-05-04 Richard Biener <rguenther@suse.de>
28558 * tree.c (array_at_struct_end_p): Handle arrays at struct
28559 end with flexarrays more conservatively. Refactor and treat
28560 arrays of arrays or aggregates more strict. Fix
28561 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
28562 * tree.c (array_at_struct_end_p): Adjust prototype.
28563 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
28564 * gimple-fold.c (get_range_strlen): Likewise.
28565 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
28567 2017-05-04 Richard Biener <rguenther@suse.de>
28569 PR tree-optimization/31130
28570 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
28572 (supports_overflow_infinity): Likewise.
28573 (is_negative_overflow_infinity): Likewise.
28574 (is_positive_overflow_infinity): Likewise.
28575 (is_overflow_infinity): Likewise.
28576 (stmt_overflow_infinity): Likewise.
28577 (overflow_infinity_range_p): Likewise.
28578 (usable_range_p): Remove as always returning true.
28579 (make_overflow_infinity): Remove.
28580 (negative_overflow_infinity): Likewise.
28581 (positive_overflow_infinity): Likewise.
28582 (avoid_overflow_infinity): Likewise.
28583 (set_value_range): Adjust accordingly.
28584 (set_value_range_to_nonnegative): Likewise, remove now unused
28585 overflow_infinity arg.
28586 (vrp_operand_equal_p): Adjust.
28587 (update_value_range): Likewise.
28588 (range_int_cst_singleton_p): Likewise.
28589 (operand_less_p): Likewise.
28590 (compare_values_warnv): Likewise.
28591 (extract_range_for_var_from_comparison_expr): Likewise.
28592 (vrp_int_const_binop): Likewise.
28593 (zero_nonzero_bits_from_vr): Likewise.
28594 (extract_range_from_multiplicative_op_1): Likewise.
28595 (extract_range_from_binary_expr_1): Likewise.
28596 (extract_range_from_unary_expr): Likewise.
28597 (extract_range_from_comparison): Likewise.
28598 (extract_range_basic): Likewise.
28599 (adjust_range_with_scev): Likewise.
28600 (compare_ranges): Likewise.
28601 (compare_range_with_value): Likewise.
28602 (dump_value_range): Likewise.
28603 (test_for_singularity): Likewise, remove strict_overflow_p parameter
28605 (simplify_cond_using_ranges): Adjust.
28607 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28609 * brig-builtins.def: Added a builtin for class_f64.
28610 * builtin-types.def: Added a builtin type needed by class_f64.
28612 2017-05-03 Jason Merrill <jason@redhat.com>
28614 * timevar.def: Add TV_CONSTEXPR.
28616 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28618 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
28620 2017-05-03 Martin Jambor <mjambor@suse.cz>
28622 * ipa-prop.c (ipa_update_after_lto_read): Removed.
28623 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
28624 * ipa-cp.c (ipcp_propagate_stage): Do not call
28625 ipa_update_after_lto_read.
28626 * ipa-inline.c (ipa_inline): Likewise.
28628 2017-05-03 Martin Jambor <mjambor@suse.cz>
28630 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
28631 tag. Added a default constructor and a destructor.
28632 (ipa_edge_args_sum_t): New class;
28633 (ipa_edge_args_sum): Declare.
28634 (ipa_edge_args_vector): Remove declaration.
28635 (IPA_EDGE_REF): Use ipa_edge_args_sum.
28636 (ipa_free_edge_args_substructures): Remove declaration.
28637 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
28638 (ipa_edge_args_info_available_for_edge_p): Likewise.
28639 * ipa-prop.c (ipa_edge_args_vector): Removed.
28640 (edge_removal_hook_holder): Likewise.
28641 (edge_duplication_hook_holder): Likewise.
28642 (ipa_edge_args_sum): New variable.
28643 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
28644 ipa_edge_args_vector.
28645 (ipa_free_edge_args_substructures): Likewise.
28646 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
28647 ipa_edge_args_vector.
28648 (ipa_edge_removal_hook): Turned into method
28649 ipa_edge_args_sum_t::remove.
28650 (ipa_edge_duplication_hook): Turned into method
28651 ipa_edge_args_sum_t::duplicate.
28652 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
28653 registering edge hooks.
28654 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
28655 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
28656 ipa_edge_args_sum instead of ipa_edge_args_vector.
28657 * ipa-profile.c (ipa_profile): Likewise.
28659 2017-05-03 Martin Jambor <mjambor@suse.cz>
28661 * symbol-summary.h (function_summary): New method exists.
28662 (function_summary::symtab_removal): Deallocate through release.
28663 (call_summary): New class.
28664 (gt_ggc_mx): New overload.
28665 (gt_pch_nx): Likewise.
28666 (gt_pch_nx): Likewise.
28668 2017-05-03 Jeff Law <law@redhat.com>
28670 PR tree-optimization/78496
28671 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
28672 from simplify_cond_using_ranges. Split off code to walk
28673 backwards through casts into ...
28674 (simplify_cond_using_ranges_2): New function.
28675 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
28676 (execute_vrp): After identifying jump threads, call
28677 simplify_cond_using_ranges_2.
28679 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
28682 * ipa-inline.h (inline_summary): Add ctor.
28683 (create_ggc): Do not use ggc_cleared_alloc.
28685 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
28686 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28688 * gcc.c (handle_braces): Support escaping in switch matching
28690 * doc/invoke.texi (Spec Files): Document it.
28691 Remove superfluous @code markup in items.
28693 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28695 * diagnostic-show-locus.c (struct column_range): New struct.
28696 (get_affected_columns): New function.
28697 (get_printed_columns): New function.
28698 (struct correction): New struct.
28699 (correction::ensure_capacity): New function.
28700 (correction::ensure_terminated): New function.
28701 (struct line_corrections): New struct.
28702 (line_corrections::~line_corrections): New dtor.
28703 (line_corrections::add_hint): New function.
28704 (layout::print_trailing_fixits): Reimplement in terms of the new
28706 (selftest::test_overlapped_fixit_printing): New function.
28707 (selftest::diagnostic_show_locus_c_tests): Call it.
28709 2017-05-03 Nathan Sidwell <nathan@acm.org>
28711 Canonicalize canonical type hashing
28712 * tree.h (type_hash_canon_hash): Declare.
28713 * tree.c (type_hash_list, attribute_hash_list): Move into
28714 type_hash_canon_hash.
28715 (build_type_attribute_qual_variant): Break out hash code calc into
28716 type_hash_canon_hash.
28717 (type_hash_canon_hash): New. Generic type hash computation.
28718 (build_range_type_1, build_array_type_1, build_function_type,
28719 build_method_type_directly, build_offset_type, build_complex_type,
28720 make_vector_type): Call it.
28722 2017-05-03 Richard Biener <rguenther@suse.de>
28724 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
28725 When all DRs have unknown misaligned do not always peel
28726 when there is a store but apply the same costing model as if
28727 there were only loads.
28729 2017-05-03 Richard Biener <rguenther@suse.de>
28732 PR tree-optimization/80492
28733 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28734 compare_base_decls returning dont-know properly.
28736 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
28738 * config/arm/iterators.md (CCSI): New mode iterator.
28739 (arch): New mode attribute.
28740 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
28741 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
28742 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
28743 code iterator for success result mode.
28744 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
28745 the corresponding new insn generators.
28747 2017-05-03 Bin Cheng <bin.cheng@arm.com>
28750 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28751 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28753 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
28755 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
28756 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
28757 (DDR_A): Wrap DDR argument in brackets.
28758 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
28759 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
28760 (DDR_REVERSED_P): Likewise.
28762 2017-05-03 Jakub Jelinek <jakub@redhat.com>
28764 PR tree-optimization/79472
28765 * tree-switch-conversion.c (struct switch_conv_info): Add
28766 contiguous_range and default_case_nonstandard fields.
28767 (collect_switch_conv_info): Compute contiguous_range and
28768 default_case_nonstandard fields, don't clear final_bb if
28769 contiguous_range and only the default case doesn't have the required
28771 (check_all_empty_except_final): Set default_case_nonstandard instead
28772 of failing if contiguous_range and the default case doesn't have empty
28774 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
28775 and only the default case doesn't have the required constants. Skip
28777 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
28778 if default_case_nonstandard.
28779 (build_constructors): Build constant 1 just once. Assert that default
28780 values aren't inserted in between cases if contiguous_range. Skip
28782 (build_arrays): Skip virtual phis.
28783 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
28784 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
28785 Handle virtual phis.
28786 (gen_inbound_check): Handle default_case_nonstandard case.
28787 (process_switch): Adjust check_final_bb caller. Call
28788 gather_default_values with the first non-default case instead of
28789 default case if default_case_nonstandard.
28791 2017-05-02 Nathan Sidwell <nathan@acm.org>
28793 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
28794 check. Fix formatting.
28796 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
28798 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
28799 errors when comparing specialized and unspecialized times.
28801 2017-05-02 David Malcolm <dmalcolm@redhat.com>
28803 * diagnostic-show-locus.c
28804 (layout::should_print_annotation_line_p): Make private.
28805 (layout::print_annotation_line): Make private.
28806 (layout::annotation_line_showed_range_p): Make private.
28807 (layout::show_ruler): Make private.
28808 (layout::print_source_line): Make private. Pass in line and
28809 line_width, rather than calling location_get_source_line. Drop
28811 (layout::print_leading_fixits): New method.
28812 (layout::print_any_fixits): Rename to...
28813 (layout::print_trailing_fixits): ...this, and make private.
28814 Don't print newline fixits.
28815 (diagnostic_show_locus): Move logic for printing one row into...
28816 (layout::print_line): ...this new function. Move the
28817 location_get_source_line call and error-handling from
28818 print_source_line to here. Call print_leading_fixits, and rename
28819 print_any_fixits to print_trailing_fixits.
28820 (selftest::test_fixit_insert_containing_newline): Update now that
28821 newlines are partially supported.
28822 (selftest::test_fixit_insert_containing_newline_2): New test.
28823 (selftest::test_fixit_replace_containing_newline): Update comments.
28824 (selftest::diagnostic_show_locus_c_tests): Call the new test.
28825 * edit-context.c (class added_line): New class.
28826 (class edited_line): Describe newline handling in comment.
28827 (edited_line::actually_edited_p): New method.
28828 (edited_line::print_content): Delete redundant decl.
28829 (edited_line::m_predecessors): New field.
28830 (edited_file::print_content): Call edited_line::print_content.
28831 (edited_file::print_diff): Update to support newlines.
28832 (edited_file::print_diff_hunk): Likewise.
28833 (edited_file::print_run_of_changed_lines): New function.
28834 (edited_file::print_diff_line): Convert to...
28835 (print_diff_line): ...this.
28836 (edited_file::get_effective_line_count): New function.
28837 (edited_line::edited_line): Initialize new field m_predecessors.
28838 (edited_line::~edited_line): Clean up m_predecessors.
28839 (edited_line::apply_fixit): Handle newlines.
28840 (edited_line::get_effective_line_count): New function.
28841 (edited_line::print_content): New function.
28842 (edited_line::print_diff_lines): New function.
28843 (selftest::test_applying_fixits_insert_containing_newline): New
28845 (selftest::test_applying_fixits_replace_containing_newline): New
28847 (selftest::insert_line): New function.
28848 (selftest::test_applying_fixits_multiple_lines): Add example of
28850 (selftest::edit_context_c_tests): Call the new tests.
28852 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28854 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
28855 parameter cand. Update dump information.
28856 (get_computation_cost): Update uses.
28858 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28860 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
28861 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
28862 (get_computation_at, rewrite_use_address): Update use of
28863 get_computation_aff.
28865 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28867 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
28868 (get_computation): Delete.
28869 (get_computation_cost): Implement like get_computation_cost_at.
28870 Use get_computation_at.
28871 (get_computation_cost_at): Delete.
28872 (rewrite_use_nonlinear_expr): Use get_computation_at.
28873 (rewrite_use_compare, remove_unused_ivs): Ditto.
28875 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28877 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
28879 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28881 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
28882 (ivopts_global_cost_for_size): Rename parameter and update uses.
28883 (iv_ca_recount_cost): Update uses.
28884 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
28885 candidates seperately in n_invs and n_cands.
28886 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
28888 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28890 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
28891 (find_inv_vars_cb): New.
28892 (find_depends): Renamed to ...
28893 (find_inv_vars): ... this.
28894 (add_candidate_1, force_var_cost): Call find_inv_vars.
28895 (split_address_cost, determine_group_iv_cost_cond): Ditto.
28897 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28899 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
28900 inv_vars. Add inv_exprs.
28901 (struct iv_cand): Rename depends_on to inv_vars.
28902 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
28903 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
28904 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
28905 (dump_cand): Dump inv_vars.
28906 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
28907 (record_invariant, find_depends, add_candidate_1): Ditto.
28908 (set_group_iv_cost, force_var_cost): Ditto.
28909 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
28910 (get_computation_cost_at, get_computation_cost): Ditto.
28911 (determine_group_iv_cost_generic): Ditto.
28912 (determine_group_iv_cost_address): Ditto.
28913 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
28914 (determine_group_iv_costs): Ditto.
28915 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
28916 (iv_ca_set_remove_invariants): Renamed to ...
28917 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
28918 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
28919 (iv_ca_set_add_invariants): Renamed to ...
28920 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
28921 (iv_ca_set_cp): Use iv_ca_set_add_invs.
28922 (iv_ca_has_deps): Support inv_vars and inv_exprs.
28923 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
28924 (create_new_ivs): Remove useless dump.
28926 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28928 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
28930 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
28931 (iv_ca_set_no_cp, create_new_iv): Ditto.
28933 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28935 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28937 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28939 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
28940 function tree_check2.
28942 2017-05-02 Martin Liska <mliska@suse.cz>
28944 * doc/gcov.texi: Add missing preposition.
28945 * gcov.c (function_info::function_info): Properly fill up
28946 all member variables.
28948 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28950 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
28952 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28954 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
28956 2017-05-02 Martin Liska <mliska@suse.cz>
28959 * lto-streamer-in.c (lto_read_tree_1): Remove
28960 LTO_STREAMER_DEBUG.
28961 * lto-streamer.c (struct tree_hash_entry): Likewise.
28962 (struct tree_entry_hasher): Likewise.
28963 (tree_entry_hasher::hash): Likewise.
28964 (tree_entry_hasher::equal): Likewise.
28965 (lto_streamer_init): Likewise.
28966 (lto_orig_address_map): Likewise.
28967 (lto_orig_address_get): Likewise.
28968 (lto_orig_address_remove): Likewise.
28969 * lto-streamer.h: Likewise.
28970 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
28971 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
28973 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
28975 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
28976 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
28977 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
28978 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
28979 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
28980 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
28981 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
28982 (mm_maskz_sub_ss): New intrinsics.
28983 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28984 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28985 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
28986 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
28987 (__builtin_ia32_subss_mask_round): New builtins.
28988 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28989 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28990 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
28992 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
28993 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28995 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28998 2017-05-02 Martin Jambor <mjambor@suse.cz>
29000 PR tree-optimization/78687
29001 * tree-sra.c (access): New field parent.
29002 (process_subtree_disqualification): New function.
29003 (disqualify_candidate): Call it.
29004 (build_accesses_from_assign): Reset write flag if creating an
29006 (build_access_subtree): Fill in parent field and also prpagate
29007 down grp_write flag.
29008 (create_artificial_child_access): New parameter set_grp_write, set
29009 grp_write to its value.
29010 (propagate_subaccesses_across_link): Also propagate grp_write flag
29012 (propagate_all_subaccesses): Push the closest parent back to work
29013 queue if add_access_to_work_queue returned true.
29015 2017-05-02 Richard Biener <rguenther@suse.de>
29017 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29018 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29019 -fstrict-overflow documentation.
29020 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29021 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29022 flag_strict_overflow.
29023 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29024 * lto-opts.c (lto_write_options): Do not stream it.
29025 * lto-wrapper.c (merge_and_complain): Do not handle it.
29026 * opts.c (default_options_table): Do not set -fstrict-overflow.
29027 (finish_options): Likewise do not clear it when sanitizing.
29028 * simplify-rtx.c (simplify_const_relational_operation): Do not
29029 test flag_strict_overflow.
29031 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29033 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29034 using enabled attribute.
29035 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29036 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29037 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29038 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29039 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29040 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29041 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29042 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29043 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29044 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29046 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29048 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29050 2017-05-02 Richard Biener <rguenther@suse.de>
29052 PR tree-optimization/80591
29054 2017-04-10 Richard Biener <rguenther@suse.de>
29056 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29059 2017-05-02 Richard Biener <rguenther@suse.de>
29061 PR tree-optimization/80549
29062 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29063 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29064 headers if they do not have a preheader.
29066 2017-05-02 Martin Liska <mliska@suse.cz>
29069 * common.opt: Fix typo.
29070 * doc/invoke.texi: Likewise.
29072 2017-05-01 Jan Beulich <jbeulich@suse.com>
29074 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29075 swapping, add (x,x,m,x,n) alternative.
29077 2017-05-01 Nathan Sidwell <nathan@acm.org>
29079 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29080 unnecessary unadjusted_alignment check.
29082 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29085 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29087 * gimplify.c (gimplify_cilk_detach): New function.
29088 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29089 * tree-core.h: Document EXPR_CILK_SPAWN.
29090 * tree.h (EXPR_CILK_SPAWN): Define.
29092 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29094 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29095 to use new fixit_hint representation, using the "replace" logic.
29096 (get_line_span_for_fixit_hint): Likewise.
29097 (layout::print_any_fixits): Likewise.
29098 (selftest::test_one_liner_many_fixits): Rename to...
29099 (selftest::test_one_liner_many_fixits_1): ...this, and update
29100 comment and expected output to reflect that the multiple fix-it
29101 hints are now consolidated into one insertion.
29102 (selftest::test_one_liner_many_fixits_2): New test.
29103 (selftest::test_diagnostic_show_locus_one_liner): Update for
29105 (selftest::test_fixit_consolidation): Update for fix-it API
29107 * diagnostic.c (print_parseable_fixits): Likewise.
29108 * edit-context.c (edited_line::m_line_events): Convert from
29109 auto_vec <line_event *> to auto_vec <line_event>.
29110 (class line_event): Convert from abstract base class to a concrete
29111 class, taking over the role of replace_event.
29112 (class insert_event): Delete.
29113 (class replace_event): Rename to class line_event. Convert to
29115 (edit_context::add_fixits): Reimplement.
29116 (edit_context::apply_insert): Delete.
29117 (edit_context::apply_replace): Rename to...
29118 (edit_context::apply_fixit): ...this. Convert to half-open range.
29119 (edited_file::apply_insert): Delete.
29120 (edited_file::apply_replace): Rename to...
29121 (edited_file::apply_fixit): ...this.
29122 (edited_line::~edited_line): Drop deletion of events.
29123 (edited_line::apply_insert): Delete.
29124 (edited_line::apply_replace): Rename to...
29125 (edited_line::apply_fixit): ...this. Convert to half-open range.
29126 Update for change to type of m_line_events.
29127 * edit-context.h (edit_context::apply_insert): Delete.
29128 (edit_context::apply_replace): Rename to...
29129 (edit_context::apply_fixit): ...this.
29131 2017-05-01 Martin Sebor <msebor@redhat.com>
29133 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29136 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29139 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29140 __get_cpuid_max returns 0.
29141 (__get_cpuid_count): Ditto.
29143 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29145 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29146 replacement expression is another instance of one of its arguments.
29148 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29151 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29152 check for stack push/pop autoinc.
29153 * config/i386/i386.c (ix86_agi_dependent): Return false
29154 if the only reason why modified_in_p returned true is that
29155 addr is SP based and set_insn is a push or pop.
29157 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29159 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29162 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29165 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29166 (account_size_time): Use two predicates - exec_pred and
29168 (evaluate_conditions_for_known_args): Compute both clause and
29170 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29171 (inline_summary_t::duplicate): Update.
29172 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29174 (compute_inline_parameters): Likewise.
29175 (estimate_edge_size_and_time): Update caluclation of time.
29176 (estimate_node_size_and_time): Compute both time and nonspecialized
29178 (estimate_ipcp_clone_size_and_time): Update.
29179 (inline_merge_summary): Update.
29180 (do_estimate_edge_time): Update.
29181 (do_estimate_edge_size): Update.
29182 (do_estimate_edge_hints): Update.
29183 (inline_read_section, inline_write_summary): Stream both new predicates.
29184 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29186 (compute_inlined_call_time): Cleanup.
29187 (big_speedup_p): Update.
29188 (edge_badness): Update.
29189 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29190 (size_time_entry): Replace predicate by exec_predicate and
29191 nonconst_predicate.
29192 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29193 (estimate_edge_time): Return also nonspec_time.
29194 (reset_edge_growth_cache): Update.
29196 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29198 PR rtl-optimization/80491
29199 * ifcvt.c (noce_process_if_block): When looking for x setter
29200 with missing else_bb, don't check only the insn right before
29201 cond_earliest, but look for the last insn that x is modified in
29202 within the same bb.
29204 PR rtl-optimization/80491
29205 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29207 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29209 PR tree-optimization/80487
29210 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
29212 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29214 PR tree-optimization/79697
29215 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
29216 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
29217 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
29219 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
29220 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
29222 2017-04-28 Martin Sebor <msebor@redhat.com>
29224 PR tree-optimization/80523
29225 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
29226 (init_target_to_host_charmap, target_to_host, target_strtol10): New
29228 (maybe_warn, format_directive, parse_directive): Use new functions.
29229 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
29231 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
29233 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
29235 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
29237 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
29238 target_header_dir): Set correctly.
29239 * configure: Regenerated.
29240 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29241 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29242 instead of SYSTEM_HEADER_DIR.
29244 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
29246 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
29247 (estimate_local_effects): Likewise.
29248 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
29249 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
29250 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
29251 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
29252 do_estimate_edge_time, estimate_edge_time): Likewise.
29253 * ipa-inline-analysis.c (estimate_node_size_and_time,
29254 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
29255 (estimate_time_after_inlining): Remove.
29257 2017-04-28 Martin Liska <mliska@suse.cz>
29259 * doc/gcov.texi: Enhance documentation of gcov.
29261 2017-04-28 Martin Liska <mliska@suse.cz>
29263 * doc/gcov.texi: Sort options in alphabetic order.
29264 * doc/gcov-dump.texi: Likewise.
29265 * doc/gcov-tool.texi: Likewise.
29266 * gcov.c (print_usage): Likewise.
29267 * gcov-dump.c (print_usage): Likewise.
29268 * gcov-tool.c (print_merge_usage_message): Likewise.
29269 (print_rewrite_usage_message): Likewise.
29270 (print_overlap_usage_message): Likewise.
29272 2017-04-28 Martin Liska <mliska@suse.cz>
29274 PR gcov-profile/53915
29275 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
29277 2017-04-28 Martin Liska <mliska@suse.cz>
29279 PR gcov-profile/79891
29280 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
29281 is marked by compiler as living on a line.
29282 (get_cycles_count): Remove usage of the union.
29283 (output_intermediate_file): Likewise.
29284 (find_source): Fix GNU coding style.
29285 (accumulate_line_counts): Remove old non-all block mode.
29286 (output_lines): Remove usage of the union.
29287 * profile.c (output_location): Include all BBs, even if
29288 belonging to a same line (and file) as a previous BB.
29290 2017-04-28 Martin Liska <mliska@suse.cz>
29292 * gcov.c (process_args): Handle new argument 'w'.
29293 (read_graph_file): Assign ID to BBs.
29294 (output_branch_count): Display BB # if verbose flag is set.
29295 (output_lines): Likewise for arcs.
29296 (print_usage): Add '--verbose' option help.
29297 * doc/gcov.texi: Document --verbose (-w) option.
29299 2017-04-28 Martin Liska <mliska@suse.cz>
29301 * gcov.c (struct block_location_info): New struct.
29302 (process_file): Fill up the new structure.
29303 (read_graph_file): Replace usage of encoding by the newly added
29305 (add_line_counts): Likewise.
29306 (accumulate_line_counts): Remove usage of the union.
29307 (function_info::function_info): New function.
29308 (function_info::~function_info): Likewise.
29309 (process_file): Call delete instead of release_function.
29310 (release_function): Release the function.
29311 (release_structures): Call delete instead of release_function.
29312 (solve_flow_graph): Replace usage of num_blocks.
29313 (find_exception_blocks): Likewise.
29314 (output_lines): Fix GNU coding style.
29316 2017-04-28 Martin Liska <mliska@suse.cz>
29319 * coverage.c (coverage_remove_note_file): New function.
29320 * coverage.h: Declare the function.
29321 * toplev.c (finalize): Clean if an error has been seen.
29323 2017-04-28 Martin Liska <mliska@suse.cz>
29325 PR gcov-profile/80031
29326 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
29327 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
29328 * gcov.c (read_graph_file): Read just number of blocks.
29329 * profile.c (branch_prob): Do not stream 0 flags per a basic
29332 2017-04-28 Martin Liska <mliska@suse.cz>
29334 * gcov-dump.c (tag_*): Add new argument to declarations.
29335 (dump_gcov_file): Likewise.
29336 (tag_blocks): Add and use new argument depth.
29337 (tag_arcs): Likewise.
29338 (tag_lines): Likewise.
29339 (tag_counters): Likewise.
29340 (tag_summary): Likewise.
29341 (dump_working_sets): Use depth to do a proper indentation.
29343 2017-04-28 Jakub Jelinek <jakub@redhat.com>
29346 * cgraph.h (symtab_node::debug_symtab): No longer inline.
29347 * symtab.c (symtab_node::debug_symtab): Move definition here.
29349 2017-04-28 Richard Biener <rguenther@suse.de>
29351 * lto-streamer.h (LTO_major_version): Bump to 7.
29353 2017-04-28 Richard Biener <rguenther@suse.de>
29355 * tree-vrp.c (assert_info): New struct.
29356 (add_assert_info): New helper.
29357 (register_edge_assert_for_2): Refactor to add asserts to a vector
29359 (register_edge_assert_for_1): Likewise.
29360 (register_edge_assert_for): Likewise.
29361 (finish_register_edge_assert_for): New helper actually registering
29362 asserts where live on edge.
29363 (find_conditional_asserts): Adjust.
29364 (find_switch_asserts): Likewise.
29365 (evrp_dom_walker::try_find_new_range): Generalize.
29366 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
29368 2017-04-27 Marek Polacek <polacek@redhat.com>
29371 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
29372 arg10 and arg11 to itype.
29374 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
29376 * doc/extend.texi (Object Size Checking): Improve grammar.
29378 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
29381 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
29382 that the logic for permitting reciprocal estimates matches that
29385 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29388 * tree.c (type_cache_hasher::equal): Only compare
29389 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
29390 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
29391 non-aggregate element types.
29392 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
29393 about the flag on ARRAY_TYPEs in the comment, formatting fix.
29395 2017-04-27 Richard Biener <rguenther@suse.de>
29397 PR middle-end/80533
29398 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
29399 stripping ARRAY_REFs from MEM_EXPR make sure we're not
29400 keeping a reference to a trailing array.
29402 2017-04-27 Richard Biener <rguenther@suse.de>
29404 PR middle-end/80539
29405 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
29406 being in loop-closed SSA form conservatively.
29407 (chrec_fold_multiply_poly_poly): Likewise.
29409 2017-04-27 Tamar Christina <tamar.christina@arm.com>
29411 PR middle-end/79665
29412 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
29413 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
29415 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29418 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
29419 (aarch64_function_arg_alignment): Return unsigned int again, but still
29420 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
29421 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29422 Don't emit -Wpsabi note.
29423 (aarch64_function_arg_boundary): Likewise.
29424 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29427 2017-04-26 Nathan Sidwell <nathan@acm.org>
29429 * tree.h (crc32_unsigned_n): Declare.
29430 (crc32_unsigned, crc32_unsigned): Make inline.
29431 * tree.c (crc32_unsigned_bits): Replace with ...
29432 (crc32_unsigned_n): ... this.
29433 (crc32_unsigned, crc32_byte): Remove.
29434 (crc32_string): Remove unnecessary braces.
29436 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29438 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
29439 * ipa-inline-analysis.c (MAX_TIME): Remove.
29440 (account_size_time): Use sreal for time.
29441 (dump_inline_summary): Update.
29442 (estimate_function_body_sizes): Update.
29443 (estimate_edge_size_and_time): Update.
29444 (estimate_calls_size_and_time): Update.
29445 (estimate_node_size_and_time): Update.
29446 (inline_merge_summary): Update.
29447 (inline_update_overall_summary): Update.
29448 (estimate_time_after_inlining): Update.
29449 (inline_read_section): Update.
29450 (inline_write_summary): Update.
29451 * ipa-inline.c (compute_uninlined_call_time): Update.
29452 (compute_inlined_call_time): Update.
29453 (recursive_inlining): Update.
29454 (inline_small_functions): Update.
29455 (dump_overall_stats): Update.
29456 * ipa-inline.h: Include sreal.h.
29457 (size_time_entry): Turn time to sreal.
29458 (inline_summary): Turn self_time nad time to sreal.
29460 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29462 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
29464 (sreal::stream_out, sreal::stream_in): New.
29465 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
29467 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29469 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
29472 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
29475 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
29477 (dimode_scalar_chain::compute_convert_gain): Ditto.
29478 (dimode_scalar_chain::make_vector_copies): Ditto.
29479 (dimode_scalar_chain::convert_reg): Ditto.
29480 (dimode_scalar_chain::convert_insn): Ditto.
29481 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
29482 (VI248_AVX512BW_1): New mode iterator.
29483 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
29484 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
29487 2017-04-25 Martin Sebor <msebor@redhat.com>
29489 PR tree-optimization/80497
29490 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
29491 constants are representable in HOST_WIDE_INT.
29492 (parse_directive): Ditto.
29494 2017-04-25 Martin Sebor <msebor@redhat.com>
29497 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
29498 (new_zero_array): Adjust signature.
29499 (dom_info::dom_init): Used unsigned rather that size_t.
29500 (dom_info::dom_info): Same.
29502 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29503 Jakub Jelinek <jakub@redhat.com>
29506 * config/arm/arm.c: Include gimple.h.
29507 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29508 returns negative, increment ncrn only if it returned positive.
29509 (arm_needs_doubleword_align): Return int instead of bool,
29510 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
29511 members, but if there is any such non-FIELD_DECL
29512 > PARM_BOUNDARY aligned decl, return -1 instead of false.
29513 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29514 returns negative, increment nregs only if it returned positive.
29515 (arm_setup_incoming_varargs): Likewise.
29516 (arm_function_arg_boundary): Emit -Wpsabi note if
29517 arm_needs_doubleword_align returns negative, return
29518 DOUBLEWORD_ALIGNMENT only if it returned positive.
29520 2017-04-25 Marek Polacek <polacek@redhat.com>
29523 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29524 first argument to type.
29526 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
29529 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
29530 type checks to test for compatibility instead of equality.
29532 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29533 Jakub Jelinek <jakub@redhat.com>
29536 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
29538 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
29539 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
29540 the alignment computation, but return their maximum in warn_alignment.
29541 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29542 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
29544 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
29545 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29548 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29550 * config/arc/simdext.md (dmpyh): Fix typo.
29552 2017-04-25 Richard Biener <rguenther@suse.de>
29554 PR tree-optimization/80492
29555 * alias.c (compare_base_decls): Handle registers with asm
29556 specification conservatively.
29557 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29558 compare_base_decls returning dont-know properly.
29560 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29562 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
29563 (legitimate_offset_address_p): New function.
29564 (arc_legitimate_address_p): Use above function.
29566 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29568 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
29570 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29572 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
29573 ACCH registers whenever they are available.
29575 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29577 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
29578 double regs fix when not used.
29580 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29582 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
29584 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
29585 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
29587 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29589 * config/arc/arc.c (arc_output_addsi): Check for h-register class
29590 when emitting short ADD instructions.
29592 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29594 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
29596 (cmpsi_cc_c_insn): Likewise.
29597 (cbranchsi4_scratch): Compute proper instruction length using
29598 compact_hreg_operand.
29599 * config/arc/predicates.md (compact_hreg_operand): New predicate.
29601 2017-04-25 Richard Biener <rguenther@suse.de>
29603 PR middle-end/80509
29604 * passes.c (pass_manager::pass_manager): Initialize
29605 m_name_to_pass_map.
29607 2017-04-25 Richard Biener <rguenther@suse.de>
29609 PR tree-optimization/79201
29610 * tree-ssa-sink.c (statement_sink_location): Handle calls.
29612 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29615 * config/s390/vector.md: Split MEM->GPR vector moves for
29616 non-s_operand addresses.
29618 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29621 * config/s390/predicates.md (reload_const_wide_int_operand): New
29623 * config/s390/s390.md ("movti"): Remove d/P alternative.
29624 ("movti_bigconst"): New pattern definition.
29626 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29629 * s390-protos.h (s390_expand_cs_hqi): Removed.
29630 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
29631 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
29632 modes as well as CCZ1mode and CCZmode.
29633 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
29634 signature of s390_emit_compare_and_swap.
29635 (s390_expand_cs_hqi): Likewise, make static.
29636 (s390_expand_cs_tdsi): Generate an explicit compare before trying
29637 compare-and-swap, in some cases.
29638 (s390_expand_cs): Wrapper function.
29639 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
29641 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
29642 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
29643 patterns for small and large integers. Forbid symref memory operands.
29644 Move expander to s390.c. Require cc register.
29645 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
29646 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
29647 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
29648 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
29649 symref memory operands. Remove CC mode and call s390_match_ccmode
29651 ("atomic_exchange<mode>"): Allow and implement all integer modes.
29653 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29655 * config/s390/s390.md (define_peephole2): New peephole to help
29656 combining the load-and-test pattern with volatile memory.
29658 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29660 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
29661 with CCZmode for TARGET_Z196.
29663 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29665 PR rtl-optimization/80501
29666 * combine.c (make_compound_operation_int): Set subreg_code to SET
29667 even for AND with mask of the sign bit of mode.
29669 PR rtl-optimization/80500
29670 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
29671 sum's initial value.
29673 2017-04-25 Julian Brown <julian@codesourcery.com>
29674 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29676 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
29678 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
29680 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
29682 2017-04-25 Julian Brown <julian@codesourcery.com>
29683 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29685 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
29686 (thunderx2t99_sha): New Reservation.
29688 2017-04-25 Julian Brown <julian@codesourcery.com>
29689 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29691 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
29692 type for 1-element load.
29694 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
29696 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
29698 2017-04-24 Martin Jambor <mjambor@suse.cz>
29700 PR tree-optimization/80293
29701 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
29702 char arrays not totally scalarizable if it is false.
29703 (analyze_all_variable_accesses): Pass correct value in the new
29704 parameter. Add a statistics counter.
29706 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
29708 PR middle-end/79931
29709 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
29711 2017-04-24 Richard Biener <rguenther@suse.de>
29713 PR tree-optimization/80494
29714 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
29715 out for complex types.
29717 2017-04-24 Richard Biener <rguenther@suse.de>
29719 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
29720 * tree-ssa-sccvn.c (print_scc): Print SCC size.
29721 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
29722 (DFS): Adjust and never fail.
29723 (sccvn_dom_walker::fail): Remove.
29724 (sccvn_dom_walker::before_dom_children): Adjust.
29725 (run_scc_vn): Likewise and never fail.
29726 * tree-ssa-pre.c (pass_pre::execute): Adjust.
29727 (pass_fre::execute): Likewise.
29729 2017-04-24 Richard Biener <rguenther@suse.de>
29731 PR tree-optimization/79725
29732 * tree-ssa-sink.c (statement_sink_location): Return whether
29733 failure reason was zero uses. Move that check later.
29734 (sink_code_in_bb): Deal with zero uses by removing the stmt
29737 2017-04-24 Richard Biener <rguenther@suse.de>
29740 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
29741 pointer-based references.
29743 2017-04-24 Richard Biener <rguenther@suse.de>
29746 * pass_manager.h (pass_manager::operator new): Remove.
29747 (pass_manager::operator delete): Likewise.
29748 * passes.c (pass_manager::operator new): Remove.
29749 (pass_manager::operator delete): Likewise.
29750 (pass_manager::pass_manager): Zero individual pass members.
29752 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
29755 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
29756 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
29757 Check "XEXP (src, 1)" operand here.
29758 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
29759 Check "XEXP (src, 1)" operand here.
29760 (dimode_scalar_chain::make_vector_copies): Detect count register
29761 of a shift instruction. Zero extend count register from QImode
29762 to DImode to satisfy vector shift pattern count operand predicate.
29763 Substitute vector shift count operand with a DImode copy.
29764 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
29767 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
29769 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
29770 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29771 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
29772 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29773 (UNSPEC_NOREX_MEM): Remove definition.
29775 2017-04-21 Richard Biener <rguenther@suse.de>
29777 PR tree-optimization/79547
29778 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29779 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
29780 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
29781 without any constraints.
29783 2017-04-21 Richard Biener <rguenther@suse.de>
29785 PR tree-optimization/78847
29786 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
29788 2017-04-21 Richard Biener <rguenther@suse.de>
29790 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
29791 (build_distinct_type_copy): Likewise.
29792 (build_variant_type_copy): Likewise.
29793 * tree.c (build_qualified_type): Pass down mem-stat info.
29794 (build_distinct_type_copy): Likewise.
29795 (build_variant_type_copy): Likewise.
29797 2017-04-21 Richard Biener <rguenther@suse.de>
29799 PR tree-optimization/80237
29800 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
29802 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
29803 for a simplified result.
29805 2016-04-21 Richard Biener <rguenther@suse.de>
29807 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
29808 sth as strict as a simple_iv but a chrec without symbols and an
29809 operand defined in the loop we are peeling (and not some subloop).
29810 (propagate_constants_for_unrolling): Propagate all constants.
29812 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29815 * config/i386/i386.c (print_reg): Remove assert for disalowed
29816 regno values, call output_operand_lossage instead.
29818 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29821 * config/i386/constraints.md (Yc): New register constraint.
29822 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
29823 Use Yc constraint for alternative 2 of operand 0. Remove
29824 preferred_for_speed attribute.
29826 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29828 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
29829 lastprivate clauses in SIMT case.
29831 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
29833 * doc/invoke.texi (-Wextra-semi): Document new warning option.
29835 2017-04-20 Richard Biener <rguenther@suse.de>
29837 PR tree-optimization/57796
29838 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
29839 as N scalar stores.
29840 (vect_model_load_cost): Cost gathers as N scalar loads.
29842 2017-04-20 Richard Biener <rguenther@suse.de>
29844 * ggc-page.c (ggc_allocated_p): Rename to ...
29845 (safe_lookup_page_table_entry): ... this and return the lookup
29847 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
29849 2017-04-20 Richard Biener <rguenther@suse.de>
29851 PR tree-optimization/80453
29852 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
29853 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
29854 from the conditions.
29855 (vn_phi_eq): Pass them down.
29856 (vn_phi_lookup): Record them.
29857 (vn_phi_insert): Likewise.
29859 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
29861 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
29862 uninitialized variable warning to avoid buffer overrun.
29864 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29867 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
29868 is suppressed for '{ 0 }' in C.
29870 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29872 * BASE-VER: Set to 8.0.0.
29874 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
29876 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
29877 priority .init_array and .fini_array section with SECTION_NOTYPE
29880 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29882 PR middle-end/80423
29883 * tree.h (build_array_type): Add typeless_storage default argument.
29884 * tree.c (type_cache_hasher::equal): Also compare
29885 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
29886 (build_array_type): Add typeless_storage argument, set
29887 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
29889 (build_nonshared_array_type): Adjust build_array_type_1 caller.
29890 (build_array_type): Likewise. Add typeless_storage argument.
29892 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29893 Jakub Jelinek <jakub@redhat.com>
29895 PR tree-optimization/80426
29896 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
29897 operation on symbolic operands, also compute the overflow for the
29898 invariant part when the operation degenerates into a negation.
29900 2017-04-19 Jakub Jelinek <jakub@redhat.com>
29903 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
29904 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
29907 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
29909 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
29912 * config/avr/avr.c (tree.h): Include it.
29913 (cgraph.h): Include it.
29914 (avr_encode_section_info): Don't warn for uninitialized progmem
29915 variable if it's just an alias.
29917 2017-04-19 Richard Biener <rguenther@suse.de>
29920 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
29921 when needed by AutoPGO.
29923 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
29926 * doc/lto.texi: Remove an extra 'that'.
29928 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
29930 PR rtl-optimization/80429
29931 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
29932 are only used in debug insns.
29934 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29935 Vladimir Makarov <vmakarov@redhat.com>
29937 * config/sparc/predicates.md (input_operand): Add comment. Return
29938 true for any memory operand when LRA is in progress.
29939 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
29941 2017-04-18 Jeff Law <law@redhat.com>
29944 * mips.md ({return,simple_return}_internal): Do not overwrite
29947 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29949 PR tree-optimization/80443
29950 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
29951 instead of adding 1, subtract -1 and similarly instead of subtracting
29954 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
29956 PR rtl-optimization/80357
29957 * haifa-sched.c (tmp_bitmap): New variable.
29958 (model_recompute): Handle duplicate use records.
29959 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
29960 (free_global_sched_pressure_data): Free it.
29962 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
29965 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29966 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29967 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29968 instead of SYSTEM_HEADER_DIR.
29970 2017-04-18 Jeff Law <law@redhat.com>
29972 PR middle-end/80422
29973 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
29974 predecessors after walking up the insn chain.
29976 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29979 * dwarf2out.c (modified_type_die): Try harder not to emit internal
29980 sizetype type into debug info.
29982 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29985 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
29986 unneeded test for TARGET_UPPER_REGS_SF.
29987 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
29989 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29992 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
29993 instead of gsi_after_labels.
29995 2017-04-18 Jeff Law <law@redhat.com>
29997 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30001 2017-04-13 Jeff Law <law@redhat.com>
30002 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30003 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30005 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30008 * config/avr/avr.c (intl.h): Include it.
30009 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30011 2017-04-18 Martin Liska <mliska@suse.cz>
30013 PR gcov-profile/78783
30014 * gcov-tool.c (gcov_output_files): Validate that destination
30015 file is either removed by the tool or by a user.
30017 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30018 Guy Benyei <guybe@mellanox.com>
30020 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30021 block, and do not negate it, the stored id is already negative.
30023 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30025 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30027 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30030 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30031 masks of options that should be turned off if the VSX vector
30032 options are turned off.
30033 (OTHER_P8_VECTOR_MASKS): Likewise.
30034 (OTHER_VSX_VECTOR_MASKS): Likewise.
30035 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30036 rs6000_disable_incompatible_switches to validate no type switches
30038 (rs6000_incompatible_switch): New function to disallow turning on
30039 other vector options if -mno-vsx, -mno-power8-vector, or
30040 -mno-power9-vector are specified.
30042 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30044 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30046 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30048 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30049 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30050 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30051 (ARG_POINTER_CFA_OFFSET): Likewise.
30053 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30055 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30056 conditions to take advantage of various optimizations.
30058 2017-04-13 Jeff Law <law@redhat.com>
30060 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30061 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30062 (zero_extendsidi2_dext): Likewise.
30064 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30067 * fold-const.c (fold_ternary_loc): Revert
30068 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30071 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30073 PR rtl-optimization/80343
30074 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30075 new scratch pseudo.
30077 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30080 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30081 to ubsan_encode_value.
30083 2017-04-13 Jeff Law <law@redhat.com>
30085 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30086 appearing in DEBUG_INSNs.
30088 2017-04-13 Martin Liska <mliska@suse.cz>
30090 PR gcov-profile/80413
30091 * gcov-io.c (gcov_write_string): Copy to buffer just when
30092 allocated size is greater than zero.
30094 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30097 * dwarf2out.c (decls_for_scope): Ignore declarations of
30098 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30100 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30103 * ipa-visibility.c (non_local_p): Fix typos.
30104 (localize_node): When localizing symbol in same comdat group,
30105 dissolve the group only when we know external symbols are going
30107 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30109 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30111 PR tree-optimization/79390
30112 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30113 order does not result in usable sequence, retry with reversed operand
30119 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30120 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30121 op0 instead of fold_convert_loc (loc, type, arg0).
30123 2017-04-12 Jeff Law <law@redhat.com>
30125 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30126 has a delay slot in the generated code.
30128 * config/cris/cris.md (cris_preferred_reload_class): Return
30129 GENNONACR_REGS rather than GENERAL_REGS.
30131 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30134 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30135 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30136 signedness of the result type.
30138 2017-04-12 Richard Biener <rguenther@suse.de>
30139 Jeff Law <law@redhat.com>
30141 PR tree-optimization/80359
30142 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30143 trim stores to TARGET_MEM_REFs.
30145 2017-04-12 Richard Biener <rguenther@suse.de>
30147 PR tree-optimization/79390
30148 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30149 threading case even more.
30151 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30154 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30155 for quad_address_p for TImode, instead of just not indexed_address.
30157 2017-04-12 Richard Biener <rguenther@suse.de>
30158 Bernd Edlinger <bernd.edlinger@hotmail.de>
30160 PR middle-end/79671
30161 * alias.c (component_uses_parent_alias_set_from): Handle
30162 TYPE_TYPELESS_STORAGE.
30163 (get_alias_set): Likewise.
30164 * tree-core.h (tree_type_common): Add typeless_storage flag.
30165 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30166 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30167 for types containing members with TYPE_TYPELESS_STORAGE.
30168 (place_field): Likewise.
30169 (layout_type): Likewise for ARRAY_TYPE.
30170 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30171 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30172 TYPE_TYPELESS_STORAGE.
30173 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30175 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30178 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30179 first argument to type.
30181 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30185 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30186 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30187 (rs6000_expand_binop_builtin): Likewise.
30188 (rs6000_expand_ternop_builtin): Likewise; also add missing
30189 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30190 vshasigma built-ins.
30191 * doc/extend.texi: Document that vec_xxpermdi's third argument
30192 must be a constant.
30194 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30196 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30197 Use shift_const cost parameter when calculating gain of STV shifts.
30199 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30201 PR rtl-optimization/70478
30202 * lra-constraints.c (process_alt_operands): Check memory for
30203 disfavoring memory insn operand.
30205 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30207 PR middle-end/80100
30208 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30209 left shift in unsigned HOST_WIDE_INT type.
30211 PR rtl-optimization/80385
30212 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
30213 (not (neg X)) into (plus X -1) for complex or non-integral modes.
30216 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
30217 if they have any depend clauses.
30219 2017-04-11 Martin Liska <mliska@suse.cz>
30222 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
30223 * ipa-split.c (split_function): Create a local comdat symbol
30224 if caller is in a comdat group.
30226 2017-04-11 Martin Liska <mliska@suse.cz>
30229 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
30232 2017-04-11 Martin Sebor <msebor@redhat.com>
30234 PR middle-end/80364
30235 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
30236 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
30238 (directive::set_width, directive::set_precision, format_character):
30240 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
30243 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
30246 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
30247 conflict, set target->arch_name instead of target->cpu_name.
30249 2017-04-11 Richard Biener <rguenther@suse.de>
30251 PR tree-optimization/80374
30252 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
30253 build_zero_cst, remove fold_convertible_p check again.
30255 2017-04-11 Martin Liska <mliska@suse.cz>
30258 * ubsan.c (instrument_object_size): Do not instrument register
30261 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30264 * config/i386/i386-builtin-types.def
30265 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
30266 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
30267 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
30268 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
30269 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
30270 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
30271 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
30272 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
30273 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
30274 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
30275 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
30276 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
30277 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
30278 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
30279 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
30280 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
30281 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
30282 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
30283 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
30284 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
30285 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
30286 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
30287 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
30288 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
30289 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
30290 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
30291 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
30292 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
30293 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
30294 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
30295 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
30296 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
30297 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
30298 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
30299 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
30300 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
30301 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
30302 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
30303 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
30304 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
30305 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
30306 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
30307 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
30308 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
30309 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
30310 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
30312 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
30313 flag to second_arg_count, handle 4 argument function type _COUNT
30314 aliases, handle second_arg_count on second argument rather than last.
30316 2017-04-10 Jeff Law <law@redhat.com>
30318 PR tree-optimization/80374
30319 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
30320 record anything if we can not convert integer_zero_node to the
30323 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30326 * config/rs6000/rs6000.c (rs6000_option_override_internal):
30327 Enhance special handling given to the TARGET_P9_MINMAX option in
30328 relation to certain other options.
30330 2017-04-10 Bin Cheng <bin.cheng@arm.com>
30332 PR tree-optimization/80153
30333 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
30334 remove POINTER_PLUS_EXPR's base part directly, rather than through
30337 2017-04-10 Richard Biener <rguenther@suse.de>
30338 Bin Cheng <bin.cheng@arm.com>
30340 PR tree-optimization/80153
30341 * tree-affine.c (aff_combination_to_tree): Get base pointer from
30342 the first element of pointer type aff_tree. Build result expr in
30344 (add_elt_to_tree): Convert to type unconditionally. Remove other
30345 fold_convert calls.
30346 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
30347 (rewrite_use_nonlinear_expr): Check invariant using iv information.
30349 2017-04-10 Richard Biener <rguenther@suse.de>
30351 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30354 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
30356 PR rtl-optimization/70478
30357 * lra-constraints.c (curr_small_class_check): New.
30358 (update_and_check_small_class_inputs): New.
30359 (process_alt_operands): Update curr_small_class_check. Disfavor
30360 alternative insn memory operands. Check available regs for small
30363 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30366 * config/mips/mips.opt (-mvirt): Update description.
30367 * doc/invoke.texi (-mvirt): Likewise.
30369 2017-04-10 Richard Biener <rguenther@suse.de>
30371 PR middle-end/80362
30372 * fold-const.c (fold_binary_loc): Look at unstripped ops when
30373 looking for NEGATE_EXPR in -A / -B to A / B folding.
30375 2017-04-10 Martin Liska <mliska@suse.cz>
30377 PR gcov-profile/80224
30378 * gcov.c (print_usage): Fix usage string.
30379 (get_gcov_intermediate_filename): Remove.
30380 (output_gcov_file): Use both for normal and intermediate format.
30381 (generate_results): Do not initialize special file for
30382 intermediate format.
30384 2017-04-10 Richard Biener <rguenther@suse.de>
30386 PR tree-optimization/80304
30387 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
30390 2017-04-10 Nathan Sidwell <nathan@acm.org>
30393 * config/rs6000/rs6000.c (rs6000_vector_type): New.
30394 (rs6000_init_builtins): Use it.
30396 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30398 * config/arm/arm.md (<mrc>): Add mode to SET source.
30399 (<mrrc>): Likewise.
30401 2017-04-10 Richard Biener <rguenther@suse.de>
30403 PR middle-end/80344
30404 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
30406 2017-04-10 Jakub Jelinek <jakub@redhat.com>
30409 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
30410 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
30411 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
30412 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
30413 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
30414 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
30415 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
30416 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
30417 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
30418 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
30419 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
30420 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
30421 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
30422 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
30423 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
30424 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
30425 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
30426 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
30427 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
30428 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
30429 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
30430 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
30431 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
30433 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
30435 PR rtl-optimization/70478
30436 * lra-constraints.c: Reverse the last patch.
30438 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
30440 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
30441 Add comment for WCHAR_T.
30443 2017-04-08 Martin Liska <mliska@suse.cz>
30446 2017-04-07 Martin Liska <mliska@suse.cz>
30449 * ipa-split.c (split_function): Add function part to a same comdat
30452 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30455 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
30457 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
30459 * rs6000/rs6000.c (vec_load_pendulum): Rename...
30460 (vec_pairing): ...to this.
30461 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
30462 (rs6000_sched_init): Adjust for name change.
30463 (struct rs6000_sched_context): Likewise.
30464 (rs6000_init_sched_context): Likewise.
30465 (rs6000_set_sched_context): Likewise.
30467 2017-04-07 Jakub Jelinek <jakub@redhat.com>
30473 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
30475 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
30476 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
30477 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
30479 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
30481 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
30483 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30485 PR rtl-optimization/70703
30486 * ira-color.c (update_conflict_hard_regno_costs): Use
30487 int64_t instead of HOST_WIDE_INT.
30489 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30491 PR rtl-optimization/70478
30492 * lra-constraints.c (process_alt_operands): Disfavor alternative
30493 insn memory operands.
30495 2017-04-07 Jeff Law <law@redhat.com>
30497 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
30498 CALL and NOTE_INSN_CALL_ARG_LOCATION.
30500 2017-04-07 Martin Liska <mliska@suse.cz>
30503 * config/aarch64/aarch64.c (aarch64_process_target_attr):
30504 Show error message instead of an ICE.
30506 2017-04-07 Martin Liska <mliska@suse.cz>
30509 * ipa-split.c (split_function): Add function part to a same comdat
30512 2017-04-07 Richard Biener <rguenther@suse.de>
30514 PR middle-end/80341
30515 * tree.c (get_unwidened): Also handle ! for_type case for
30517 * convert.c (do_narrow): Split out from ...
30518 (convert_to_integer_1): ... here. Do not pass final truncation
30519 type to get_unwidened for TRUNC_DIV_EXPR.
30521 2017-04-07 Richard Biener <rguenther@suse.de>
30523 * tree-affine.c (wide_int_ext_for_comb): Take type rather
30525 (aff_combination_const): Adjust.
30526 (aff_combination_scale): Likewise.
30527 (aff_combination_add_elt): Likewise.
30528 (aff_combination_add_cst): Likewise.
30529 (aff_combination_convert): Likewise.
30530 (add_elt_to_tree): Likewise. Remove unused argument.
30531 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
30533 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
30535 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
30537 * config/arm/arm.c (arm_default_short_enums): Use
30538 ARM_DEFAULT_SHORT_ENUMS.
30539 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
30541 2017-04-06 Jakub Jelinek <jakub@redhat.com>
30544 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
30545 members with redundant out-of-class redeclaration.
30547 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30550 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
30551 * config/i386/i386.md (*zero_extendsidi2):
30552 Add (?*x,*x) and (?*v,*v) alternatives.
30554 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30557 * config/i386/i386.c (ix86_expand_builtin)
30558 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
30559 mode from insn data. Convert operands to insn operand mode.
30560 Copy operands that don't satisfy insn predicate to a register.
30562 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
30564 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
30567 2017-04-06 Richard Biener <rguenther@suse.de>
30569 PR tree-optimization/80334
30570 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
30571 preserve alignment of accesses.
30573 2017-04-06 Richard Biener <rguenther@suse.de>
30575 PR tree-optimization/80262
30576 * tree-sra.c (build_ref_for_offset): Preserve address-space
30578 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
30579 Drop useless address-space information on MEM_REF offsets.
30581 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
30583 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
30585 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30587 PR rtl-optimization/70703
30588 * ira-color.c (update_conflict_hard_regno_costs): Use
30589 HOST_WIDE_INT instead of long.
30591 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
30594 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
30595 not defined for x86_64 target. Add -mmmx target option when __SSE2__
30597 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
30598 for x86_64 target. Handle -m3dnowa option.
30600 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30602 PR rtl-optimization/70703
30603 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
30604 (update_conflict_hard_regno_costs): Use long instead of unsigned
30605 arithmetic for cost calculation.
30607 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30608 Bernd Edlinger <bernd.edlinger@hotmail.de>
30611 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
30614 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
30617 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
30618 ptr_mode with Pmode throughout.
30619 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
30620 into probe_stack_range and use DImode.
30622 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
30625 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
30626 call_eh_return is true.
30628 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30630 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
30631 Initialize last_match_fntype_index.
30633 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30636 * tree-nvr.c: Include internal-fn.h.
30637 (pass_return_slot::execute): Ignore internal calls without
30640 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30641 Richard Biener <rguenther@suse.de>
30644 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
30645 captures used multiple times, except for the last use.
30646 * generic-match-head.c: Include gimplify.h.
30648 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30650 PR tree-optimization/79390
30651 * target.h (struct noce_if_info): Declare.
30652 * targhooks.h (default_noce_conversion_profitable_p): Declare.
30653 * target.def (noce_conversion_profitable_p): New target hook.
30654 * ifcvt.h (struct noce_if_info): New type, moved from ...
30655 * ifcvt.c (struct noce_if_info): ... here.
30656 (noce_conversion_profitable_p): Renamed to ...
30657 (default_noce_conversion_profitable_p): ... this. No longer
30659 (noce_try_store_flag_constants, noce_try_addcc,
30660 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
30661 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
30662 instead of noce_conversion_profitable_p.
30663 * config/i386/i386.c: Include ifcvt.h.
30664 (ix86_option_override_internal): Don't override
30665 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
30666 (ix86_noce_conversion_profitable_p): New function.
30667 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
30668 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
30669 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
30670 * doc/tm.texi: Regenerated.
30672 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30674 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
30677 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30680 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
30681 instructions for small multiply cores.
30683 2017-04-04 Jeff Law <law@redhat.com>
30685 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
30687 (mips_expand_vec_perm_const): Initialize elements in orig_perm
30688 that are not set by the loop over the elements.
30690 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30693 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
30694 int mode, convert_modes it to mode as unsigned, otherwise use
30695 lowpart_subreg to mode rather than SImode.
30696 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
30697 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
30698 Use DImode instead of SImode for the shift count operand.
30699 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
30702 2017-04-04 Richard Biener <rguenther@suse.de>
30704 PR middle-end/80281
30705 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
30706 arithmetic done for the negate or the plus. Simplify.
30707 (A - (-B) -> A + B): Likewise.
30708 * fold-const.c (split_tree): Make sure to not negate pointers.
30710 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
30712 PR rtl-optimization/60818
30713 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
30714 a compare of comparisons with the thing compared if this results
30715 in a different machine mode.
30717 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
30719 * alias.c (base_alias_check): Fix typo in comment.
30720 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
30721 * cgraphunit.c (symbol_table::compile): Likewise.
30722 * collect2.c (maybe_run_lto_and_relink): Likewise.
30723 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
30724 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
30725 * config/avr/avr.c (avr_map_op_t): Likewise.
30726 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
30727 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
30728 * config/epiphany/epiphany.md (movcc): Likewise.
30729 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
30730 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
30732 * config/mips/mips.c (mips_save_restore_reg): Likewise.
30733 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
30734 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
30735 * config/sh/sh.c (sh_rtx_costs): Likewise.
30736 * fold-const.c (fold_truth_andor): Likewise.
30737 * genautomata.c (collapse_flag): Likewise.
30738 * gengtype.h (struct type::u::s): Likewise.
30739 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
30740 * input.c (FORMAT_AMOUNT): Likewise.
30741 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
30742 (known_aggs_to_agg_replacement_list): Likewise.
30743 * ipa-inline-analysis.c: Likewise.
30744 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
30745 * ipa-polymorphic-call.c
30746 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
30747 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
30748 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
30750 * modulo-sched.c (apply_reg_moves): Likewise.
30751 * omp-expand.c (build_omp_regions_1): Likewise.
30752 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
30753 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
30754 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
30755 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
30756 * value-prof.c: Likewise.
30757 * var-tracking.c (val_reset): Likewise.
30759 2017-04-03 Richard Biener <rguenther@suse.de>
30761 PR tree-optimization/80275
30762 * fold-const.c (split_address_to_core_and_offset): Handle
30765 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
30767 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
30768 descriptors is at least equal to that of functions.
30770 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30772 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
30774 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30777 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
30778 (mov<IMOD4:mode>): New expander.
30779 (*mov<IMOD4:mode>_internal): New insn and split pattern.
30781 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
30783 PR rtl-optimization/79405
30784 * fwprop.c (propagations_left): New variable.
30785 (forward_propagate_into): Decrement it.
30786 (fwprop_init): Initialize it.
30787 (fw_prop): If the variable has reached zero, stop propagating.
30788 (fwprop_addr): Ditto.
30790 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30793 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
30794 a FUNCTION_DECL, pass it as decl instead of origin to
30797 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
30799 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
30802 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
30805 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
30806 TARGET_VSX_SMALL_INTEGER.
30808 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30810 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30811 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
30813 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30815 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
30816 extraction from odd-numbered MSA register.
30818 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30820 PR middle-end/80173
30821 * expmed.c (store_bit_field_1): Don't attempt to create
30822 a word subreg out of hard registers wider than word if they
30823 have HARD_REGNO_NREGS of 1 for their mode.
30825 PR middle-end/80163
30826 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
30827 conversions to integer types wider than word and pointer.
30830 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
30831 (rtx_equal_for_cselib_p): Pass 0 to it.
30832 * cselib.c (cselib_hasher::equal): Likewise.
30833 (rtx_equal_for_cselib_1): Add depth argument. If depth
30834 is 128, don't look up VALUE locs and punt. Increment
30835 depth in recursive calls when walking VALUE locs.
30837 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
30839 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
30840 (make_gcov_file_name): Use the canonical path name for generating
30842 (read_line): Fix handling of files with ascii null bytes.
30844 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
30846 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
30847 to initialise a vector register instead
30848 of using a const_int.
30850 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30852 PR translation/80189
30853 * gimplify.c (omp_default_clause): Use %qs instead of %s in
30854 diagnostic messages.
30856 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
30859 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
30860 (dfp_diex_<mode>): Update mode of operand 1.
30861 * doc/extend.texi (dxex, dxexq): Document change to return type.
30862 (diex, diexq): Document change to argument type.
30864 2017-03-30 Martin Jambor <mjambor@suse.cz>
30867 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
30868 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
30869 it reflects the signature changes performed at the callee side.
30870 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
30871 to cgraph_build_function_type_skip_args.
30872 (build_function_decl_skip_args): Adjust call to the above function.
30874 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30877 * config/i386/sse.md
30878 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
30879 register as dest whenever it is a MEM not rtx_equal_p to the
30880 corresponding dup operand, and when forcing into reg move the
30881 reg into the memory afterwards.
30882 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
30883 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
30884 for the force_reg mode.
30885 (avx512vl_vextractf128<mode>): Use register as dest either
30886 always when a MEM, or when it is a MEM not rtx_equal_p to the
30887 corresponding dup operand, or even not when it is a CONST_VECTOR
30888 depending on the mode and lo vs. hi.
30889 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
30891 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
30892 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
30893 Likewise. Require that operands[2] is even.
30894 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
30895 Remove extraneous parens. Require that operands[2] is a multiple
30897 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
30898 operands[0] is a MEM if <mask_applied>, the predicates/constraints
30899 disallow memory then.
30901 2017-03-30 Richard Biener <rguenther@suse.de>
30903 PR tree-optimization/77498
30904 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
30905 to non-constants over backedges.
30907 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
30909 PR rtl-optimization/80233
30910 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
30911 as last_combined_insn. Do not test for BARRIER_P separately.
30913 2017-03-29 Andreas Schwab <schwab@suse.de>
30916 * calls.c (prepare_call_address): Convert funexp to Pmode before
30917 copying to temp reg.
30919 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30921 PR tree-optimization/80158
30922 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
30923 Handle possible future case of more than one alternate
30925 (replace_rhs_if_not_dup): Likewise.
30926 (replace_one_candidate): Likewise.
30928 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
30930 PR rtl-optimization/80193
30931 * ira.c (ira): Do not check allocation for LRA.
30933 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
30935 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
30936 (nvptx_output_simt_exit): Declare.
30937 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
30938 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
30939 (init_softstack_frame): Move initialization of crtl->is_leaf to...
30940 (nvptx_declare_function_name): ...here. Emit declaration of local
30941 memory space buffer for omp_simt_enter insn.
30942 (nvptx_output_unisimt_switch): New.
30943 (nvptx_output_softstack_switch): New.
30944 (nvptx_output_simt_enter): New.
30945 (nvptx_output_simt_exit): New.
30946 * config/nvptx/nvptx.h (struct machine_function): New fields
30947 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
30948 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
30949 (UNSPECV_SIMT_EXIT): Ditto.
30950 (omp_simt_enter_insn): New insn.
30951 (omp_simt_enter): New expansion.
30952 (omp_simt_exit): New insn.
30953 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
30955 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
30956 (expand_GOMP_SIMT_ENTER_ALLOC): New.
30957 (expand_GOMP_SIMT_EXIT): New.
30958 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
30959 (GOMP_SIMT_ENTER_ALLOC): Ditto.
30960 (GOMP_SIMT_EXIT): Ditto.
30961 * target-insns.def (omp_simt_enter): New insn.
30962 (omp_simt_exit): Ditto.
30963 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
30965 (lower_rec_simd_input_clauses): Implement SIMT privatization.
30966 (lower_rec_input_clauses): Likewise.
30967 (lower_lastprivate_clauses): Handle SIMT privatization.
30969 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
30970 (ompdevlow_adjust_simt_enter): New.
30971 (find_simtpriv_var_op): New.
30972 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
30973 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
30975 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
30976 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
30977 (copy_decl_for_dup_finish): Ditto.
30979 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
30981 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
30984 * config/i386/i386.c (ix86_option_override_internal): Always
30985 allow -mpreferred-stack-boundary=3 for 64-bit targets.
30987 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30989 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
30991 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30993 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
30994 mark new edge's irreducible flag accordign to it.
30995 (vect_do_peeling): Check loop preheader edge's irreducible flag
30996 and pass it to function slpeel_add_loop_guard.
30998 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31000 PR tree-optimization/80218
31001 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31002 Update block frequencies and counts.
31004 2017-03-28 Richard Biener <rguenther@suse.de>
31006 PR tree-optimization/78644
31007 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31008 of a simplification result we may not use it at all.
31010 2017-03-28 Richard Biener <rguenther@suse.de>
31013 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31014 without arguments, generate default definition of a SSA name.
31016 2017-03-28 Richard Biener <rguenther@suse.de>
31018 PR middle-end/80222
31019 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31020 TYPE_REF_CAN_ALIAS_ALL references.
31021 * fold-const.c (fold_indirect_ref_1): Likewise.
31023 2017-03-28 Martin Liska <mliska@suse.cz>
31026 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31027 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31029 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31030 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31032 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31033 (EXTRA_SPECS): Define.
31034 (SUBTARGET_EXTRA_SPECS): Likewise.
31035 (SUBTARGET_CPP_SPEC): Likewise.
31036 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31037 SUBTARGET_EXTRA_SPECS.
31038 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31040 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31042 * config/arc/simdext.md (vst64_insn): Update pattern.
31043 (vld32wh_insn): Likewise.
31044 (vld32wl_insn): Likewise.
31045 (vld64_insn): Likewise.
31046 (vld32_insn): Likewise.
31048 2017-03-28 Marek Polacek <polacek@redhat.com>
31051 * fold-const.c (fold_comparison): Use protected_set_expr_location
31052 instead of SET_EXPR_LOCATION.
31054 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31056 * tree.c (add_expr): Avoid name lookup warning.
31058 2017-03-27 Jeff Law <law@redhat.com>
31060 PR tree-optimization/80216
31061 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31062 function name. Limit recursion depth.
31063 (record_temporary_equivalences): Corresponding changes.
31065 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31067 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31070 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31073 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31075 * cfgcleanup.c (reg_note_cfa_p): New array.
31076 (insns_have_identical_cfa_notes): New function.
31077 (old_insns_match_p): Don't cross-jump in between /f
31078 and non-/f instructions. If both i1 and i2 are frame related,
31079 verify all CFA notes, their order and content.
31081 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31084 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31085 HImode and SImode with zero extend to DImode to one insn.
31086 (bswap<mode>2_extenddi): Likewise.
31087 (bswapsi2_extenddi): Likewise.
31088 (bswaphi2_extendsi): Likewise.
31089 (bswaphi2): Combine bswap HImode and SImode into one insn.
31090 Separate memory insns from swapping register.
31091 (bswapsi2): Likewise.
31092 (bswap<mode>2): Likewise.
31093 (bswaphi2_internal): Delete, no longer used.
31094 (bswapsi2_internal): Likewise.
31095 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31096 store, and gpr<-gpr swap insns.
31097 (bswap<mode>2_store): Likewise.
31098 (bswaphi2_reg): Register only splitter, combine with the splitter.
31099 (bswaphi2 splitter): Likewise.
31100 (bswapsi2_reg): Likewise.
31101 (bswapsi2 splitter): Likewise.
31102 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31103 the insns into load, store, and register/register insns.
31104 (bswapdi2_ldbrx): Likewise.
31105 (bswapdi2_load): Likewise.
31106 (bswapdi2_store): Likewise.
31107 (bswapdi2_reg): Likewise.
31109 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31111 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31112 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31114 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31117 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31119 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31120 special handling for target option conflicts between dform
31121 options (-mpower9-dform, -mpower9-dform-vector,
31122 -mpower9-dform-scalar) and -mno-direct-move.
31124 2017-03-27 Richard Biener <rguenther@suse.de>
31126 PR tree-optimization/80181
31127 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31129 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31131 * config/arc/predicates.md (move_double_src_operand): Replace the
31132 call to move_double_src_operand with a call to address_operand.
31134 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31136 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31137 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31138 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31140 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31142 * config/arc/predicates.md (long_immediate_loadstore_operand):
31143 Consider scaled addresses cases.
31145 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31147 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31148 restored when in interrupt.
31149 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31150 doesn't have delay slot.
31152 2017-03-27 Richard Biener <rguenther@suse.de>
31155 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31156 inlined thunk clones.
31158 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31161 * asan.c (instrument_derefs): Copy over last operand from
31162 original COMPONENT_REF to the new COMPONENT_REF with
31163 DECL_BIT_FIELD_REPRESENTATIVE.
31164 * ubsan.c (instrument_object_size): Likewise.
31166 2017-03-27 Richard Biener <rguenther@suse.de>
31168 PR tree-optimization/80170
31169 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31170 sure DR/SCEV didnt fold in constants we do not see when looking
31171 at the reference base alignment.
31173 2017-03-27 Richard Biener <rguenther@suse.de>
31175 PR middle-end/80171
31176 * gimple-fold.c (fold_ctor_reference): Properly guard against
31177 NULL return value from canonicalize_constructor_val.
31179 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31182 * config/i386/i386.c (ix86_expand_builtin)
31183 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31184 flags reg setting and flags reg using instructions.
31185 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31186 clobbering instructions to zero extend op2.
31188 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31190 * doc/install.texi (Configuration) <--with-aix-soname>:
31191 Update link to AIX ld.
31193 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31195 PR rtl-optimization/80160
31196 PR rtl-optimization/80159
31197 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31198 reg_alternate_class into account.
31200 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31203 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31204 to consider in curr_insn_transform.
31206 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31208 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31209 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31210 and emit_mode_inner.
31212 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31214 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
31215 argument to the overloaded builtin variants. Use the new flag to
31216 deprecate certain builtin variants.
31217 * config/s390/s390-builtin-types.def: Add new builtin types.
31218 * config/s390/s390-builtins.h: Support new flags field for
31219 overloaded builtins.
31220 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
31221 (s390_macro_to_expand): Enable vector float data type.
31222 (s390_cpu_cpp_builtins_internal): Indicate support of the new
31223 builtins by incrementing the __VEC__ version number.
31224 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
31226 (s390_resolve_overloaded_builtin): Emit error messages depending
31227 on the builtin flags.
31228 * config/s390/s390.c (s390_expand_builtin): Support additional
31229 flags argument. Change error message to match the messages
31230 emitted in s390-c.c.
31231 * config/s390/s390.md: New UNSPEC_* constants.
31232 (op_type): Add new instruction types.
31233 * config/s390/vecintrin.h: Add new builtins and test data class
31235 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
31236 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
31237 (VEC_INEXACT, VEC_NOINEXACT): New constants.
31238 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
31239 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
31240 ("vec_mergel<mode>"): V_HW -> VEC_HW.
31242 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
31243 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
31244 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
31245 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
31247 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
31248 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
31249 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
31250 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
31252 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
31253 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
31254 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
31255 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
31256 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
31257 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
31258 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
31260 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
31261 ("vec_scatter_element<V_HW_4:mode>_DI")
31262 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
31263 ("vec_fpint<mode>", "vflls")
31264 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
31265 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
31266 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
31267 ("*vec_cmphe<mode>_cc"): ... these.
31269 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
31270 mode constant instead of magic value.
31272 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31274 * config/s390/s390.c (s390_expand_vec_compare): Support other
31275 vector floating point modes than just V2DF.
31276 (s390_expand_vcond): Likewise.
31277 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
31278 (s390_cannot_change_mode_class): Prevent mode changes between TF
31279 and V1TF in vector registers.
31280 * config/s390/s390.md (DF, SF): New mode attributes.
31281 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
31282 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
31283 SFmode support for VRs.
31284 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
31286 (VFT, VF_HW): New mode iterators.
31287 (vw, sdx): New mode attributes.
31288 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
31289 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
31290 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
31291 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
31292 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
31293 also the new vector floating point modes. Renaming to ...
31295 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
31296 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
31297 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
31298 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
31299 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
31300 ("vec_unordered<mode>"): ... these.
31302 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
31303 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
31304 ("*vec_extendv2df"): New insn definitions.
31306 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31308 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
31309 ("mulditi3_2", "*muldi3_sign"): New patterns.
31310 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
31311 rename the pattern definition.
31313 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31315 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
31317 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
31319 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31321 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
31322 instruction if possible.
31323 * config/s390/vector.md (vec_halfnumelts): New mode
31325 ("*vec_vllezlf<mode>"): New pattern.
31327 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31329 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
31330 ("popcountv4si2", "popcountv2di2"): Rename to ...
31331 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
31332 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
31334 ("popcount<mode>2_vxe"): New pattern.
31336 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31338 * common/config/s390/s390-common.c (processor_flags_table): Add
31340 * config.gcc: Add arch12.
31341 * config/s390/driver-native.c (s390_host_detect_local_cpu):
31342 Default to arch12 for unknown CPU model numbers.
31343 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
31344 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
31345 PROCESSOR_max sanity check.
31346 * config/s390/s390-opts.h (enum processor_type): Add
31348 * config/s390/s390.c (processor_table): Add arch12.
31349 (s390_expand_builtin): Add check for B_VXE flag.
31350 (s390_issue_rate): Add PROCESSOR_ARCH12.
31351 (s390_get_sched_attrmask): Likewise.
31352 (s390_get_unit_mask): Likewise.
31353 (s390_sched_score): Enable z13 scheduling for arch12.
31354 (s390_sched_reorder): Likewise.
31355 (s390_sched_variable_issue): Likewise.
31356 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
31358 (s390_tune_attr): Use z13 scheduling also for arch12.
31359 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
31360 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
31361 (TARGET_VXE_P): New macros.
31362 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
31363 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
31364 * config/s390/s390.opt: Add arch12 as processor_type.
31366 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31368 * config/s390/s390.md
31369 ("fixuns_truncdddi2", "fixuns_trunctddi2")
31370 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
31371 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
31373 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
31374 Rename expanders to ...
31376 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
31377 ("fixuns_truncdddi2_emu"): ... these.
31379 ("fixuns_trunc<mode>si2_emu"): New expander.
31381 ("*fixuns_truncdfdi2_z13"): Rename to ...
31382 ("*fixuns_truncdfdi2_vx"): ... this.
31384 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31386 * config/s390/2964.md: Remove the single element vector compare
31387 instructions which are no longer used.
31388 * config/s390/s390.c (s390_select_ccmode): Remove handling of
31390 (s390_canonicalize_comparison): Remove handling of DFmode
31392 (s390_expand_vec_compare_scalar): Remove function.
31393 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
31394 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
31396 ("*cmp<mode>_ccs"): Add wfcdb instruction.
31398 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31400 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
31402 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
31403 will anyway by matched by mov<mode>_64dfp.
31405 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31407 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
31408 vlef/vstef. Add missing operand to vleif.
31410 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31412 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
31413 pair for all vector types with 64 bit elements.
31414 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
31415 * config/s390/vector.md (V_HW_64): ... here.
31416 (V_128_NOSINGLE): New mode iterator.
31417 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
31418 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
31419 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
31420 ("*vec_load_pairv2di"): Change to ...
31421 ("*vec_load_pair<mode>"): ... this one.
31423 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31425 * config/s390/constraints.md: Add comments.
31426 (jKK): Reject element sizes > 8 bytes.
31427 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
31429 * config/s390/s390.md: Add the s_operand checks formerly in
31430 s390_split_ok_p to various splitters where they are still
31432 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
31433 for 128 bit vectors. Plus two splitters.
31435 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31437 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
31440 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31443 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
31444 error if the boundary argument is not constant.
31446 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31448 PR rtl-optimization/80112
31449 * loop-doloop.c (doloop_condition_get): Don't check condition
31450 if cmp isn't SET with IF_THEN_ELSE src.
31452 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31454 PR tree-optimization/80158
31455 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
31456 replacing a candidate statement, also replace it for the
31457 candidate's alternate interpretation.
31458 (replace_rhs_if_not_dup): Likewise.
31459 (replace_one_candidate): Likewise.
31461 2017-03-24 Richard Biener <rguenther@suse.de>
31463 PR tree-optimization/80167
31464 * graphite-isl-ast-to-gimple.c
31465 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
31467 (translate_isl_ast_to_gimple::get_rename): Likewise.
31469 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31471 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
31472 handling of certain combinations of target options, including the
31473 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
31474 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
31476 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31479 * config/arm/arm.md (*load_multiple): Add reload_completed to
31480 matching condition.
31482 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31483 Richard Biener <rguenth@suse.de>
31485 PR tree-optimization/79908
31486 PR tree-optimization/80136
31487 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31488 been cast away, gimplify_and_add suffices.
31490 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
31492 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
31494 2017-03-23 Richard Biener <rguenther@suse.de>
31496 PR tree-optimization/80032
31497 * gimplify.c (gimple_push_cleanup): Forced unconditional
31498 cleanups still have to go to the conditional_cleanups
31501 2017-03-22 Jakub Jelinek <jakub@redhat.com>
31503 PR tree-optimization/80072
31504 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
31506 (next_operand_entry_id): Change type to unsigned int.
31507 (sort_by_operand_rank): Make sure to return the right return value
31508 even if unsigned fields are bigger than INT_MAX.
31509 (struct oecount): Change cnt and id type to unsigned int.
31510 (oecount_hasher::equal): Formatting fix.
31511 (oecount_cmp): Make sure to return the right return value
31512 even if unsigned fields are bigger than INT_MAX.
31513 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
31516 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
31517 TREE_READONLY on result if writing it more than once.
31520 * doc/invoke.texi (-fsanitize=thread): Document that with
31521 -fnon-call-exceptions atomics are not able to throw
31525 * tsan.c: Include tree-eh.h.
31526 (instrument_builtin_call): Call maybe_clean_eh_stmt or
31527 maybe_clean_or_replace_eh_stmt where needed.
31528 (instrument_memory_accesses): Add cfg_changed argument.
31529 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
31530 if it returned true.
31531 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
31533 PR rtl-optimization/63191
31534 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
31535 wrapper function, moved the whole old content into ...
31536 (ix86_delegitimize_address_1): ... this. New inline function.
31537 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
31538 true as last argument instead of ix86_delegitimize_address.
31540 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31542 * config/aarch64/aarch64.c (generic_branch_cost): Copy
31543 cortexa57_branch_cost.
31545 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31547 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
31549 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31552 * doc/md.texi (Constraints): Document wA constraint.
31553 * config/rs6000/constraints.md (wA): New.
31554 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
31555 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
31556 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
31557 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
31559 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
31562 * gimplify.c (is_oacc_declared): New function.
31563 (oacc_default_clause): Use it to set default flags for acc declared
31564 variables inside parallel regions.
31565 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
31566 declared variables.
31567 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
31568 declare attribute to any decl as necessary.
31570 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
31573 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
31574 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
31575 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
31576 (arm_arch_lpae): This.
31577 * config/arm/arm.c (arm_arch7ve): Rename into ...
31578 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
31579 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
31582 2017-03-22 Martin Liska <mliska@suse.cz>
31585 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
31586 error message instead of an ICE.
31588 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31590 * doc/extend.texi (6.11 Additional Floating Types): Revise.
31592 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
31594 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31596 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31599 2017-03-21 Martin Sebor <msebor@redhat.com>
31601 * doc/extend.texi: Use "cannot" instead of "can't."
31602 * doc/hostconfig.texi: Same.
31603 * doc/install.texi: Same.
31604 * doc/invoke.texi: Same.
31605 * doc/loop.texi: Same.
31606 * doc/md.texi: Same.
31607 * doc/objc.texi: Same.
31608 * doc/rtl.texi: Same.
31609 * doc/tm.texi: Same.
31610 * doc/tm.texi.in: Same.
31611 * doc/trouble.texi: Same.
31613 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
31616 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
31617 (collect_checksum_attributes): Set it.
31618 (die_checksum_ordered): Use it.
31620 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31622 PR tree-optimization/79908
31623 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
31624 change: For a VA_ARG whose LHS has been cast away, use
31625 force_gimple_operand to construct the side effects.
31627 2017-03-21 David Malcolm <dmalcolm@redhat.com>
31629 PR translation/80001
31630 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
31631 more amenable to translation.
31632 (oacc_loop_auto_partitions): Likewise.
31634 2017-03-21 Marek Polacek <polacek@redhat.com>
31635 Martin Sebor <msebor@redhat.com>
31637 PR tree-optimization/80109
31638 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
31639 on INTEGRAL_TYPE_P.
31641 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31642 Segher Boessenkool <segher@kernel.crashing.org>
31645 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
31646 check reg_used_between_p between insn and one of succ or succ2
31647 depending on if succ is artificial insn not inserted into insn
31650 2017-03-21 Martin Liska <mliska@suse.cz>
31652 PR gcov-profile/80081
31653 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
31654 * doc/gcc.texi: Include gcov-dump stuff.
31655 * doc/gcov-dump.texi: New file.
31657 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
31659 PR rtl-optimization/79150
31660 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
31661 conditional jump, if the jump is the last insn of the loop.
31663 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31664 Richard Biener <rguenth@suse.de>
31666 PR tree-optimization/79908
31667 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31668 been cast away, use force_gimple_operand to construct the side
31671 2017-03-21 Martin Liska <mliska@suse.cz>
31673 PR libfortran/79956
31674 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
31677 2017-03-21 Brad Spengler <spender@grsecurity.net>
31680 * plugin.c (htab_hash_plugin): New function.
31681 (add_new_plugin): Use it and adjust.
31682 (parse_plugin_arg_opt): Adjust.
31683 (init_one_plugin): Likewise.
31685 2017-03-21 Richard Biener <rguenther@suse.de>
31687 PR tree-optimization/80032
31688 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
31689 if set force the cleanup to happen unconditionally.
31690 (gimplify_target_expr): Push inserted clobbers with force_uncond
31691 to avoid them being removed by control-dependent DCE.
31693 2017-03-21 Richard Biener <rguenther@suse.de>
31695 PR tree-optimization/80122
31696 * tree-inline.c (copy_bb): Do not expans va-arg packs or
31697 va_arg_pack_len when the inlined call stmt requires pack
31699 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
31701 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31704 * tsan.c (instrument_builtin_call): If the memory model argument
31705 is not a constant, assume it is valid.
31708 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
31711 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31713 PR rtl-optimization/79910
31714 * combine.c (can_combine_p): Do not allow combining an I0 or I1
31715 if its dest is used by an insn before I2 (other than the combined
31716 insns themselves, which are properly handled already).
31718 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31721 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31723 * combine.c (record_used_regs): New static function.
31724 (try_combine): Handle situations where there is an additional
31725 instruction between I2 and I3 which needs to have a LOG_LINK
31729 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31731 * combine.c (try_combine): Delete redundant i1 test. Call
31732 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31734 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31737 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
31738 alternatives 13/14.
31740 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31742 PR tree-optimization/80054
31743 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
31744 the optimization if a PHI or any of its arguments is not dominated
31745 by the candidate's basis. Use gphi* rather than gimple* as
31747 (replace_profitable_candidates): Clean up a gimple* variable that
31748 should be a gphi* variable.
31750 2017-03-20 Martin Sebor <msebor@redhat.com>
31753 * doc/extend.texi (attribute constructor): Document present limitation.
31755 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
31758 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
31759 __POWER9_VECTOR__ #ifdef control, change template definition to
31760 use Power9-specific built-in function.
31761 (vec_any_eq): Likewise.
31762 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
31763 to control outcomes from this test.
31764 (vector_ae_<mode>p): For VEC_F modes, likewise.
31766 2017-03-20 Ian Lance Taylor <iant@google.com>
31768 * config/i386/i386.c (ix86_function_regparm): Save an extra
31769 register for -fsplit-stack with DECL_STATIC_CHAIN.
31771 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31774 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
31775 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
31777 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31779 * config/riscv/riscv.c (riscv_print_operand): Use "fence
31781 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
31784 2017-03-20 Marek Polacek <polacek@redhat.com>
31787 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
31789 2017-03-20 Richard Biener <rguenther@suse.de>
31791 PR tree-optimization/80113
31792 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
31793 allocate extra SSA name for PHI def.
31794 (add_close_phis_to_outer_loops): Likewise.
31795 (add_close_phis_to_merge_points): Likewise.
31796 (copy_loop_close_phi_args): Likewise.
31797 (copy_cond_phi_nodes): Likewise.
31799 2017-03-20 Martin Liska <mliska@suse.cz>
31801 PR middle-end/79753
31802 * tree-chkp.c (chkp_build_returned_bound): Do not build
31803 returned bounds for a LHS that's not a BOUNDED_P type.
31805 2017-03-20 Martin Liska <mliska@suse.cz>
31809 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
31810 COMPLEX_CST and VECTOR_CST.
31812 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31815 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
31816 target operand. A new splitter adds the clobber statement in case
31817 the target operand is dead anyway.
31819 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
31821 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
31822 to age-old versions of binutils and glibc.
31824 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
31826 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
31828 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31830 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
31832 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31834 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
31835 requirement for binutils 2.13.
31837 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31839 * combine.c (try_combine): Delete redundant i1 test. Call
31840 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31842 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
31844 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
31845 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
31847 <riscv64-*-elf>: Re-arrange section
31848 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
31849 <riscv32-*-linux>: Likewise.
31850 <riscv64-*-elf>: Likewise
31851 <riscv64-*-linux>: Likewise.
31853 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
31856 * aarch64.opt(verbose-cost-dump): Fix typo.
31858 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
31861 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
31862 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
31864 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31866 * reload.c (find_reloads): When reloading a nonoffsettable address,
31867 use RELOAD_OTHER for it and its address reloads.
31869 PR rtl-optimization/79910
31870 * combine.c (record_used_regs): New static function.
31871 (try_combine): Handle situations where there is an additional
31872 instruction between I2 and I3 which needs to have a LOG_LINK
31875 2017-03-17 Jeff Law <law@redhat.com>
31877 PR tree-optimization/71437
31878 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
31879 conditional in the hash table first.
31880 (vrp_dom_walker::before_dom_children): Extract condition from
31881 ASSERT_EXPR. Record condition, its inverion and any implied
31882 conditions as well.
31884 2017-03-17 Marek Polacek <polacek@redhat.com>
31885 Markus Trippelsdorf <markus@trippelsdorf.de>
31887 PR tree-optimization/80079
31888 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
31891 2017-03-17 Richard Biener <rguenther@suse.de>
31893 PR middle-end/80075
31894 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
31895 Properly verify the LHS before the RHS possibly claims to be
31897 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
31900 2017-03-17 Martin Jambor <mjambor@suse.cz>
31902 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
31903 (List of -O2 options): Likewise.
31904 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
31907 2017-03-17 Tom de Vries <tom@codesourcery.com>
31909 * gcov-dump.c (print_usage): Print bug_report_url.
31911 2017-03-17 Richard Biener <rguenther@suse.de>
31913 PR middle-end/80050
31914 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
31915 (parser::peek): Likewise.
31917 2017-03-17 Richard Biener <rguenther@suse.de>
31919 PR tree-optimization/80048
31920 * sese.c (free_sese_info): Properly release rename_map and
31921 copied_bb_map elements.
31923 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
31925 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
31926 Add linked-list forward and backlinks. Insert on
31927 construction, remove on destruction.
31928 (class pass_store_merging): Add m_stores_head field.
31929 (pass_store_merging::terminate_and_process_all_chains):
31930 Iterate over m_stores_head list.
31931 (pass_store_merging::terminate_all_aliasing_chains):
31933 (pass_store_merging::execute): Check for debug stmts first.
31934 Push new chains onto the m_stores_head stack.
31936 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
31939 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
31940 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
31941 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
31943 2017-03-16 Jeff Law <law@redhat.com>
31945 PR tree-optimization/71437
31946 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
31947 member function. Implementation moved into after_dom_children
31948 member function and into the threader's thread_outgoing_edges
31950 (dom_opt_dom_walker::after_dom_children): Simplify by moving
31951 some code into new thread_outgoing_edges.
31952 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
31953 definition. Simplify marker handling (do it here). Assume we always
31954 have the available expression and the const/copies tables.
31955 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
31957 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
31958 * tree-vrp.c (equiv_stack): No longer file scoped.
31959 (vrp_dom_walker): New class.
31960 (vrp_dom_walker::before_dom_children): New member function.
31961 (vrp_dom_walker::after_dom_children): Likewise.
31962 (identify_jump_threads): Setup domwalker. Use it rather than
31963 walking edges in a random order by hand. Simplify setup/finalization.
31964 (finalize_jump_threads): Remove.
31965 (vrp_finalize): Do not call identify_jump_threads here.
31966 (execute_vrp): Do it here instead and call thread_through_all_blocks
31969 PR tree-optimization/71437
31970 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
31972 (simplify_stmt_for_jump_threading): Add basic_block argument. All
31974 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
31975 (dom_opt_dom_walker::thread_across_edge): Remove
31976 handle_dominating_asserts argument. All callers changed.
31977 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
31978 changes. Remove calls to lhs_of_dominating_assert. Other
31979 uses of handle_dominating_asserts turn into unconditional code
31980 (simplify_control_stmt_condition_1): Likewise.
31981 (simplify_control_stmt_condition): Likewise.
31982 (thread_through_normal_block, thread_across_edge): Likewise.
31983 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
31984 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
31985 object if it is not an SSA_NAME.
31986 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
31987 before calling into the VRP specific simplifiers.
31988 (identify_jump_threads): Remove handle_dominating_asserts
31991 2017-03-16 Jakub Jelinek <jakub@redhat.com>
31994 * tree-diagnostic.c (default_tree_printer): No longer static.
31995 * tree-diagnostic.h (default_tree_printer): New prototype.
31997 2017-03-16 Tamar Christina <tamar.christina@arm.com>
31999 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32000 Change ins into fmov.
32002 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32004 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32005 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32006 Use h_con constraint for operand 1.
32007 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32008 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32010 2017-03-15 Jeff Law <law@redhat.com>
32012 PR tree-optimization/71437
32013 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32014 (record_temporary_equivalences): Use it.
32016 PR tree-optimization/71437
32017 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32018 tree-ssa-scopedtables.
32019 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32020 (record_conditions, record_cond, vuse_eq): Likewise.
32021 (record_edge_info): Adjust to API tweak of record_conditions.
32022 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32023 (record_temporary_equivalences, optimize_stmt): Likewise.
32024 (eliminate_redundant_computations): Likewise.
32025 (record_equivalences_from_stmt): Likewise.
32026 * tree-ssa-scopedtables.c: Include options.h and params.h.
32027 (vuse_eq): New function, moved from tree-ssa-dom.c
32028 (build_and_record_new_cond): Likewise.
32029 (record_conditions): Likewise. Accept vector of conditions rather
32030 than edge_equivalence structure for first argument.
32031 for the first argument.
32032 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32033 from tree-ssa-dom.c.
32034 (avail_exprs_stack::record_cond): Likewise.
32035 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32036 from tree-ssa-dom.c.
32037 (avail_exprs_stack): Add new member functions lookup_avail_expr
32039 (record_conditions): Declare.
32041 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32044 * lra-constraints.c (process_alt_operands): Increase reject for
32045 reloading an input/output operand.
32047 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32050 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32051 insns to convert from signed/unsigned char/short to IEEE 128-bit
32053 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32055 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32058 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32059 subreg of inner mode for values already in registers.
32061 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32063 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32064 iteration reg is used after the loop.
32066 2017-03-14 Martin Sebor <msebor@redhat.com>
32068 PR tree-optimization/79800
32069 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32070 precision in negative-positive range.
32071 (format_floating): Call non-const overload with adjusted precision.
32073 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32076 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32079 2017-03-14 Martin Sebor <msebor@redhat.com>
32081 PR middle-end/80020
32082 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32083 * builtins.def (aligned_alloc): Use it.
32086 * Makefile.in (GTFILES): Add calls.c.
32087 * calls.c: Include "gt-calls.h".
32089 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32091 PR rtl-optimization/79728
32092 * regs.h (struct target_regs): New field
32093 x_contains_allocatable_regs_of_mode.
32094 (contains_allocatable_regs_of_mode): New macro.
32095 * reginfo.c (init_reg_sets_1): Initialize it, and change
32096 contains_reg_of_mode so it includes global regs as well.
32097 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32098 rather than contains_regs_of_mode.
32100 2017-03-14 Martin Liska <mliska@suse.cz>
32102 * doc/invoke.texi: Document options that can't be combined with
32103 -fcheck-pointer-bounds.
32105 2017-03-14 Martin Liska <mliska@suse.cz>
32107 PR middle-end/79831
32108 * doc/invoke.texi (-Wchkp): Document the option.
32110 2017-03-14 Martin Liska <mliska@suse.cz>
32112 * Makefile.in: Install gcov-dump.
32114 2017-03-14 Martin Liska <mliska@suse.cz>
32116 * multiple_target.c (expand_target_clones): Bail out for
32117 an invalid attribute.
32119 2017-03-14 Richard Biener <rguenther@suse.de>
32121 * alias.c (struct alias_set_entry): Pack properly.
32122 * cfgloop.h (struct loop): Likewise.
32123 * cse.c (struct set): Likewise.
32124 * ipa-utils.c (struct searchc_env): Likewise.
32125 * loop-invariant.c (struct invariant): Likewise.
32126 * lra-remat.c (struct cand): Likewise.
32127 * recog.c (struct change_t): Likewise.
32128 * rtl.h (struct address_info): Likewise.
32129 * symbol-summary.h (function_summary): Likewise.
32130 * tree-loop-distribution.c (struct partition): Likewise.
32131 * tree-object-size.c (struct object_size_info): Likewise.
32132 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32133 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32134 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32135 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32136 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32137 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32138 (struct _stmt_vec_info): Likewise.
32140 2017-03-14 Martin Liska <mliska@suse.cz>
32143 * multiple_target.c (create_dispatcher_calls): Check that
32144 a target can create a function dispatcher.
32146 2017-03-14 Martin Liska <mliska@suse.cz>
32149 * multiple_target.c (expand_target_clones): Drop local.local
32150 flag for default implementation.
32152 2017-03-14 Richard Biener <rguenther@suse.de>
32154 PR tree-optimization/80030
32155 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32157 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32159 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32160 gcc_fallthrough() instead of __attribute__((fallthrough));
32162 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32164 * doc/gcc.texi: Remove "up" link to (DIR).
32165 * doc/gccint.texi: Ditto.
32167 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32169 * doc/install.texi (Specific) <avr>: Remove reference to
32172 2017-03-13 Jeff Law <law@redhat.com>
32174 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32175 attribute rather than comments.
32177 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32178 match_scratch operand is highest.
32180 2017-03-13 Martin Liska <mliska@suse.cz>
32182 PR middle-end/78339
32183 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32184 is a CHKP clone, use original declaration.
32186 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32188 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32189 (arc_conditional_register_usage): Use a different allocation order
32190 when optimizing for size.
32191 * common/config/arc/arc-common.c (arc_option_optimization_table):
32192 Section anchors default on when optimizing for size.
32194 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32196 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32198 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32200 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32201 * config/arc/arc.md (cpu_facility): Add cd variant.
32202 (*movqi_insn): Add code density variant.
32203 (*movhi_insn): Likewise.
32204 (*movqi_insn): Likewise.
32205 (*addsi3_mixed): Likewise.
32206 (subsi3_insn): Likewise.
32208 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32210 * config/arc/arc.md (movsi_cond_exec): Update constraint.
32212 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32214 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
32215 expressions with MINUS and UNARY ops.
32217 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32220 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
32222 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
32223 between vec_select and vector argument.
32224 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
32225 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
32226 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
32227 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
32228 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
32229 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
32231 2017-03-13 Richard Biener <rguenther@suse.de>
32234 * params.def (vect-max-peeling-for-alignment): Fix typo.
32236 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32238 * doc/install.texi (Specific) <mips-*-*>: Remove description of
32239 issue that only occurred with binutils below 2.18.
32241 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32243 * doc/install.texi (Specific) <cris-axis-elf>: No longer
32244 refer to binutils 2.11/2.12 minimum.
32246 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32248 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
32249 ftp.kernel.org and simplify binutils requirement.
32251 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
32253 * doc/invoke.texi (Warning Options): Fix spelling of link-time
32255 (Optimize Options): Ditto. Also remove redundancy.
32257 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32259 PR translation/79848
32260 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
32262 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
32263 to G_ to avoid double translation.
32265 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32267 PR translation/79923
32268 * auto-profile.c (get_combined_location): Convert leading
32269 character of diagnostics to lower case and remove trailing period.
32270 (read_profile): Likewise for various diagnostics.
32271 * config/arm/arm.c (arm_option_override): Remove trailing period
32272 from various diagnostics.
32273 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
32274 (msp430_expand_delay_cycles): Likewise.
32276 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32279 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
32280 full command-line argument, rather than just "str".
32281 (aarch64_validate_march): Likewise.
32282 (aarch64_validate_mtune): Likewise.
32284 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
32286 PR rtl-optimization/78911
32287 * lra-assigns.c (must_not_spill_p): New function.
32288 (spill_for): Use it.
32290 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32292 PR tree-optimization/79981
32293 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
32294 ATOMIC_COMPARE_EXCHANGE ifn result.
32295 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
32296 IFN_ATOMIC_COMPARE_EXCHANGE.
32298 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32301 * opts.c (parse_sanitizer_options): Add missing question mark to
32302 "did you mean" message.
32304 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32306 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
32308 (VMULEUH_UNS): Likewise.
32309 (VMULOUB_UNS): Likewise.
32310 (VMULOUH_UNS): Likewise.
32311 * config/rs6000/rs6000.c (builtin_function_type): Remove
32312 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
32314 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32317 * read-rtl-function.c (function_reader::read_rtx_operand): Update
32318 x with result of extra_parsing_for_operand_code_0.
32319 (function_reader::extra_parsing_for_operand_code_0): Convert
32320 return type from void to rtx, returning x. When reading
32321 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
32322 larger size containing struct block_symbol.
32324 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
32326 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
32327 -mfloat128-hardware without -m64.
32329 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
32332 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
32333 entries to the case statement that marks unsigned arguments to
32334 overloaded functions.
32336 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32338 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
32339 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
32341 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
32344 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
32345 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
32347 2017-03-10 Martin Liska <mliska@suse.cz>
32351 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
32352 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
32355 2017-03-10 Olivier Hainque <hainque@adacore.com>
32357 * tree-switch-conversion (array_value_type): Start by resetting
32358 candidate type to it's main variant.
32360 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32362 PR rtl-optimization/79909
32363 * combine.c (try_combine): Use simplify_replace_rtx on individual
32364 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
32365 of the whole CALL_INSN_FUNCTION_USAGE.
32367 PR tree-optimization/79972
32368 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
32369 get_range_info on SSA_NAMEs. Formatting fixes.
32371 2017-03-10 Richard Biener <rguenther@suse.de>
32372 Jakub Jelinek <jakub@redhat.com>
32374 PR tree-optimization/77975
32375 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
32376 edge to be constant.
32377 (get_val_for): For constant x return it. Formatting fix.
32378 (loop_niter_by_eval): Avoid pointless looping if the next iteration
32379 would use the same bases as the current one.
32381 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32383 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
32384 instead of vec_select for V1TImode.
32385 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
32387 (VSX_LE_128): Add V1TI to this mode iterator.
32388 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
32389 (*vsx_le_perm_store_<mode>): Likewise.
32390 (pre-reload splitter for VSX stores): Likewise.
32391 (post-reload splitter for VSX stores): Likewise.
32392 (*vsx_xxpermdi2_le_<mode>): Likewise.
32393 (*vsx_lxvd2x2_le_<mode>): Likewise.
32394 (*vsx_stxvd2x2_le_<mode>): Likewise.
32396 2017-03-09 Michael Eager <eager@eagercon.com>
32398 Correct failures with --enable-checking=yes,rtl.
32400 * config/microblaze/microblaze.c (microblaze_expand_shift):
32401 Replace GET_CODE test with CONST_INT_P and INTVAL test with
32402 test for const0_rtx.
32403 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
32404 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
32406 2017-03-09 Richard Biener <rguenther@suse.de>
32408 PR tree-optimization/79977
32409 * graphite-scop-detection.c (scop_detection::merge_sese):
32410 Handle the case of extra exits to blocks dominating the entry.
32412 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
32414 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
32417 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
32419 PR rtl-optimization/79949
32420 * lra-constraints.c (process_alt_operands): Check memory when
32421 trying to predict a cycle. Print about the overall increase.
32423 2017-03-09 Richard Biener <rguenther@suse.de>
32425 PR middle-end/79971
32426 * gimple-expr.c (useless_type_conversion_p): Preserve
32427 TYPE_SATURATING for fixed-point types.
32429 2017-03-09 Richard Biener <rguenther@suse.de>
32432 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
32433 alignment of BLKmode params.
32435 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32438 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
32439 (VALL_NO_V2Q): Likewise.
32441 * config/aarch64/aarch64-simd.md
32442 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
32444 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
32445 VALL_NO_V2Q mode iterator.
32446 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
32448 2017-03-09 Martin Liska <mliska@suse.cz>
32450 PR tree-optimization/79631
32451 * tree-chkp-opt.c (chkp_is_constant_addr): Call
32452 tree_int_cst_sign_bit just for INTEGER constants.
32454 2017-03-09 Martin Liska <mliska@suse.cz>
32458 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
32461 2017-03-09 Marek Polacek <polacek@redhat.com>
32464 * tree.c (inchash::add_expr): Handle TREE_VEC.
32466 2017-03-09 Martin Liska <mliska@suse.cz>
32469 (chkp_narrow_size_and_offset): New function.
32470 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
32471 (void chkp_parse_bit_field_ref): New function.
32472 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
32473 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
32475 2017-03-09 Martin Liska <mliska@suse.cz>
32478 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
32479 (chkp_find_bounds_1): Remove gcc_unreachable.
32481 2017-03-09 Jakub Jelinek <jakub@redhat.com>
32484 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
32485 BUILT_IN_SYNC*, determine the access type from the size suffix and
32486 always build a MEM_REF with that type. Handle forgotten
32487 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
32490 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
32491 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
32492 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
32493 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
32494 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
32495 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
32496 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
32497 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
32498 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
32499 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
32500 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
32501 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
32502 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
32503 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
32504 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
32505 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
32506 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
32507 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
32508 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
32509 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
32510 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
32511 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
32512 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
32513 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
32514 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
32515 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
32516 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
32517 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
32518 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
32519 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
32520 definitions outside of __OPTIMIZE__ guarded section.
32523 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
32524 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
32525 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
32526 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
32529 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32531 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
32532 ("vfenez<mode>"): Add missing constraints.
32534 2017-03-08 Martin Sebor <msebor@redhat.com>
32537 * config/nds32/nds32.c (nds32_option_override):
32538 Fix misspelled diagnostic.
32540 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32543 * gimplify.c (gimplify_omp_for): Replace index var in outer
32544 taskloop statement with an artificial variable and add
32545 OMP_CLAUSE_PRIVATE clause for it.
32547 2017-03-08 Richard Biener <rguenther@suse.de>
32549 PR tree-optimization/79955
32550 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
32551 for accesses that are completely outside of the variable.
32553 2017-03-08 Andrew Haley <aph@redhat.com>
32555 PR tree-optimization/79943
32556 * tree-ssa-loop-split.c (compute_new_first_bound): When
32557 calculating the new upper bound, (END-BEG) should be added, not
32560 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32562 * config/avr/avr.md (setmemhi): Make sure match_dup
32563 operand number comes before match_scratch.
32565 2017-03-08 Richard Biener <rguenther@suse.de>
32567 PR tree-optimization/79920
32568 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
32569 with ncopies == 1 to ...
32570 (vect_transform_slp_perm_load): ... here. Properly compute
32571 all element loads by iterating VF times over the group. Do
32572 not handle ncopies (computed in a broken way) in
32573 vect_create_mask_and_perm.
32575 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32578 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
32579 is a uniform vector, use uniform_vector_p return value instead of
32580 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
32582 2017-03-07 Marek Polacek <polacek@redhat.com>
32584 PR middle-end/79809
32585 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
32586 (alloca_call_type): Likewise.
32588 2017-03-07 Martin Liska <mliska@suse.cz>
32590 * gcov.c (process_args): Put comment to correct location.
32592 2017-03-07 Martin Liska <mliska@suse.cz>
32594 PR middle-end/68270
32595 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
32596 Use array_at_struct_end_p instead of DECL_CHAIN (field).
32597 (chkp_narrow_bounds_for_field): Likewise.
32598 (chkp_parse_array_and_component_ref): Pass one more argument to
32601 2017-03-07 Richard Biener <rguenther@suse.de>
32603 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
32606 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
32608 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
32609 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
32611 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32614 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
32615 to end of description.
32616 (PARAM_MAX_STORES_TO_MERGE): Likewise.
32618 2017-03-07 Jakub Jelinek <jakub@redhat.com>
32620 PR rtl-optimization/79901
32621 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
32623 (*avx512f_<code><mode>3<mask_name>): ... this.
32624 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
32625 iterator instead of VI8_AVX2_AVX512BW.
32627 PR rtl-optimization/79901
32628 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
32629 min/max expander, expand it using expand_vec_cond_expr.
32632 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
32635 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32638 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
32639 to void * for PCH reasons.
32640 * dwarf2out.c (output_loc_operands, output_die): Cast
32641 v.val_vec.array to unsigned char *.
32643 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
32646 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
32649 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
32651 PR rtl-optimization/79571
32652 * lra-constraints.c (process_alt_operands): Calculate static
32653 reject and subtract it from overall when only addresses will be
32656 2017-03-06 Julia Koval <julia.koval@intel.com>
32659 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
32660 incoming stack boundary to 128 for 64-bit targets.
32662 2017-03-06 Richard Biener <rguenther@suse.de>
32664 PR tree-optimization/79894
32665 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
32666 to NULL after folding it.
32668 2017-03-06 Richard Biener <rguenther@suse.de>
32670 PR tree-optimization/79824
32671 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
32672 check disabling peeling for gaps.
32674 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
32676 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
32677 attributes): Document gettimeofday.
32679 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32681 * config/s390/s390.c (s390_option_override_internal): Set
32682 PARAM_MIN_VECT_LOOP_BOUND
32684 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32686 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
32687 * config/s390/s390.md: Likewise.
32689 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32692 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
32693 (<avx2_avx512>_perm<mode>): Rename to ...
32694 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
32696 (<avx512>_perm<mode>_mask): Rename to ...
32697 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
32699 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
32700 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
32701 instead of VI8F_256_512.
32702 (avx512f_perm<mode>): New define_expand.
32703 (avx512f_perm<mode>_mask): Likewise.
32704 (avx512f_perm<mode>_1<mask_name>): New define_insn.
32705 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
32707 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32709 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
32710 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
32712 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
32714 2017-03-06 Martin Liska <mliska@suse.cz>
32717 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
32718 when having a SSA NAME w/o VAR_DECL assigned to it.
32720 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32722 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
32723 msa_dpsub_<su>_d): Fix MODE for vec_select.
32725 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32727 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
32729 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
32731 2017-03-06 Richard Biener <rguenther@suse.de>
32733 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
32734 * plugin.c (register_plugin_info): Likewise.
32735 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
32737 2017-03-05 Jakub Jelinek <jakub@redhat.com>
32739 * config/i386/sse.md (sse_storehps, sse_storelps,
32740 avx_<castmode><avxsizesuffix>_<castmode>,
32741 avx512f_<castmode><avxsizesuffix>_<castmode>,
32742 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
32743 in condition that at least one operand is not a MEM.
32745 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32747 PR middle-end/79805
32748 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
32749 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
32751 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
32752 gimple_call_nothrow_p flag based on whether original builtin can throw.
32753 If it can, emit following stmts on the fallthrough edge.
32754 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
32755 don't create new bb if inserting just debug stmts on the edge, try to
32756 insert them on the fallthru bb or just reset debug stmts.
32758 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
32761 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
32762 restore recog_data (including the operand rtxes inside it) around
32763 the call to get_insn_template.
32765 2017-03-03 Martin Sebor <msebor@redhat.com>
32767 PR tree-optimization/79699
32768 * context.c (context::~context): Free MPFR caches to avoid
32769 a memory leak on program exit.
32771 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32773 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
32774 Use wide_int::ulow () instead of .elt (0).
32776 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32778 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
32779 (*pushxf): Limit oF constraint to 32bit targets and add oC
32780 constraint for 64bit targets.
32781 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
32782 (*pushdf): Change rmF constraint to rmC.
32784 2017-03-03 Martin Liska <mliska@suse.cz>
32786 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
32787 Remove unused variable.
32789 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32792 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
32793 is a memory operand, increase num_memory.
32794 (ix86_expand_args_builtin): Likewise.
32796 2017-03-03 Jan Hubicka <jh@suse.cz>
32799 * ipa-devirt.c (maybe_record_node): Properly handle
32800 __cxa_pure_virtual visibility.
32802 2017-03-03 Martin Liska <mliska@suse.cz>
32804 PR tree-optimization/79803
32805 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
32807 (pass_loop_prefetch::execute): Disabled optimization if an
32808 assumption about L1 cache size is not met.
32810 2017-03-03 Martin Liska <mliska@suse.cz>
32812 PR rtl-optimization/79574
32813 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
32814 (hash_scan_set): Likewise.
32815 (dump_hash_table): Likewise.
32816 (hoist_code): Likewise.
32818 2017-03-03 Richard Biener <rguenther@suse.de>
32820 * fixed-value.c (fixed_from_string): Restore use of elt (1)
32821 in place of uhigh ().
32822 (fixed_convert_from_real): Likewise.
32824 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32827 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
32829 2017-03-03 Richard Biener <rguenther@suse.de>
32831 PR middle-end/79818
32832 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
32833 TYPE_OVERFLOW_UNDEFINED check.
32835 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32837 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
32839 (vector_ae_<mode>_p): Likewise.
32840 (vector_nez_<mode>_p): Likewise.
32841 (vector_ne_v2di_p): Likewise.
32842 (vector_ae_v2di_p): Likewise.
32843 (vector_ne_<mode>_p): Likewise.
32844 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
32846 (vsx_tsqrt<mode>2_fe): Likewise.
32848 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
32851 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
32853 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32855 PR rtl-optimization/79780
32856 * cprop.c (one_cprop_pass): When second and further conditional trap
32857 in a single basic block is turned into an unconditional trap, turn it
32858 into a deleted note to avoid RTL verification failures.
32860 2017-03-02 Richard Biener <rguenther@suse.de>
32862 * fold-const.c (const_binop): Use ulow () instead of elt (0).
32864 2017-03-02 Richard Biener <rguenther@suse.de>
32866 PR tree-optimization/79345
32868 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
32869 param and abort the walk, returning -1 if it is hit.
32870 (walk_aliased_vdefs): Take a limit param and pass it on.
32871 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
32872 defaulting to 0 and return a signed int.
32873 * tree-ssa-uninit.c (struct check_defs_data): New struct.
32874 (check_defs): New helper.
32875 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
32876 about uninitialized memory.
32877 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
32878 bogus uninitialized warning.
32879 (fixed_convert_from_real): Likewise.
32881 2017-03-02 Bin Cheng <bin.cheng@arm.com>
32883 PR tree-optimization/66768
32884 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
32885 iv_use if base object can't be determined.
32887 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32889 PR tree-optimization/79345
32890 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
32891 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
32892 (get_pattern_stats): Initialize it.
32893 * genemit.c (gen_expand): Verify match_scratch numbers come after
32894 match_operand/match_dup numbers.
32895 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
32896 match_scratch numbers.
32897 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
32899 * config/s390/s390.md (trunctdsd2): Likewise.
32901 2017-03-02 Richard Biener <rguenther@suse.de>
32903 * wide-int.h (wide_int_storage::operator=): Implement in terms
32906 2017-03-02 Richard Biener <rguenther@suse.de>
32908 PR tree-optimization/79777
32909 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
32910 the to insert expression to sth existing.
32912 2017-03-01 Martin Sebor <msebor@redhat.com>
32914 PR middle-end/79692
32915 * gimple-ssa-sprintf.c
32916 (directive::known_width_and_precision): New function.
32917 (format_integer): Use it.
32918 (get_mpfr_format_length): Consider the full range of precision
32919 when computing %g output with the # flag. Set the likely byte
32920 count to 3 rather than 1 when precision is indeterminate.
32921 (format_floating): Correct the lower bound of precision.
32923 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32925 * doc/invoke.texi: Document default code model for 64-bit Linux.
32927 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32930 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
32931 udiv rather than div since input pattern is unsigned.
32933 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
32935 * config/i386/i386.c (print_reg): Warn for values of
32936 unsupported size in integer register.
32938 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
32941 * config/rs6000/predicates.md (current_file_function_operand): Do
32942 not allow self calls to be local if the function is replaceable.
32944 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
32947 * config/rs6000/altivec.h (vec_ctz and others): Change the
32948 preprocessor macro that controls conditional compilation from
32949 _ARCH_PWR9 to __POWER9_VECTOR__.
32950 (vec_all_ne): Change parameterization of __altivec_scalar_pred
32951 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
32952 control (instead of _ARCH_PWR9 control) so that template
32953 definition uses power9-specific function.
32954 (vec_any_eq): Likewise.
32955 (vec_all_ne): Change macro definition to use a power9-specific
32956 expansion under #ifdef __POWER9_VECTOR__ control (instead of
32957 _ARCH_PWR9 control).
32958 (vec_any_eq) Likewise.
32959 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
32960 expansion for CMPNEF to remove support for xvcmpnesp instruction.
32961 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
32962 support for xvcmpnedp instruction.
32963 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
32964 macro expansion so that Power9 implementation of vec_all_ne does
32965 not use the AltiVec predicate framework.
32966 (VCMPNEH_P): Likewise.
32967 (VCMPNEW_P): Likewise.
32968 (VCMPNED_P): Likewise.
32969 (VCMPNEFP_P): Likewise.
32970 (VCMPNEDP_P): Likewise.
32971 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
32972 implementation of vec_any_eq to not use AltiVec predicate
32974 (VCMPAEH_P): Likewise.
32975 (VCMPAEW_P): Likewise.
32976 (VCMPAED_P): Likewise.
32977 (VCMPAEFP_P): Likewise.
32978 (VCMPAEDP_P): Likewise.
32979 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
32980 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
32981 not use the AltiVec predicate framework.
32982 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
32983 of vec_any_eq to not use AltiVec predicate framework.
32984 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32985 support for predefined __POWER9_VECTOR__ macro to indicate that
32986 Power9 instruction selection is enabled.
32987 (altivec_overloaded_builtins): Remove extraneous
32988 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
32989 function argument types RS6000_BTI_bool_V16QI and
32990 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
32991 entry for overloaded function argument types RS6000_BTI_bool_V4SI
32992 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
32993 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
32994 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
32995 Power9 for implementations of vec_cmpne. Change the signature for
32996 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
32997 (representing vec_all_ne) to remove the previously described first
32998 argument of type RS6000_BTI_INTSI, as this was an artifact of
32999 reliance on the AltiVec predicate framework, which is no longer
33000 used in the implementation of these functions. Add
33001 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33002 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33003 since, unlike the AltiVec predicate framework implementation, we
33004 do not share function descriptors between vec_alle and vec_anyeq.
33005 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33006 set of modes that receive special treatment even when
33007 TARGET_P9_VECTOR is true. The special treatment emits code that
33008 does not depend on Power9 instructions.
33009 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33010 define_expand to not rely on AltiVec predicate framework.
33011 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33013 (vector_ne_v2di_p): Change this define_expand to not rely on
33014 AltiVec predicate framework.
33015 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33017 (vector_ne_<mode>_p): Change this define_expand to not rely on
33018 AltiVec predicate framework.
33019 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33021 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33022 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33023 define_insn pattern.
33024 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33025 define_insn pattern because the xvcmpne<VSs>. instruction is not
33027 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33028 instruction is not supported.
33030 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33032 * config/nvptx/nvptx.c: Include intl.h.
33034 2017-03-01 Martin Jambor <mjambor@suse.cz>
33037 * ipa-prop.h (ipa_bits): Removed field known.
33038 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33039 to pointers. Adjusted their comments to warn about their sharing.
33040 (ipcp_transformation_summary): Change bits to a vector of pointers.
33041 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33042 (ipa_get_ipa_bits_for_value): Declare.
33043 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33044 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33045 (ipa_bits_hash_table): Likewise.
33046 (ipa_vr_ggc_hash_traits): Likewise.
33047 (ipa_vr_hash_table): Likewise.
33048 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33049 being pointers and vr_known being removed.
33050 (ipa_set_jf_unknown): Likewise.
33051 (ipa_get_ipa_bits_for_value): New function.
33052 (ipa_set_jfunc_bits): Likewise.
33053 (ipa_get_value_range): New overloaded functions.
33054 (ipa_set_jfunc_vr): Likewise.
33055 (ipa_compute_jump_functions_for_edge): Use the above functions to
33056 construct bits and vr parts of jump functions.
33057 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33058 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33060 (ipcp_grow_transformations_if_necessary): Also allocate
33061 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33063 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33064 them. Fix too long lines.
33065 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33066 vr_known being removed.
33067 (ipa_read_jump_function): Use new setter functions to construct bits
33068 and vr parts of jump functions or set them to NULL.
33069 (write_ipcp_transformation_info): Adjust for bits being pointers.
33070 (read_ipcp_transformation_info): Likewise.
33071 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33073 Include gt-ipa-prop.h.
33074 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33076 (ipcp_store_bits_results): Likewise.
33077 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33078 Do not write to existing jump functions but use a temporary instead.
33080 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33083 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33084 attempt to use its first operand as BIT_FIELD_REF base.
33086 2017-03-01 Richard Biener <rguenther@suse.de>
33088 PR middle-end/79721
33089 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33090 interpolating formula in wrapping arithmetic.
33091 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33093 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33095 PR tree-optimization/79734
33096 * tree-vect-generic.c (expand_vector_condition): Optimize
33097 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33098 Handle VEC_COND_EXPR where comparison has different inner width from
33099 type's inner width.
33101 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33103 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33104 markup, and similar issues. Remove @opindex entries for things
33105 that aren't options. Add missing -mmpy-option entries.
33107 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33109 PR tree-optimization/79737
33110 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33111 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33112 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33113 instead of byte_size. Formatting fix.
33114 (shift_bytes_in_array_right): Formatting fix.
33116 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33119 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33120 condition on optimize for the leaf function test.
33122 2017-02-28 Martin Liska <mliska@suse.cz>
33125 * read-rtl-function.c (function_reader::handle_unknown_directive):
33126 Bail out when one uses -flto.
33128 2017-02-28 Martin Liska <mliska@suse.cz>
33130 * common.opt: Replace space with tabular for options of <number>
33132 * config/i386/i386.opt: Show <number> value for
33133 -mlarge-data-threshold.
33134 * opts.c (print_filtered_help): Do not display number in hexadecimal
33137 2017-02-28 Martin Liska <mliska@suse.cz>
33139 * common.opt: Fix --help=option -Q for options which are of
33142 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33144 * config/i386/i386.c (print_reg): Error out for values
33145 of 8-bit size in invalid integer register.
33147 2017-02-28 Martin Sebor <msebor@redhat.com>
33149 PR tree-optimization/79691
33150 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33152 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33155 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33156 gcc_unreachable with output_operand_lossage.
33158 2017-02-28 Richard Biener <rguenther@suse.de>
33160 PR tree-optimization/79740
33161 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33163 (visit_nary_op): Insert the nary into the hashtable if we
33164 pattern-matched sth.
33165 * tree-ssa-pre.c (eliminate_insert): Robustify.
33167 2017-02-28 Richard Biener <rguenther@suse.de>
33169 PR middle-end/79731
33170 * fold-const.c (decode_field_reference): Reject out-of-bound
33173 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33175 * config/i386/i386.c: Include intl.h.
33176 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33177 instead of just cond ? "..." : "...".
33178 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33179 * coverage.c (read_counts_file): Likewise.
33180 * omp-offload.c: Include intl.h.
33181 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33182 of just cond ? "..." : "...".
33183 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33184 of just cond ? "..." : "...".
33186 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33189 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33191 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33192 'tune for' CPU name.
33193 * config/arm/arm-cpu-data.h: Regenerated.
33195 2017-02-28 Richard Biener <rguenther@suse.de>
33197 PR tree-optimization/79732
33198 * tree-inline.c (expand_call_inline): Do not shadow var.
33200 2017-02-28 Richard Biener <rguenther@suse.de>
33202 PR tree-optimization/79723
33203 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33204 address-space properly.
33206 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33208 * doc/optinfo.texi (Optimization groups): Fix option used for
33210 * doc/invoke.texi (-fopt-info): Document "omp".
33211 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
33212 (OPTGROUP_ALL): Add OPTGROUP_OMP.
33213 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
33214 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
33215 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
33217 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
33219 * dumpfile.c (optgroup_options): Instead of "openmp", associate
33220 OPTGROUP_OMP with "omp".
33222 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
33225 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
33226 for arithmetic shift of unsigned V2DI.
33228 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
33230 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
33231 arc/linux.h headers.
33232 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
33233 (LINK_SPEC): Likewise.
33234 (ARC_TLS_EXTRA_START_SPEC): Likewise.
33235 (EXTRA_SPECS): Likewise.
33236 (STARTFILE_SPEC): Likewise.
33237 (ENDFILE_SPEC): Likewise.
33238 (LIB_SPEC): Likewise.
33239 (TARGET_SDATA_DEFAULT): Likewise.
33240 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
33241 (MULTILIB_DEFAULTS): Likewise.
33242 (DWARF2_UNWIND_INFO): Likewise.
33243 * config/arc/big.h: New file.
33244 * config/arc/elf.h: Likewise.
33245 * config/arc/linux.h: Likewise.
33246 * config/arc/t-uClibc: Remove.
33248 2017-02-27 Bin Cheng <bin.cheng@arm.com>
33250 PR tree-optimization/77536
33251 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
33252 (tree_transform_and_unroll_loop): Use above function to compute the
33253 estimated niter of unrolled loop and use it when scaling profile.
33254 Also use count info rather than frequency if it's non-zero.
33255 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
33256 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
33257 (vect_transform_loop): Call above function.
33259 2017-02-27 Richard Biener <rguenther@suse.de>
33261 PR tree-optimization/45397
33262 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
33263 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
33264 (visit_nary_op): Add pattern matching for CSEing sign-changed
33265 or truncated operations with wider ones.
33267 2017-02-27 Richard Biener <rguenther@suse.de>
33269 PR tree-optimization/79690
33270 * tree-vect-stmts.c (vectorizable_store): Use vector type
33271 built from the DR with address-space.
33273 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
33275 * doc/invoke.texi (Optimize Options): Refine the description
33276 of asan-use-after-return.
33278 2017-02-25 Alan Modra <amodra@gmail.com>
33280 PR rtl-optimization/79584
33281 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
33282 base, not ad->base_term, the reg within base. Remove assertion
33283 that ad->base == ad->base_term. Replace gen_int_mode using
33284 bogus mode with const0_rtx.
33286 2017-02-25 Jakub Jelinek <jakub@redhat.com>
33288 PR middle-end/79396
33289 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
33290 FMA_EXPR like tcc_binary or tcc_unary.
33292 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
33295 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
33297 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
33298 (output_loc_operands): Handle DW_OP_call_ref and
33299 DW_OP_GNU_variable_value.
33300 (struct variable_value_struct): New type.
33301 (struct variable_value_hasher): Likewise.
33302 (variable_value_hash): New variable.
33303 (string_types): Remove.
33304 (copy_loc_descr): New function.
33305 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
33306 (prepend_loc_descr_to_each): New function.
33307 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
33308 instead of add_loc_descr_to_each if the first argument is single
33309 location list and the second has multiple.
33310 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
33311 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
33312 when looking for variable value which doesn't have other location info.
33313 (loc_list_from_tree): Formatting fix.
33314 (gen_array_type_die): Simplify DW_AT_string_length handling.
33315 (adjust_string_types): Remove.
33316 (gen_subprogram_die): Don't call adjust_string_types nor test/set
33317 string_types. Call resolve_variable_values.
33318 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
33319 (resolve_addr_in_expr): Likewise. Add A argument.
33320 (copy_deref_exprloc): Remove deref argument. Adjust for the
33321 original expression being DW_OP_GNU_variable_value with optionally
33322 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
33323 optionally after it.
33324 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
33325 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
33326 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
33327 (variable_value_hasher::hash, variable_value_hasher::equal): New
33329 (resolve_variable_value_in_expr, resolve_variable_value,
33330 resolve_variable_values, note_variable_value_in_expr,
33331 note_variable_value): New functions.
33332 (dwarf2out_early_finish): Call note_variable_value on all toplevel
33335 2017-02-24 Jakub Jelinek <jakub@redhat.com>
33338 * opts.h (handle_generated_option): Add GENERATED_P argument.
33339 * opts-common.c (handle_option): Adjust function comment.
33340 (handle_generated_option): Add GENERATED_P argument, pass it to
33342 (control_warning_option): Pass false to handle_generated_option
33344 * opts.c (maybe_default_option): Pass true to handle_generated_option
33346 * optc-gen.awk: Likewise.
33348 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33350 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
33351 a REG, look at the REG it is a SUBREG of.
33352 (splitter for cmpeqsi_t): Ditto.
33354 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33356 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
33357 the special USEs with the pattern of the insn, not the insn itself.
33359 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
33362 * doc/invoke.texi: Document -mload-store-pairs.
33364 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33365 Sandra Loosemore <sandra@codesourcery.com>
33367 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
33368 argument isn't a CONST_INT.
33369 (nios2_alternate_compare_const): Assert op is a CONST_INT.
33370 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
33371 (nios2_validate_compare): Bypass alternate compare logic if *op2
33372 is not a CONST_INT.
33373 (ldstwm_operation_p): Return false if first_base is not a REG or
33374 if first_offset is not a CONST_INT.
33376 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33378 * config/cris/cris.md: Use correct operand in a define_peephole2.
33380 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33382 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
33384 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33386 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
33387 this_insn if it is an INSN or JUMP_INSN.
33388 (force_offsettable): Look at base, not at addr.
33389 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
33390 on things that aren't necessarily CONST_INTs.
33392 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
33394 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
33395 -mfpmath=sse is the default also for x86-32 targets with SSE2
33396 instruction set when @option{-ffast-math} is enabled
33398 2017-02-24 Jeff Law <law@redhat.com>
33400 PR rtl-optimizatoin/79286
33401 * ira.c (update_equiv_regs): Drop may_trap_p exception to
33404 2017-02-24 Richard Biener <rguenther@suse.de>
33406 PR tree-optimization/79389
33407 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
33410 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
33412 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
33413 function comment to reflect reality.
33414 (loop_exits_before_overflow): Fix typo in function description.
33416 2017-02-24 Richard Biener <rguenther@suse.de>
33418 PR tree-optimization/79389
33419 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
33420 properly that a threading opportunity exists. Detect conditional
33421 copy/constant propagation opportunities.
33423 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
33425 * config/visium/visium.md (type): Add trap.
33426 (b): New mode attribute.
33427 (*btst): Rename into...
33428 (*btst<mode>): ...this and adjust.
33429 (*cbranchsi4_btst_insn): Rename into...
33430 (*cbranch<mode>4_btst_insn): ...this and adjust.
33431 (trap): New define_insn.
33433 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33435 PR tree-optimization/79389
33436 * ifcvt.c (struct noce_if_info): Add rev_cond field.
33437 (noce_reversed_cond_code): New function.
33438 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
33439 reversed_comparison_code. Formatting fix.
33440 (noce_try_store_flag): Test rev_cond != NULL in addition to
33441 reversed_comparison_code.
33442 (noce_try_store_flag_constants): Likewise.
33443 (noce_try_store_flag_mask): Likewise.
33444 (noce_try_addcc): Use rev_cond if non-NULL instead of
33445 reversed_comparison_code.
33446 (noce_try_cmove_arith): Likewise. Formatting fixes.
33447 (noce_try_minmax, noce_try_abs): Clear rev_cond.
33448 (noce_find_if_block): Initialize rev_cond.
33449 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
33450 instead of false as last argument never attempt to reverse it
33453 2017-02-23 Bin Cheng <bin.cheng@arm.com>
33455 PR tree-optimization/79663
33456 * tree-predcom.c (combine_chains): Process refs in reverse order
33457 only for ZERO length chains, and add explaining comment.
33459 2017-02-23 Jeff Law <law@redhat.com>
33461 PR tree-optimization/79578
33462 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
33463 in call to operand_equal_p.
33465 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
33468 * config/i386/cpuid.h: Fix another undefined behavior.
33470 2017-02-23 Richard Biener <rguenther@suse.de>
33472 PR tree-optimization/79683
33473 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
33474 vector types for data-refs.
33476 2017-02-23 Martin Liska <mliska@suse.cz>
33478 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
33480 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33482 PR middle-end/79665
33483 * internal-fn.c (get_range_pos_neg): Moved to ...
33484 * tree.c (get_range_pos_neg): ... here. No longer static.
33485 * tree.h (get_range_pos_neg): New prototype.
33486 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
33487 are known to be in between 0 and signed maximum inclusive, try to
33488 expand both unsigned and signed divmod and use the cheaper one from
33491 2017-02-22 Jeff Law <law@redhat.com>
33493 PR tree-optimization/79578
33494 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
33495 to compare base operands.
33497 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
33500 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
33501 gpc_reg_operand instead of fpr_reg_operand.
33503 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
33505 * config/mips/mips.c (mips_return_in_memory): Force FP
33506 vector types to be returned in memory for o32 ABI.
33508 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33510 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
33511 instead of DW_TAG_member for static data member declarations and don't
33512 set no_linkage_name for static inline data members.
33513 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
33516 2017-02-22 Martin Liska <mliska@suse.cz>
33518 * doc/invoke.texi: Replace inequality signs with square brackets
33521 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
33524 * lra-constraints.c (simplify_operand_subreg): Handle
33525 WORD_REGISTER_OPERATIONS targets.
33527 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33530 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
33531 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
33532 elimination by swapping fld*.
33534 2017-02-22 Richard Biener <rguenther@suse.de>
33536 PR tree-optimization/79673
33537 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
33538 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
33539 irrelevant address-space qualifiers and avoiding a
33540 ADDR_SPACE_CONVERT_EXPR from fold_convert.
33542 2017-02-22 Richard Biener <rguenther@suse.de>
33544 PR tree-optimization/79666
33545 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
33546 to not symbolically negate if that may introduce undefined
33549 2017-02-22 Martin Liska <mliska@suse.cz>
33552 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
33553 * data-streamer-out.c (streamer_write_gcov_count_stream):
33555 * value-prof.c (stream_out_histogram_value): Make assert more
33556 precise based on type of counter.
33558 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
33561 * config/i386/i386.md (standard_x87sse_constant_load splitter):
33562 Use nonimmediate_operand instead of memory_operand for operand 1.
33563 (float-extend standard_x87sse_constant_load splitter): Ditto.
33565 2017-02-21 Jeff Law <law@redhat.com>
33567 PR tree-optimization/79621
33568 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
33569 blocks with edges to themselves.
33571 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33574 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
33575 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
33576 Use gimple_call_builtin_p.
33579 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
33580 on temporarily removed DEBUG_INSNs.
33582 PR tree-optimization/79649
33583 * tree-loop-distribution.c (classify_partition): Give up on
33584 non-generic address space loads/stores.
33586 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
33588 * doc/loop.texi (Loop manipulation): Remove nonexistent
33589 tree_ssa_loop_version from the documentation.
33590 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
33592 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33595 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
33596 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
33597 * config/rs6000/rs6000.c: Include except.h.
33598 (rs6000_expand_split_stack_prologue): Call
33599 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
33601 2017-02-21 Martin Jambor <mjambor@suse.cz>
33604 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
33605 have been analyzed.
33607 2017-02-21 Martin Jambor <mjambor@suse.cz>
33609 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
33610 for backward compatibility only.
33611 * doc/invoke.texi (Option Summary): Remove all references to
33612 -fipa-cp-alignment.
33614 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
33618 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33620 * lra-constraints.c (curr_insn_transform): Handle
33621 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33623 2017-02-21 Martin Liska <mliska@suse.cz>
33625 * config/i386/i386.opt: Replace -masm-dialect with -masm.
33627 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
33629 PR translation/79638
33630 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
33632 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
33635 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
33636 (arm_function_ok_for_sibcall): Return false for an indirect call by
33637 descriptor if all the argument registers are used.
33638 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
33639 alignment of the function.
33641 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33643 PR tree-optimization/61441
33644 * simplify-rtx.c (simplify_const_unary_operation): For
33645 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
33646 the sNaN unmodified.
33648 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33650 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33651 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33652 instead of SYSTEM_HEADER_DIR.
33654 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
33655 Martin Liška <mliska@suse.cz>
33657 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
33658 Fix typos and grammar, use active voice, and clarify.
33660 2017-02-20 Marek Polacek <polacek@redhat.com>
33662 PR middle-end/79537
33663 * gimplify.c (gimplify_expr): Handle unused *&&L;.
33666 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
33668 2017-02-20 Jakub Jelinek <jakub@redhat.com>
33671 * config/i386/i386.c (ix86_expand_builtin): Handle
33672 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
33673 ix86_builtins_isa[fcode].isa as a requirement of those
33674 flags and any other flag in the bitmask.
33675 (ix86_init_mmx_sse_builtins): Use 0 instead of
33676 ~OPTION_MASK_ISA_64BIT as mask.
33677 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
33678 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
33679 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
33680 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
33682 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33685 * lra-constraints.c (split_reg): Check requested split mode
33686 is supported by the register.
33688 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33690 * lra-constraints.c (simplify_operand_subreg): Remove early
33693 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33696 * lra-constraints.c (curr_insn_transform): Tighten condition
33697 for converting SUBREG reloads from OP_OUT to OP_INOUT.
33699 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33702 * lra-constraints.c (curr_insn_transform): Handle
33703 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33705 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
33708 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
33710 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
33712 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
33715 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
33718 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33720 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
33723 2017-02-18 Jakub Jelinek <jakub@redhat.com>
33725 * final.c (last_columnnum, override_columnnum): New variables.
33726 (final_start_function): Set last_columnnum, pass it to begin_prologue
33727 hook and pass 0 to dwarf2out_begin_prologue.
33728 (final_scan_insn): Update override_columnnum. Pass last_columnnum
33729 to source_line debug hook.
33730 (notice_source_line): Compute last_columnnum and for debug_column_info
33731 return true on column changes.
33732 * debug.h (struct gcc_debug_hooks): Add column argument to
33733 source_line and begin_prologue hooks.
33734 (debug_nothing_int_charstar_int_bool): Remove prototype.
33735 (debug_nothing_int_int_charstar,
33736 debug_nothing_int_int_charstar_int_bool): New prototypes.
33737 (dwarf2out_begin_prologue): Add column argument.
33738 * debug.c (do_nothing_debug_hooks): Adjust source_line and
33739 begin_prologue hooks.
33740 (debug_nothing_int_charstar_int_bool): Remove.
33741 (debug_nothing_int_int_charstar,
33742 debug_nothing_int_int_charstar_int_bool): New functions.
33743 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
33744 through to dwarf2out_source_line.
33745 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
33746 (dwarf2out_source_line): Add column argument, emit it if requested.
33747 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
33749 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33750 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33751 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
33752 through to dwarf2out_begin_prologue.
33753 (vmsdbgout_source_line): Add column argument, pass it through to
33754 dwarf2out_source_line.
33755 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
33756 dbxout_source_line caller.
33757 (dbxout_source_line): Add column argument.
33759 * common.opt (gno-column-info, gcolumn-info): New options.
33760 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
33761 (check_die): Also test for multiple DW_AT_decl_column attributes.
33762 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
33763 DW_AT_decl_column if requested.
33764 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
33766 (gen_variable_die): Likewise.
33767 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
33768 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
33771 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
33772 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
33773 (ix86_handle_option): Handle OPT_m3dnowa.
33774 * doc/invoke.texi (-m3dnowa): Document.
33775 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
33776 -m3dnowa instead of -m3dnow -march=athlon.
33779 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
33780 instead of gcc_assert for K, r and R code checks. Formatting fixes.
33782 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33785 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
33786 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
33787 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
33788 generator for vsx_xxpermdi_<mode>_be.
33789 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
33790 force big-endian semantics.
33791 (vsx_xxpermdi_<mode>_be): New define_expand with same
33792 implementation as previous version of vsx_xxpermdi_<mode>.
33794 2017-02-17 Jakub Jelinek <jakub@redhat.com>
33796 PR tree-optimization/79327
33797 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
33798 variable, its initialization and use.
33800 2017-02-17 Julia Koval <julia.koval@intel.com>
33802 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
33803 (OPTION_MASK_ISA_PKU_UNSET): New.
33804 (ix86_handle_option): Handle -mrdpid.
33805 * config/i386/cpuid.h (bit_RDPID): New.
33806 * config/i386/driver-i386.c (host_detect_local_cpu):
33807 Detect RDPID feature.
33808 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
33809 * config/i386/i386-c.c (ix86_target_macros_internal):
33811 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
33812 (ix86_valid_target_attribute_inner_p): Add "rdpid".
33813 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
33814 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
33815 * config/i386/i386.md (define_insn "rdpid"): New.
33816 * config/i386/i386.opt Add -mrdpid.
33817 * config/i386/immintrin.h (_rdpid_u32): New.
33819 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
33821 PR rtl-optimization/79541
33822 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
33823 instead of transforming it into USE.
33825 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
33827 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
33828 If HONOR_SNANS (SFmode) force the input to a register.
33829 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
33830 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
33831 an frsp or similar insn.
33833 2017-02-17 Martin Liska <mliska@suse.cz>
33835 PR rtl-optimization/79577
33836 * params.def (selsched-max-sched-times): Increase minimum to 1.
33838 2017-02-17 Martin Liska <mliska@suse.cz>
33840 PR rtl-optimization/79574
33841 * gcse.c (want_to_gcse_p): Prevent integer overflow.
33843 2017-02-17 Martin Liska <mliska@suse.cz>
33845 PR tree-optimization/79529
33846 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
33847 ssa_defined_default_def_p to handle cases which are implicitly
33849 * tree-ssa.c (ssa_defined_default_def_p): New function.
33850 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
33851 which are implicitly defined.
33852 * tree-ssa.h (ssa_defined_default_def_p): Declare.
33854 2017-02-17 Richard Biener <rguenther@suse.de>
33856 PR middle-end/79576
33857 * params.def (max-ssa-name-query-depth): Limit to 10.
33859 2017-02-17 Richard Biener <rguenther@suse.de>
33861 PR tree-optimization/79552
33862 * tree-ssa-structalias.c (visit_loadstore): Properly verify
33865 2017-02-17 Richard Biener <rguenther@suse.de>
33868 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
33870 2017-02-17 Marek Polacek <polacek@redhat.com>
33872 PR middle-end/79536
33873 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
33874 (fold_negate_expr): New wrapper.
33876 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
33878 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
33879 Correct terminology and de-emphasize pre-standard behavior.
33881 2017-02-16 Alan Modra <amodra@gmail.com>
33883 PR rtl-optimization/79286
33884 * ira.c (def_dominates_uses): New function.
33885 (update_equiv_regs): Don't create an equivalence for insns that
33886 may trap where the register def does not dominate the use.
33888 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
33890 PR rtl-optimization/78127
33891 * lra.c (lra): Call lra_eliminate before finish the loop after
33894 2017-02-16 Richard Biener <rguenther@suse.de>
33896 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
33898 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
33899 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
33900 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
33901 (isl_val_int_from_wi): New function.
33902 (extract_affine_gmp): Rename to ...
33903 (extract_affine_wi): ... this, take a widest_int.
33904 (extract_affine_int): Just wrap extract_affine_wi.
33905 (add_param_constraints): Use isl_val_int_from_wi.
33906 (add_loop_constraints): Likewise, and extract_affine_wi.
33908 2017-02-15 Jeff Law <law@redhat.com>
33910 PR middle-end/79521
33911 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
33912 ira_init_register_move_cost_if_necessary.
33914 2017-02-15 Martin Sebor <msebor@redhat.com>
33916 PR middle-end/32003
33917 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
33918 removed in a prior commit.
33920 2017-02-15 Bin Cheng <bin.cheng@arm.com>
33922 PR tree-optimization/79347
33923 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
33924 counters during peeling.
33926 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
33928 * Makefile.in (site.exp): Remove "set ISLVER".
33930 2017-02-15 Jakub Jelinek <jakub@redhat.com>
33933 * real.c (real_from_integer): Call real_convert even for decimal.
33935 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
33938 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
33940 2017-02-14 Andrew Pinski <apinski@cavium.com>
33942 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
33943 cores and change the partno/implementer to be correct.
33944 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
33945 the 'B" as the implementer.
33946 * config/aarch64/aarch64-tune.md: Regenerate.
33948 2017-02-14 Carl Love <cel@us.ibm.com>
33950 * config/rs6000/rs6000.c: Add case statement entry to make the
33951 xvcvuxdsp built-in argument unsigned.
33952 * config/rs6000/vsx.md: Fix the source and return operand types so they
33953 match the instruction definitions from the ISA document. Fix typo
33954 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
33957 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
33960 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
33961 member early_clobber_alts.
33962 * lra-lives.c (reg_early_clobber_p): New.
33963 (process_bb_lives): Use it.
33964 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
33965 (debug_operand_data): Initialize early_clobber_alts.
33966 (setup_operand_alternative): Set up early_clobber_alts.
33967 (collect_non_operand_hard_regs): Ditto. Pass early clobber
33968 alternatives to new_insn_reg.
33969 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
33971 (lra_update_insn_regno_info): Pass the new arg.
33973 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33975 PR middle-end/79505
33976 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
33977 (new_oacc_loop_raw): Don't clear already cleared fields.
33980 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
33981 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
33982 _mm512_prefetch_i64gather_ps): New inline functions and macros.
33984 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
33987 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
33989 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
33992 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
33993 the extra instruction to the right place to store 128-bit constant
33996 2017-02-14 Martin Sebor <msebor@redhat.com>
33998 PR middle-end/79448
33999 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34000 warning for strings of unknown length.
34002 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34004 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34006 2017-02-14 Jeff Law <law@redhat.com>
34009 * ira-costs.c (scan_one_insn): Initialize register move costs
34010 for pseudos seen in USE/CLOBBER insns.
34012 PR tree-optimization/79095
34013 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34014 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34015 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34016 if the operands are known to be not equal, then the resulting range
34018 (intersect_ranges): If the new range is ~[0,0] and the old range is
34019 wide, then prefer ~[0,0].
34020 * tree-vrp.c (overflow_comparison_p_1): New function.
34021 (overflow_comparison_p): New function.
34022 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34023 if NAME is used in an overflow test.
34024 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34025 overflow check that can be expressed as an equality test, then adjust
34026 ops to be that equality test.
34028 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34030 * config/s390/s390-builtin-types.def: Remove flags argument.
34031 * config/s390/s390.c (s390_init_builtins): Likewise.
34033 2017-02-14 Martin Liska <mliska@suse.cz>
34035 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34036 vector. Fix trailing white spaces.
34038 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34040 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34043 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34045 PR rtl-optimization/68664
34046 * config/arm/arm.c (arm_sched_can_speculate_insn):
34047 New function. Declare prototype.
34048 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34050 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34052 PR rtl-optimization/68664
34053 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34055 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34057 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34059 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34060 max skip bytes for function, loop and jump.
34062 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34064 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34065 ABS_EXPR for gimple dump.
34067 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34070 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34072 PR tree-optimization/79408
34073 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34074 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34075 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34076 also if rhs1 is INTEGER_CST.
34078 2017-02-14 Richard Biener <rguenther@suse.de>
34080 PR middle-end/79432
34081 * tree-into-ssa.c (insert_phi_nodes): When the function can
34082 have abnormal edges rewrite SSA names with broken use-def
34083 dominance out of SSA and register them for PHI insertion.
34085 2017-02-13 Martin Sebor <msebor@redhat.com>
34087 PR middle-end/79496
34088 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34089 clearing info.nowrite flag when snprintf size argument is a range.
34091 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34093 * cprop.c (cprop_jump): Add missing space in string literal.
34094 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34095 (get_constraint_for_component_ref): Likewise.
34096 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34097 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34098 * lra-constraints.c (process_alt_operands): Likewise.
34099 * ipa-inline.c (inline_small_functions): Likewise.
34100 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34101 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34102 * trans-mem.c (diagnose_tm_1_op): Likewise.
34103 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34104 (grid_parallel_clauses_gridifiable): Likewise.
34106 * config/nvptx/mkoffload.c (process): Add space in between
34109 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34110 "MOD4_SSE_REGS" and "ALL_REGS".
34112 * spellcheck.c (test_data): Add , in between "foo" and "food".
34114 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34117 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34118 boundary crossing check and subsequent code generation agree.
34120 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34122 * config/aarch64/aarch64.c (has_memory_op): Delete.
34123 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34126 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34128 PR rtl-optimization/79388
34129 PR rtl-optimization/79450
34130 * combine.c (distribute_notes): When removing TEM_INSN for which
34131 corresponding dest has last value recorded, invalidate that last
34134 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34136 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34137 of explicit '@'. Add missing assembly comment marker on branch costs
34140 2017-02-13 Nathan Sidwell <nathan@acm.org>
34142 * gengtype-lex.l (<in_struct>): Add '/'.
34144 2017-02-13 Martin Liska <mliska@suse.cz>
34147 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34149 2017-02-13 Richard Biener <rguenther@suse.de>
34151 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34153 * configure: Re-generate.
34154 * config.in: Likewise.
34155 * graphite-dependences.c: Simplify as if
34156 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34157 * graphite-isl-ast-to-gimple.c: Likewise.
34158 * graphite-optimize-isl.c: Likewise.
34159 * graphite-poly.c: Likewise.
34160 * graphite-sese-to-poly.c: Likewise.
34161 * graphite.h: Likewise.
34162 * toplev.c: Include isl/version.h and use isl_version () for
34163 printing the ISL version.
34164 * doc/install.texi: Update ISL requirement.
34166 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34168 * doc/standards.texi (Standards): Update reference to
34171 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34173 * doc/extend.texi (Named Address Spaces): sourceware.org now
34175 * doc/install.texi (Binaries): Ditto.
34178 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34180 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34181 standard terminology "stringize"/"stringizing" throughout.
34182 * doc/cppinternals.texi: Likewise.
34184 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34186 * doc/extend.texi: Fix some spelling mistakes and typos.
34187 * doc/invoke.texi: Likewise.
34189 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34192 * params.def (inline-min-speedup) Change from 10 to 8.
34194 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34196 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34199 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34202 * ipa-inline-analysis.c (get_minimal_bb): New function.
34203 (record_modified): Use it.
34204 (remap_edge_change_prob): Handle also ancestor functions.
34206 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34208 * doc/contrib.texi (Contributors): Remove broken link into
34209 the Mauve CVS repository.
34211 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34213 PR middle-end/79454
34214 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
34215 result computation whenever lhs doesn't have vector mode, not
34216 just when it has BLKmode.
34218 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34220 * doc/makefile.texi (profiledbootstrap): Refer to the
34221 installation instructions only in textual form.
34223 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34226 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
34228 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34230 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
34231 (Specific): Update mingw-w64 reference.
34233 (Specific): Remove broken link to Renesas RX processor.
34235 2017-02-10 Richard Biener <rguenther@suse.de>
34237 * toplev.c (process_options): Do not mention obsolete graphite
34238 options when printing sorry message about missing graphite support.
34239 Mention -floop-nest-optimize.
34241 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
34243 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
34244 (vtst_p16): Likewise.
34245 (vtstq_p8): Likewise.
34246 (vtstq_p16): Likewise.
34248 (vtstq_p64): Likewise.
34249 * config/arm/arm_neon.h (vgetq_lane_p64): New.
34250 (vset_lane_p64): New.
34251 (vsetq_lane_p64): New.
34253 2017-02-10 Jakub Jelinek <jakub@redhat.com>
34255 PR tree-optimization/79411
34256 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
34257 stmt operands are SSA_NAMEs used in abnormal phis.
34258 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
34261 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34264 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
34267 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34269 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
34271 2017-02-09 Jakub Jelinek <jakub@redhat.com>
34273 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
34277 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
34278 not arbitrary TREE_CONSTANT.
34281 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
34282 "omp declare target link" attribute unless is_global_var.
34283 * omp-offload.c (find_link_var_op): Likewise.
34285 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
34286 Chung-Lin Tang <cltang@codesourcery.com>
34288 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
34290 (gimplify_adjust_omp_clauses): Don't delete TILE.
34291 (gimplify_omp_for): Deal with TILE.
34292 * internal-fn.c (expand_GOACC_TILE): New function.
34293 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
34295 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
34296 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
34298 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
34299 avoid DIV for outermost collapse var.
34300 (expand_oacc_for): Insert tile element loop as needed. Adjust.
34301 Remove out of date comments, fix whitespace.
34302 * omp-general.c (omp_extract_for_data): Deal with tiling.
34303 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
34304 adjust OLF_DIM_BASE value.
34305 (struct omp_for_data): Add tiling field.
34306 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
34307 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
34308 for auto loops. Remove default auto determining, moved to
34309 oacc_loop_fixed_partitions.
34310 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
34311 stmts, add e_mask field.
34312 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
34313 (oacc_thread_numbers): Use oacc_dim_call.
34314 (oacc_xform_tile): New.
34315 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
34316 (finish_oacc_loop): Adjust for ifns vector.
34317 (oacc_loop_discover_walk): Append loop abstraction sites to list,
34318 add case for GOACC_TILE fns.
34319 (oacc_loop_xform_loop): Delete.
34320 (oacc_loop_process): Iterate over call list directly, and add
34321 handling for GOACC_TILE fns.
34322 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
34324 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
34325 vector partitioning to outer loops. Assign 2 partitions to loops
34326 when available. Add TILE handling.
34327 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
34328 (execite_oacc_device_lower): Process GOACC_TILE fns,
34329 ignore unknown specs.
34330 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
34331 * tree.c (omp_clause_num_ops): Adjust TILE ops.
34332 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
34334 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
34336 * configure.ac (ACX_BUGURL): Update.
34337 * configure: Regenerate.
34339 2017-02-09 Richard Biener <rguenther@suse.de>
34341 PR tree-optimization/69823
34342 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
34343 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
34345 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34347 * config/arc/arc-c.def: Add __NPS400__ definition.
34348 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
34349 (TARGET_NPS400): Define.
34351 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34353 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
34355 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
34356 pointer, arch_info.
34357 (arc_cpu_types): Fill the arch_info field with a pointer into the
34358 arc_arch_types table.
34359 (arc_selected_cpu): Declare.
34360 * config/arc/arc.c (arc_selected_cpu): Make global.
34361 (arc_selected_arch): Delete.
34362 (arc_base_cpu): Delete.
34363 (arc_override_options): Remove references to deleted variables,
34364 update access to arch information.
34365 (ARC_OPT): Update access to arch information.
34366 (ARC_OPTX): Likewise.
34367 * config/arc/arc.h (arc_base_cpu): Remove declaration.
34368 (TARGET_ARC600): Update access to arch information.
34369 (TARGET_ARC601): Likewise.
34370 (TARGET_ARC700): Likewise.
34371 (TARGET_EM): Likewise.
34372 (TARGET_HS): Likewise.
34373 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
34376 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
34379 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
34380 condition/operands for integer GE/LE/GEU/LEU operations.
34382 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
34384 PR translation/79397
34385 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
34388 2017-02-08 Martin Jambor <mjambor@suse.cz>
34391 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
34392 whether allocation happened.
34393 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
34394 nothing was allocated.
34396 2017-02-08 Jakub Jelinek <jakub@redhat.com>
34398 PR tree-optimization/79408
34399 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
34400 constant, but SSA_NAME with a known integer range, use the minimum
34401 of that range instead of op1 to determine if modulo can be replaced
34402 with its first operand.
34404 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34406 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
34408 2017-02-08 Richard Biener <rguenther@suse.de>
34410 PR tree-optimization/71824
34411 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34412 Check all loops contained in the merged region.
34414 2017-02-07 Andrew Pinski <apinski@cavium.com>
34416 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
34418 2017-02-07 Andrew Pinski <apinski@cavium.com>
34420 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
34421 (thunderxt88): Likewise.
34422 (thunderxt81): Disable LSE and change v8.1 to v8.
34423 (thunderxt83): Likewise.
34425 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34426 Richard Biener <rguenther@suse.de>
34428 PR middle-end/79399
34429 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
34430 type from int to size_t.
34431 * ira-costs.c (struct_costs_size): Change type from int to size_t.
34433 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34435 PR rtl-optimization/79386
34436 * cprop.c (bypass_conditional_jumps): Initialize
34437 bypass_last_basic_block already before splitting bbs after
34438 unconditional traps...
34439 (bypass_conditional_jumps): ... rather than here.
34442 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
34443 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
34444 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
34445 fix -masm=intel patterns.
34447 2017-02-07 Richard Biener <rguenther@suse.de>
34449 PR tree-optimization/79256
34450 PR middle-end/79278
34451 * builtins.c (get_object_alignment_2): Use min_align_of_type
34452 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
34453 and ADJUST_FIELD_ALIGN.
34455 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
34457 * doc/tm.texi: Regenerate.
34458 * stor-layout.c (layout_decl): Adjust.
34459 (update_alignment_for_field): Likewise.
34460 (place_field): Likewise.
34461 (min_align_of_type): Likewise.
34462 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
34463 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
34464 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
34465 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
34466 * config/frv/frv.c (frv_adjust_field_align): Likewise.
34467 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
34468 * config/i386/i386.c (x86_field_alignment): Likewise.
34469 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
34470 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
34471 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
34472 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
34473 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
34474 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
34478 2017-01-30 Richard Biener <rguenther@suse.de>
34480 PR tree-optimization/79256
34481 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34482 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34485 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
34487 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
34488 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
34489 builtins to SImode and emit a zero-extend, if necessary.
34491 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34493 * docs/invoke.texi (RISC-V Options): Alphabetize.
34495 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34497 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
34500 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34502 * config/riscv/riscv.c: New file.
34503 * common/config/riscv/riscv-common.c: Likewise.
34504 * config.gcc: Likewise.
34505 * config/riscv/constraints.md: Likewise.
34506 * config/riscv/elf.h: Likewise.
34507 * config/riscv/generic.md: Likewise.
34508 * config/riscv/linux.h: Likewise.
34509 * config/riscv/multilib-generator: Likewise.
34510 * config/riscv/peephole.md: Likewise.
34511 * config/riscv/pic.md: Likewise.
34512 * config/riscv/predicates.md: Likewise.
34513 * config/riscv/riscv-builtins.c: Likewise.
34514 * config/riscv/riscv-c.c: Likewise.
34515 * config/riscv/riscv-ftypes.def: Likewise.
34516 * config/riscv/riscv-modes.def: Likewise.
34517 * config/riscv/riscv-opts.h: Likewise.
34518 * config/riscv/riscv-protos.h: Likewise.
34519 * config/riscv/riscv.h: Likewise.
34520 * config/riscv/riscv.md: Likewise.
34521 * config/riscv/riscv.opt: Likewise.
34522 * config/riscv/sync.md: Likewise.
34523 * config/riscv/t-elf-multilib: Likewise.
34524 * config/riscv/t-linux: Likewise.
34525 * config/riscv/t-linux-multilib: Likewise.
34526 * config/riscv/t-riscv: Likewise.
34527 * configure.ac: Likewise.
34528 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
34529 Waterman as RISC-V maintainers.
34530 * doc/install.texi: Add RISC-V entries.
34531 * doc/invoke.texi: Add RISC-V options section.
34532 * doc/md.texi: Add RISC-V constraints section.
34533 * configure: Regenerated.
34535 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
34538 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
34539 false values to be constant vectors with all 0 or all 1 bits set.
34540 (vcondu<mode><mode>): Likewise.
34541 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
34543 (fpmask_comparison_operator): Update comment.
34544 (vecint_comparison_operator): New predicate.
34545 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
34546 vector conditionals when the true and false values are constant
34547 vectors with all 0 bits or all 1 bits set.
34549 2017-02-06 Martin Sebor <msebor@redhat.com>
34551 PR tree-optimization/79376
34552 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
34554 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
34556 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
34557 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
34558 to simplify split condition.
34560 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34562 * omp-expand.c (oxpand_omp_atomic_fetch_op,
34563 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
34566 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
34568 PR rtl-optimization/68664
34569 * target.def (can_speculate_insn): New hook.
34570 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
34571 * doc/tm.texi: Regenerate.
34572 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
34573 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
34574 (rs6000_sched_can_speculate_insn): New function.
34576 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34578 PR tree-optimization/79284
34579 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
34580 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
34581 vectorizable_mask_load_store, vectorizable_operation,
34582 vect_is_simple_cond, get_same_sized_vectype): Use it instead
34583 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
34584 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
34585 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
34586 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
34587 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
34588 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
34589 is_gimple_assign (stmt). Replace another such test with
34590 is_gimple_assign (stmt).
34592 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34595 * config/avr/avr.c (rtl-iter.h): Include it.
34596 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
34597 (avr_legitimate_combined_insn): ...and implementation.
34599 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
34601 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
34602 * config/s390/s390.c (s390_const_operand_ok)
34603 (s390_canonicalize_comparison, s390_extract_part)
34604 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
34605 (s390_contiguous_bitmask_p, s390_rtx_costs)
34606 (legitimize_pic_address): Likewise.
34607 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
34608 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
34609 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
34610 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
34611 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
34613 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34615 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
34616 REGNO($0) == REGNO($1).
34618 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34620 * config/s390/linux.h(SIZE_TYPE): Add comment.
34622 2017-02-06 Julian Brown <julian@codesourcery.com>
34623 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34624 Virendra Pathak <virendra.pathak@broadcom.com>
34626 * config/aarch64/aarch64-cores.def: Change the scheduler
34628 * config/aarch64/aarch64.md: Include thunderx2t99.md.
34629 * config/aarch64/thunderx2t99.md: New file.
34631 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34633 * doc/standards.texi (Go Language): Update link to language
34636 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34638 * tree-eh.c (lower_resx): Sanitize profile.
34639 (cleanup_empty_eh_move_lp): Likewise.
34641 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34644 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
34646 * cfgloopmanip.h (loop_version): Update prototype.
34647 * modulo-sched.c (sms_schedule): Update call of loop_version.
34648 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
34649 * tree-parloops.c (gen_parallel_loop): Likewise.
34650 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
34651 * tree-ssa-loop-split.c (split_loop): Likewise.
34652 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
34653 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
34655 2017-02-05 Martin Liska <mliska@suse.cz>
34658 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
34660 (print_operand_address): Initialize a struct to zero.
34662 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34664 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
34665 garbage collector only in textual form.
34667 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34669 * doc/extend.texi (x86 specific memory model extensions for
34670 transactional memory): Simplify a phrase.
34672 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
34675 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
34676 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
34677 (atomic_storedi_1): Likewise.
34679 2017-02-04 Jakub Jelinek <jakub@redhat.com>
34681 PR tree-optimization/79338
34682 * tree-parloops.c (gather_scalar_reductions): Don't call
34683 vect_analyze_loop_form for loop->inner before destroying loop's
34686 2017-02-03 Martin Sebor <msebor@redhat.com>
34688 PR tree-optimization/79327
34689 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
34690 when precision has resulted in leading zeros.
34691 (format_integer): Adjust the likely counter to assume an unknown
34692 argument that may be zero is non-zero.
34694 2017-02-03 Jason Merrill <jason@redhat.com>
34697 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
34698 avoid copying non-taken branch.
34700 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34702 PR tree-optimization/79340
34703 * tree-vect-loop.c (vectorizable_reduction): Release
34704 vec_defs elements after safe_splicing them into other vectors.
34707 PR tree-optimization/79327
34708 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
34709 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
34711 (format_integer): Use wide_int_to_tree instead of build_int_cst
34712 + to_?hwi. If argmin is NULL, just set argmin and argmax to
34713 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
34714 of shortest and longest sequence.
34716 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
34718 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
34719 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
34721 2017-02-03 Walter Lee <walt@tilera.com>
34724 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
34725 after initial stackframe link reg save.
34726 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
34728 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34731 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
34732 wu for stxssp alternative.
34734 2017-02-03 Martin Sebor <msebor@redhat.com>
34736 PR tree-optimization/79352
34737 * gimple-fold.c (get_range_strlen): Add argument.
34738 (get_range_strlen): Change return type to bool.
34739 (get_maxval_strlen): Pass in a dummy argument.
34740 * gimple-fold.h (get_range_strlen): Change return type to bool.
34741 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
34742 * tree.h (array_at_struct_end_p): Add argument.
34743 * tree.c (array_at_struct_end_p): Handle it.
34745 2017-02-03 Martin Liska <mliska@suse.cz>
34748 * multiple_target.c (create_dispatcher_calls): Redirect edge
34749 from a caller of a dispatcher.
34750 (expand_target_clones): Make the clones local.
34751 (ipa_target_clone): Do both target clones and resolvers.
34752 (ipa_dispatcher_calls): Remove the pass.
34753 (pass_dispatcher_calls::gate): Likewise.
34754 (make_pass_dispatcher_calls): Likewise.
34755 * passes.def (pass_target_clone): Put as very first IPA early
34758 2017-02-03 Martin Liska <mliska@suse.cz>
34760 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
34761 in case of a function with ifunc attribute.
34763 2017-02-03 Martin Liska <mliska@suse.cz>
34765 * cgraph.c (cgraph_node::dump): Dump function version info.
34766 * symtab.c (symtab_node::dump_base): Add missing new line.
34768 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34770 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
34771 (ifcombine_ifandif): Use it.
34773 2017-02-03 Martin Liska <mliska@suse.cz>
34775 * doc/invoke.texi: Document default value for
34776 use-after-scope-direct-emission-threshold.
34778 2017-02-03 Martin Liska <mliska@suse.cz>
34780 PR tree-optimization/79339
34781 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
34782 (format_floating): Likewise.
34784 2017-02-03 Martin Liska <mliska@suse.cz>
34787 * ipa-prop.c (ipa_node_params_t::insert): Remove current
34789 (ipa_node_params_t::remove): Likewise.
34790 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
34791 initialization from removed ipa_node_params_t::insert.
34792 (ipa_node_params::~ipa_node_params): Move from removed
34793 ipa_node_params_t::release.
34794 * symbol-summary.h (symbol_summary::m_released): New member.
34795 Do not release a summary twice. Do not allow to call finalizer
34796 for types of a summary that live in GGC memory.
34798 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34800 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
34803 2017-02-02 Martin Sebor <msebor@redhat.com>
34805 PR middle-end/79275
34806 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
34807 (format_string): Tighten up the range of output for non-constant
34808 strings and correct the expected range for wide non-constant strings.
34810 2017-02-02 Martin Sebor <msebor@redhat.com>
34812 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
34814 PR middle-end/32003
34815 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
34817 (-fdump-tree-@var): Add to index and document how to come up
34818 with pass-specific option and dump file names.
34819 (-fdump-passes): Clarify where to look for output.
34821 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34823 PR middle-end/77445
34824 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
34825 statistics of the analyzed path; allow threading for speed when
34826 any of BBs along the path are optimized for speed.
34828 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
34830 PR middle-end/78468
34831 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
34832 settings of the virtual registers.
34835 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
34837 * explow.c (get_dynamic_stack_size): Take known alignment of stack
34838 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
34841 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34843 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
34844 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
34846 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34848 * config/s390/s390.md: Add missing comments with the expanded
34850 * config/s390/vector.md: Likewise.
34851 * config/s390/vx-builtins.md: Likewise.
34853 2017-02-02 Jakub Jelinek <jakub@redhat.com>
34856 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
34857 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
34858 conditions on a single line.
34860 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34862 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34863 __S390_VX__ to __VX__.
34865 2017-02-01 Andrew Pinski <apinski@cavium.com>
34867 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
34868 stmt_info to record_stmt_cost.
34869 (vect_get_known_peeling_cost): Pass stmt_info if known to
34871 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
34872 cpu_vector_cost field into
34873 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
34874 field into vec_int_stmt_cost and vec_fp_stmt_cost.
34875 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
34876 splitting of scalar_stmt_cost and vec_stmt_cost.
34877 (thunderx_vector_cost): Likewise.
34878 (cortexa57_vector_cost): LIkewise.
34879 (exynosm1_vector_cost): Likewise.
34880 (xgene1_vector_cost): Likewise.
34881 (thunderx2t99_vector_cost): Improve after the splitting of the two
34883 (aarch64_builtin_vectorization_cost): Update for the splitting of
34884 scalar_stmt_cost and vec_stmt_cost.
34886 2017-02-01 Torvald Riegel <triegel@redhat.com>
34887 Richard Henderson <rth@redhat.com>
34889 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
34890 conditional on existance of a fast atomic load.
34891 * optabs-query.c (can_atomic_load_p): New function.
34892 * optabs-query.h (can_atomic_load_p): Declare it.
34893 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
34894 no fast atomic load is available for the particular size of access.
34895 (expand_atomic_compare_and_swap): Likewise.
34896 (expand_atomic_load): Likewise.
34897 (expand_atomic_store): Likewise.
34898 (expand_atomic_fetch_op): Likewise.
34899 * testsuite/lib/target-supports.exp
34900 (check_effective_target_sync_int_128): Remove x86 because it provides
34901 no fast atomic load.
34902 (check_effective_target_sync_int_128_runtime): Likewise.
34904 2017-02-01 Richard Biener <rguenther@suse.de>
34906 * graphite.c: Include tree-vectorizer.h for find_loop_location.
34907 (graphite_transform_loops): Provide opt-info for optimized nests.
34908 * tree-parloop.c (parallelize_loops): Provide opt-info for
34909 parallelized loops.
34911 2017-02-01 Richard Biener <rguenther@suse.de>
34913 PR middle-end/79315
34914 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
34915 was not set before.
34917 2017-02-01 Richard Biener <rguenther@suse.de>
34919 PR tree-optimization/71824
34920 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34921 Verify the loops are valid in the merged SESE region.
34922 (scop_detection::can_represent_loop_1): Check analyzing the
34923 evolution of the number of iterations in the region succeeds.
34925 2017-01-31 Ian Lance Taylor <iant@golang.org>
34927 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
34928 REG_ARGS_SIZE note to 32-bit push insns and call insn.
34930 2017-01-31 David Malcolm <dmalcolm@redhat.com>
34932 PR preprocessor/79210
34933 * input.c (get_substring_ranges_for_loc): Replace line_width
34934 assertion with error-handling.
34936 2017-01-31 Richard Biener <rguenther@suse.de>
34938 PR tree-optimization/77318
34939 * graphite-sese-to-poly.c (extract_affine): Fix assert.
34940 (create_pw_aff_from_tree): Take loop parameter.
34941 (add_condition_to_pbb): Pass loop of the condition to
34942 create_pw_aff_from_tree.
34944 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34946 * config/s390/s390.c (s390_asan_shadow_offset): New function.
34947 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
34949 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
34953 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
34955 (convert_int_to_float128): Likewise.
34956 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
34957 (convert_int_to_float128): Likewise.
34958 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
34959 (UNSPEC_IEEE128_CONVERT): Likewise.
34960 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
34961 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
34962 Use local variables for IBM extended format.
34963 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
34964 (fix_trunc<mode>si2_fprs): Likewise.
34965 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
34966 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
34967 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
34968 to know that we can now have integers of all sizes in vector
34970 (fix<uns>_<mode>di2_hw): Likewise.
34971 (float<uns>_<mode>si2_hw): Likewise.
34972 (fix_<mode>si2_hw): Likewise.
34973 (fixuns_<mode>si2_hw): Likewise.
34974 (float<uns>_<mode>di2_hw): Likewise.
34975 (float_<mode>di2_hw): Likewise.
34976 (float_<mode>si2_hw): Likewise.
34977 (floatuns_<mode>di2_hw): Likewise.
34978 (floatuns_<mode>si2_hw): Likewise.
34979 (xscvqp<su>wz_<mode>): Delete, no longer used.
34980 (xscvqp<su>dz_<mode>): Likewise.
34981 (xscv<su>dqp_<mode>): Likewise.
34982 (ieee128_mfvsrd_64bit): Likewise.
34983 (ieee128_mfvsrd_32bit): Likewise.
34984 (ieee128_mfvsrwz): Likewise.
34985 (ieee128_mtvsrw): Likewise.
34986 (ieee128_mtvsrd_64bit): Likewise.
34987 (ieee128_mtvsrd_32bit): Likewise.
34989 2017-01-31 Martin Liska <mliska@suse.cz>
34992 * ipa-prop.c (ipa_free_all_node_params): Call release method
34993 instead of ~sumbol_summary to not to trigger double times
34996 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
34998 PR tree-optimization/71691
34999 * bitmap.h (class auto_bitmap): New.
35000 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35001 is_maybe_undefined instead of ssa_undefined_value_p.
35003 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35005 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35006 __S390_ARCH_LEVEL__ to __ARCH__.
35008 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35010 PR tree-optimization/79267
35011 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35012 if should_remove_lhs_p is true.
35014 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35017 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35018 (add_alignment_attribute): New.
35019 (base_type_die): Add alignment attribute.
35020 (subrange_type_die): Likewise.
35021 (modified_type_die): Likewise.
35022 (gen_array_type_die): Likewise.
35023 (gen_descr_array_type_die: Likewise.
35024 (gen_enumeration_type_die): Likewise.
35025 (gen_subprogram_die): Likewise.
35026 (gen_variable_die): Likewise.
35027 (gen_field_die): Likewise.
35028 (gen_ptr_to_mbr_type_die): Likewise.
35029 (gen_struct_or_union_type_die): Likewise.
35030 (gen_subroutine_type_die): Likewise.
35031 (gen_typedef_die): Likewise.
35032 (base_type_cmp): Compare alignment attribute.
35034 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35037 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35038 (setb_unsigned) New pattern for setb with CCUNS.
35039 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35040 subfc./subfe sequence to avoid overflow problems. Generate a
35041 shorter sequence with cmpld/setb for power9.
35042 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35043 for generating subfc. instruction.
35044 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35045 now uses this instruction.
35047 2017-01-30 Ian Lance Taylor <iant@google.com>
35050 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35051 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35053 2017-01-30 Martin Sebor <msebor@redhat.com>
35055 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35056 Move constant to the right of a relational operator.
35057 (get_mpfr_format_length, format_character, format_string): Ditto.
35058 (should_warn_p, maybe_warn): Same.
35060 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35062 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35065 * asan.c (get_translation_unit_decl): Remove function.
35066 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35068 2017-01-30 Martin Liska <mliska@suse.cz>
35070 PR gcov-profile/79259
35071 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35072 -fprofile-generate.
35074 2017-01-30 Martin Liska <mliska@suse.cz>
35077 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35078 Initialize variables with NULL value.
35080 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35083 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35085 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35087 2017-01-30 Richard Biener <rguenther@suse.de>
35089 PR tree-optimization/79276
35090 * tree-vrp.c (process_assert_insertions): Properly adjust common
35091 when removing a duplicate.
35093 2017-01-30 Richard Biener <rguenther@suse.de>
35095 PR tree-optimization/79256
35096 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35097 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35099 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35101 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35104 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35105 ("*r<noxa>sbg_<mode>_sll_bitmask")
35106 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35107 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35108 Use contiguous_bitmask_nowrap_operand.
35110 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35113 * config/rs6000/altivec.h (vec_xl): Revise #define.
35114 (vec_xst): Likewise.
35116 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35118 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35120 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35122 PR rtl-optimization/79194
35123 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35124 traps before call to bypass_conditional_jumps.
35126 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35128 PR tree-optimization/71374
35129 * lra-constraints.c (check_conflict_input_operands): New.
35130 (match_reload): Use it.
35132 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35135 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35136 account to calculate conflict_set.
35138 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35140 PR rtl-optimization/78559
35141 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35142 other_insn in combine.
35144 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35146 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35147 uint16_type_node for BT_UINT16.
35149 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35151 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35152 "RTL Tests" to menu.
35153 (GIMPLE Tests): New node.
35154 (RTL Tests): New node.
35156 2017-01-27 Richard Biener <rguenther@suse.de>
35158 PR tree-optimization/79245
35159 * tree-loop-distribution.c (distribute_loop): Apply cost
35160 modeling also to detected patterns.
35162 2017-01-27 Richard Biener <rguenther@suse.de>
35164 PR tree-optimization/71433
35165 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35166 (compare_assert_loc): New function.
35167 (process_assert_insertions): Sort and optimize assert locations
35168 to remove duplicates and push down identical assertions on
35169 edges to their destination block.
35171 2017-01-27 Richard Biener <rguenther@suse.de>
35173 PR tree-optimization/79244
35174 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35175 out SSA names even if abnormal.
35177 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35179 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35180 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35181 instead of MPFR_RNDN.
35183 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35186 * arm.c (arm_option_override): Don't call build_target_option_node
35187 until after doing all option overrides.
35188 (arm_valid_target_attribute_tree): Likewise.
35190 2017-01-27 Martin Liska <mliska@suse.cz>
35192 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35193 for {cd}tors and C++ {cd}tors.
35195 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35197 * config/s390/s390.md ("*setmem_long_and")
35198 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35200 2017-01-26 Martin Sebor <msebor@redhat.com>
35202 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35205 2017-01-26 Martin Sebor <msebor@redhat.com>
35207 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35208 HAVE_DFmode before using XFmode or DFmode.
35209 (parse_directive): Avoid using the z length modifier to avoid
35210 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
35212 PR middle-end/78703
35213 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
35214 to accept adjustment as an array.
35215 (get_int_range): New function.
35216 (struct directive): Make width and prec arrays.
35217 (directive::set_width, directive::set_precision): Call get_int_range.
35218 (format_integer, format_floating): Handle width and precision ranges.
35219 (format_string, parse_directive): Same.
35221 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35224 * dwarf2out.c (generate_skeleton_bottom_up): For children with
35225 comdat_type_p set, just clone them, but keep the children in the
35229 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
35230 which have direct callers with -fvar-tracking-assignments enabled
35232 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
35233 inside of type units.
35235 2017-01-26 Martin Sebor <msebor@redhat.com>
35237 PR middle-end/78703
35238 * gimple-ssa-sprintf.c (struct result_range): Add likely and
35240 (struct format_result): Replace number_chars, number_chars_min,
35241 and number_chars_max with a single member of struct result_range.
35243 (format_result::operator+=): Adjust.
35244 (struct fmtresult): Remove bounded. Handle likely and unlikely
35246 (fmtresult::adjust_for_width_or_precision): New function.
35247 (fmtresult:type_max_digits): New function.
35248 (bytes_remaining): Handle likely and unlikely counters.
35249 (min_bytes_remaining): Remove.
35250 (format_percent): Simplify.
35251 (format_integer, format_floating): Set likely and unlikely counters.
35252 (get_string_length, format_character, format_string): Same.
35253 (format_plain, should_warn_p): New function.
35254 (maybe_warn): Call should_warn_p. Update diagnostic messages
35255 and handle those for all directives, including plain strings.
35256 (format_directive): Handle likely and unlikely counters.
35257 Remove unnecessary quoting from diagnostics. Add an informational
35259 (add_bytes): Remove.
35260 (pass_sprintf_length::compute_format_length): Simplify.
35261 (try_substitute_return_value): Handle likely and unlikely counters.
35263 2017-01-26 Carl Love <cel@us.ibm.com>
35265 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
35266 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
35268 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
35271 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
35272 endianess for subregs into account.
35273 * lra-constraints.c (lra_constraints): Do risky transformations
35274 always on the first iteration.
35275 * lra-lives.c (check_pseudos_live_through_calls): Add arg
35276 last_call_used_reg_set.
35277 (process_bb_lives): Define and use last_call_used_reg_set.
35278 * lra.c (lra): Always continue after lra_constraints on the first
35281 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
35283 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
35285 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
35287 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35289 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
35290 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
35291 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
35292 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
35293 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
35294 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
35295 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
35296 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
35297 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
35299 2017-01-26 Marek Polacek <polacek@redhat.com>
35302 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
35303 for the third operand.
35305 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35307 PR middle-end/79236
35308 * omp-low.c (struct omp_context): Add simt_stmt field.
35309 (scan_omp_for): Return omp_context *.
35310 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
35311 context to the _simt_ SIMD stmt.
35312 (lower_omp_for): For combined SIMD with sibling _simt_
35313 SIMD, make sure to use the same decls in _looptemp_
35314 clauses as in the sibling.
35316 2017-01-26 David Sherwood <david.sherwood@arm.com>
35318 PR middle-end/79212
35319 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
35322 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35325 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
35326 emit fld b; fld a; if possible.
35328 * brig-builtins.def: Update copyright years.
35329 * config/arm/arm_acle_builtins.def: Update copyright years.
35331 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
35334 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
35335 constraint instead of o for the stxsd instruction.
35337 2017-01-25 Carl Love <cel@us.ibm.com>
35339 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
35340 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
35342 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
35344 * doc/invoke.texi (C++ Dialect Options): Fix typo.
35346 2017-01-25 Richard Biener <rguenther@suse.de>
35348 PR tree-optimization/69264
35349 * target.def (vector_alignment_reachable): Improve documentation.
35350 * doc/tm.texi: Regenerate.
35351 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
35353 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
35354 earlier changes with respect to TYPE_USER_ALIGN.
35355 (vector_alignment_reachable_p): Likewise. Improve dumping.
35357 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35360 * config/arm/arm.md (xordi3): Force constant operand into a register
35363 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35365 * doc/invoke.texi (-fstore-merging): Correct default optimization
35366 levels at which it is enabled.
35367 (-O): Move -fstore-merging from list to...
35370 2017-01-25 Richard Biener <rguenther@suse.de>
35373 * omp-expand.c: Include debug.h.
35374 (expand_omp_taskreg): Make sure to generate early debug before
35375 outlining anything from a function.
35376 (expand_omp_target): Likewise.
35377 (grid_expand_target_grid_body): Likewise.
35379 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
35382 * asan.c (get_translation_unit_decl): New function.
35383 (asan_add_global): Extract modules file name from globals
35384 TRANSLATION_UNIT_DECL name.
35386 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
35389 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
35390 for long calls with APCS frame and VFP.
35392 2017-01-24 David Malcolm <dmalcolm@redhat.com>
35394 * cfg.c (original_copy_tables_initialized_p): New function.
35395 * cfg.h (original_copy_tables_initialized_p): New decl.
35396 * cfgrtl.c (relink_block_chain): Guard the call to
35397 free_original_copy_tables with a call to
35398 original_copy_tables_initialized_p.
35399 * cgraph.h (symtab_node::native_rtl_p): New decl.
35400 * cgraphunit.c (symtab_node::native_rtl_p): New function.
35401 (symtab_node::needed_p): Don't assert for early assembly output
35402 for __RTL functions.
35403 (cgraph_node::finalize_function): Set "force_output" for __RTL
35405 (cgraph_node::analyze): Bail out early for __RTL functions.
35406 (analyze_functions): Update assertion to support __RTL functions.
35407 (cgraph_node::expand): Bail out early for __RTL functions.
35408 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
35410 * function.h (struct function): Update comment for field
35412 * gimple-expr.c: Include "tree-pass.h".
35413 (gimple_has_body_p): Return false for __RTL functions.
35414 * Makefile.in (OBJS): Add run-rtl-passes.o.
35415 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
35417 (gcc::pass_manager::get_clean_slate): New accessor.
35418 * passes.c: Include "insn-addr.h".
35419 (should_skip_pass_p): Add logging. Update logic for running
35420 "expand" to be compatible with both __GIMPLE and __RTL. Guard
35421 property-provider override so it is only done for gimple passes.
35423 (skip_pass): New function.
35424 (execute_one_pass): Call skip_pass when skipping passes.
35425 * read-md.c (md_reader::read_char): Support filtering
35426 the input to a subset of line numbers.
35427 (md_reader::md_reader): Initialize fields
35428 m_first_line and m_last_line.
35429 (md_reader::read_file_fragment): New function.
35430 * read-md.h (md_reader::read_file_fragment): New decl.
35431 (md_reader::m_first_line): New field.
35432 (md_reader::m_last_line): New field.
35433 * read-rtl-function.c (function_reader::create_function): Only
35434 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
35435 curr_properties. Set DECL_INITIAL to a dummy block.
35436 (read_rtl_function_body_from_file_range): New function.
35437 * read-rtl-function.h (read_rtl_function_body_from_file_range):
35439 * run-rtl-passes.c: New file.
35440 * run-rtl-passes.h: New file.
35442 2017-01-24 Jeff Law <law@redhat.com>
35444 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
35447 2017-01-24 Bin Cheng <bin.cheng@arm.com>
35449 PR tree-optimization/79159
35450 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
35451 (record_nonwrapping_iv): Improve boundary using above function if no
35452 value range information.
35454 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
35455 Martin Jambor <mjambor@suse.cz>
35457 * brig-builtins.def: New file.
35458 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
35459 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
35460 (DEF_HSAIL_SAT_BUILTIN): Likewise.
35461 (DEF_HSAIL_INTR_BUILTIN): Likewise.
35462 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
35463 * builtin-types.def (BT_INT8): New.
35464 (BT_INT16): Likewise.
35465 (BT_UINT8): Likewise.
35466 (BT_UINT16): Likewise.
35467 (BT_FN_ULONG): Likewise.
35468 (BT_FN_UINT_INT): Likewise.
35469 (BT_FN_UINT_ULONG): Likewise.
35470 (BT_FN_UINT_LONG): Likewise.
35471 (BT_FN_UINT_PTR): Likewise.
35472 (BT_FN_ULONG_PTR): Likewise.
35473 (BT_FN_INT8_FLOAT): Likewise.
35474 (BT_FN_INT16_FLOAT): Likewise.
35475 (BT_FN_UINT32_FLOAT): Likewise.
35476 (BT_FN_UINT16_FLOAT): Likewise.
35477 (BT_FN_UINT8_FLOAT): Likewise.
35478 (BT_FN_UINT64_FLOAT): Likewise.
35479 (BT_FN_UINT16_UINT32): Likewise.
35480 (BT_FN_UINT32_UINT16): Likewise.
35481 (BT_FN_UINT16_UINT16_UINT16): Likewise.
35482 (BT_FN_INT_PTR_INT): Likewise.
35483 (BT_FN_UINT_PTR_UINT): Likewise.
35484 (BT_FN_LONG_PTR_LONG): Likewise.
35485 (BT_FN_ULONG_PTR_ULONG): Likewise.
35486 (BT_FN_VOID_UINT64_UINT64): Likewise.
35487 (BT_FN_UINT8_UINT8_UINT8): Likewise.
35488 (BT_FN_INT8_INT8_INT8): Likewise.
35489 (BT_FN_INT16_INT16_INT16): Likewise.
35490 (BT_FN_INT_INT_INT): Likewise.
35491 (BT_FN_UINT_FLOAT_UINT): Likewise.
35492 (BT_FN_FLOAT_UINT_UINT): Likewise.
35493 (BT_FN_ULONG_UINT_UINT): Likewise.
35494 (BT_FN_ULONG_UINT_PTR): Likewise.
35495 (BT_FN_ULONG_ULONG_ULONG): Likewise.
35496 (BT_FN_UINT_UINT_UINT): Likewise.
35497 (BT_FN_VOID_UINT_PTR): Likewise.
35498 (BT_FN_UINT_UINT_PTR: Likewise.
35499 (BT_FN_UINT32_UINT64_PTR): Likewise.
35500 (BT_FN_INT_INT_UINT_UINT): Likewise.
35501 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
35502 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
35503 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
35504 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
35505 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
35506 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
35507 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
35508 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
35509 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
35510 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
35511 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
35512 * doc/frontends.texi: List BRIG FE.
35513 * doc/install.texi (Testing): Add BRIG tesring requirements.
35514 * doc/invoke.texi (Overall Options): Mention BRIG.
35515 * doc/standards.texi (Standards): Doucment BRIG HSA version.
35517 2017-01-24 Richard Biener <rguenther@suse.de>
35519 PR translation/79208
35520 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
35522 2017-01-24 Martin Jambor <mjambor@suse.cz>
35525 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
35526 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
35527 and known_contexts.
35529 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
35531 PR middle-end/79123
35532 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
35533 casts from signed to unsigned really don't have a range.
35535 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
35537 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
35538 GMP_RNDx for compatiblity.
35540 2017-01-24 Martin Liska <mliska@suse.cz>
35543 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
35544 that would prevent us to call alloca with -1 as argument.
35546 2017-01-24 Jakub Jelinek <jakub@redhat.com>
35548 * dwarf2out.c (output_compilation_unit_header, output_file_names):
35549 Avoid -Wformat-security warning.
35551 2017-01-23 Andrew Pinski <apinski@cavium.com>
35553 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
35556 2017-01-23 Martin Sebor <msebor@redhat.com>
35558 PR middle-end/78703
35559 * gimple-ssa-sprintf.c (warn_level): New global.
35560 (format_integer): Use it here and throughout the rest of the file.
35561 Use the same switch to compute sign as base.
35562 (maybe_warn): New function.
35563 (format_directive): Factor out warnings into maybe_warn.
35564 Add debugging output. Use warn_level.
35565 (add_bytes): Use warn_level.
35566 (pass_sprintf_length::compute_format_length): Add debugging output.
35567 (try_substitute_return_value): Same.
35568 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
35570 PR middle-end/78703
35571 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
35572 (struct fmtresult, format_integer, format_floating): Adjust.
35573 (fmtresult::fmtresult): Set max correctly in two argument ctor.
35574 (get_string_length, format_string,format_directive): Same.
35575 (pass_sprintf_length::compute_format_length): Same.
35576 (try_substitute_return_value): Simplify slightly.
35578 PR middle-end/78703
35579 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
35580 (fmtresult::operator+=): Outlined.
35581 (struct fmtresult): Add ctors.
35582 (struct conversion_spec): Rename...
35583 (struct directive): ...to this. Add and remove data members.
35584 (directive::set_width, directive::set_precision): New functions.
35585 (format_percent): Use fmtresult ctor.
35586 (get_width_and_precision): Remove.
35587 (format_integer): Make naming changes. Avoid computing width and
35589 (format_floating): Same. Adjust indentation.
35590 (format_character, format_none): New functions.
35591 (format_string): Moved character handling to format_character.
35592 (format_directive): Remove arguments, change return type.
35593 (parse_directive): New function.
35594 (pass_sprintf_length::compute_format_length): Move directive
35595 parsing to parse_directive.
35597 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35599 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
35600 (assign_assembler_name_if_needed): ... this.
35601 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
35602 (assign_assembler_name_if_needed): ... this.
35603 (free_lang_data_in_cgraph): Adjust callers.
35604 * cgraphunit.c (cgraph_node::analyze): Likewise.
35605 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
35608 2017-01-23 Richard Biener <rguenther@suse.de>
35610 PR tree-optimization/79088
35611 PR tree-optimization/79188
35612 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
35613 resetting loop bounds after last path deletion. Reset loop
35614 bounds of the target loop, make code match the comments.
35615 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
35616 Make sure loops need no fixups.
35618 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35620 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
35621 exponent support with double type for first argument.
35622 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
35623 type returned by __builtin_vec_extract_sig,
35624 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
35625 functions from "vector int" to "vector unsigned int" or from
35626 "vector long long int" to "vector unsigned long long int".
35627 Changed type returned by __builtin_vec_extract_exp,
35628 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
35629 functions from "vector int" to "vector unsigned int" or from
35630 "vector long long int" to "vector unsigned long long int".
35631 Changed return type of __builtin_vec_test_data_class,
35632 __builtin_vec_test_data_class_sp, and
35633 __builtin_vec_test_data_class_dp from "vector int" to
35634 "vector bool int" or from "vector long long int" to "vector bool
35635 long long int" and changed second argument type from "unsigned
35636 int" to "int". Added new overloaded function forms "vector float
35637 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
35638 "vector float __builtin_vec_insert_exp_sp (vector float, vector
35639 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
35640 double, vector unsigned long long int)" and "vector double
35641 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
35642 long int)". Changed return type of
35643 __builtin_scalar_test_data_class and
35644 __builtin_scalar_test_data_class_sp and
35645 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
35646 int" and changed second argument from "unsigned int" to "int".
35647 Changed type returned by __builtin_scalar_test_neg,
35648 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
35649 from "int" to "bool int". Added new overloaded function form
35650 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
35651 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
35652 exponent double-precision with floating point first argument.
35653 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
35654 documentation of scalar_test_data_class, scalar_test_neg,
35655 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
35656 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
35657 vec_test_data_class built-in functions to reflect refinements in
35658 their type signatures.
35660 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
35662 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
35664 (aarch64_elf_asm_destructor): Likewise.
35666 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
35668 PR rtl-optimization/78634
35669 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
35670 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
35671 * ifcvt.c (noce_try_cmove): Add missing cost check.
35673 PR rtl-optimization/71724
35674 * combine.c (if_then_else_cond): Look for situations where it is
35675 beneficial to undo the work of one of the recursive calls.
35677 2017-01-23 Bin Cheng <bin.cheng@arm.com>
35679 PR tree-optimization/70754
35680 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
35681 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
35682 combined stmt before it if not NULL.
35683 (combine_chains): Process refs reversely and compute dominance point
35686 2017-01-23 Martin Liska <mliska@suse.cz>
35688 PR tree-optimization/79196
35689 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
35690 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
35692 (strlen_optimize_stmt): Call the renamed function.
35694 2017-01-23 Michael Matz <matz@suse.de>
35696 PR tree-optimization/78384
35697 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
35699 2017-01-23 Richard Biener <rguenther@suse.de>
35701 PR tree-optimization/79186
35702 * tree-vrp.c (register_new_assert_for): Make sure we've seen
35703 both incoming edges before moving an assert.
35705 2017-01-23 Martin Jambor <mjambor@suse.cz>
35707 * ipa-prop.c (load_from_param_1): Removed.
35708 (load_from_unmodified_param): Bits from load_from_param_1 put back
35710 (load_from_param): Removed.
35711 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
35712 with stmt. Reverted back to use of load_from_unmodified_param.
35714 2017-01-23 Martin Jambor <mjambor@suse.cz>
35717 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
35718 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
35719 field a pointer to garbage collected vector, mark lattices and
35720 ipcp_orig_node with GTY((skip)).
35721 (ipa_get_param_count): Adjust to descriptors being a pointer.
35722 (ipa_get_param): Likewise.
35723 (ipa_get_type): Likewise.
35724 (ipa_get_param_move_cost): Likewise.
35725 (ipa_set_param_used): Likewise.
35726 (ipa_get_controlled_uses): Likewise.
35727 (ipa_set_controlled_uses): Likewise.
35728 (ipa_is_param_used): Likewise.
35729 (ipa_node_params_t): Move into garbage collector. New methods insert
35731 (ipa_node_params_sum): Annotate wth GTY(()).
35732 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
35734 (ipa_load_from_parm_agg): Adjust declaration.
35735 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
35736 * ipa-profile.c (ipa_profile): Likewise.
35737 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
35738 (ipa_populate_param_decls): Make descriptors parameter garbage
35740 (ipa_dump_param): Adjust to descriptors being a pointer.
35741 (ipa_alloc_node_params): Likewise.
35742 (ipa_initialize_node_params): Likewise.
35743 (load_from_param_1): Make descriptors parameter garbage collected.
35744 (load_from_unmodified_param): Likewise.
35745 (load_from_param): Likewise.
35746 (ipa_load_from_parm_agg): Likewise.
35747 (ipa_node_params::~ipa_node_params): Removed.
35748 (ipa_free_all_node_params): Remove call to delete operator.
35749 (ipa_node_params_t::insert): New.
35750 (ipa_node_params_t::remove): Likewise.
35751 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
35752 copy known_csts and known_contexts vectors.
35753 (ipa_read_node_info): Adjust to descriptors being a pointer.
35754 (ipcp_modif_dom_walker): Make m_descriptors field garbage
35756 (ipcp_transform_function): Make descriptors variable garbage
35759 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
35761 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
35762 * config/i386/avx512dqintrin.h: Ditto.
35763 * config/i386/avx512fintrin.h: Ditto.
35764 * config/i386/i386.c: Handle new builtins.
35765 * config/i386/i386-builtin.def: Add new builtins.
35766 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
35767 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
35769 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35770 Martin Liska <mliska@suse.cz>
35772 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
35773 * asan.c (asan_expand_poison_ifn): Support stores and use
35774 appropriate ASAN report function.
35775 * internal-fn.c (expand_ASAN_POISON_USE): New function.
35776 * internal-fn.def (ASAN_POISON_USE): Declare.
35777 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
35778 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
35779 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
35780 ASAN_POISON calls w/o LHS.
35781 * tree-ssa.c (execute_update_addresses_taken): Create clobber
35782 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
35783 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
35784 * gimplify.c (asan_poison_variables): Add attribute
35785 use_after_scope_memory to variables that really needs to live
35787 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
35788 having the attribute.
35790 2017-01-23 Martin Liska <mliska@suse.cz>
35792 * asan.c (create_asan_shadow_var): New function.
35793 (asan_expand_poison_ifn): Likewise.
35794 * asan.h (asan_expand_poison_ifn): New declaration.
35795 * internal-fn.c (expand_ASAN_POISON): Likewise.
35796 * internal-fn.def (ASAN_POISON): New builtin.
35797 * sanopt.c (pass_sanopt::execute): Expand
35798 asan_expand_poison_ifn.
35799 * tree-inline.c (copy_decl_for_dup_finish): Make function
35801 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
35802 * tree-ssa.c (is_asan_mark_p): New function.
35803 (execute_update_addresses_taken): Rewrite local variables
35804 (identified just by use-after-scope as addressable) into SSA.
35806 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35808 * doc/install.texi (Specific): opensource.apple.com uses https
35809 now. Remove trailing slash.
35811 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35813 * README.Portability: Remove note on an Irix compatibility issue.
35815 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
35817 * gcov.c (INCLUDE_ALGORITHM): Define.
35818 (INCLUDE_VECTOR): Define.
35819 No longer include <vector> and <algorithm> directly.
35821 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
35823 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
35825 * doc/invoke.texi (Code Gen Options): Ditto.
35827 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
35830 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
35832 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
35834 rtl-optimization/79125
35835 * cprop.c (local_cprop_pass): Handle cases where we make an
35836 unconditional trap.
35838 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
35842 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
35843 read from, for big endian.
35845 2017-01-20 Jiong Wang <jiong.wang@arm.com>
35847 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
35848 register pauth builtins for LP64 only.
35850 2017-01-20 Marek Polacek <polacek@redhat.com>
35853 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
35856 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35858 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
35860 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
35861 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
35862 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
35864 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35867 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
35868 in big-endian mode when they are not a single duplicated value.
35870 2017-01-20 Richard Biener <rguenther@suse.de>
35872 * BASE-VER: Bump to 7.0.1.
35874 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35876 * omp-low.c (omplow_simd_context): New struct. Use it...
35877 (lower_rec_simd_input_clauses): ...here and...
35878 (lower_rec_input_clauses): ...here to hold common data. Adjust all
35879 references to idx, lane, max_vf, is_simt.
35881 2017-01-20 Graham Markall <graham.markall@embecosm.com>
35883 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
35886 2017-01-20 Martin Jambor <mjambor@suse.cz>
35888 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
35889 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
35891 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
35892 (GTFILES): Rename hsa.c to hsa-common.c.
35893 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
35894 * hsa-dump.c: Likewise.
35895 * hsa-gen.c: Likewise.
35896 * hsa-regalloc.c: Likewise.
35897 * ipa-hsa.c: Likewise.
35898 * omp-expand.c: Likewise.
35899 * omp-low.c: Likewise.
35900 * toplev.c: Likewise.
35902 2017-01-20 Marek Polacek <polacek@redhat.com>
35905 * doc/invoke.texi: Document -Wduplicated-branches.
35906 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
35907 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
35908 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
35909 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
35910 return 0 only when not OEP_LEXICOGRAPHIC.
35911 (fold_build_cleanup_point_expr): Use the expression
35912 location when building CLEANUP_POINT_EXPR.
35913 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
35914 * tree.c (add_expr): Handle error_mark_node.
35916 2017-01-20 Martin Liska <mliska@suse.cz>
35919 * tree-profile.c (init_ic_make_global_vars): Do not call
35921 (gimple_init_gcov_profiler): Likewise.
35923 2017-01-20 Martin Liska <mliska@suse.cz>
35926 * cgraph.h (maybe_create_reference): Remove argument and
35928 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
35930 * ipa-cp.c (create_specialized_node): Likewise.
35931 * symtab.c (symtab_node::maybe_create_reference): Handle
35932 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
35934 2017-01-20 Martin Liska <mliska@suse.cz>
35936 * read-rtl-function.c (function_reader::create_function): Use
35937 build_decl instread of build_decl_stat.
35939 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
35941 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
35942 * config/i386/avx512dqintrin.h: Ditto.
35943 * config/i386/avx512fintrin.h: Ditto.
35944 * config/i386/i386-builtin-types.def: Add new types.
35945 * config/i386/i386.c: Handle new types.
35946 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
35947 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
35948 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
35949 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
35950 (__builtin_ia32_kshiftridi): New.
35951 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
35953 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
35957 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
35958 define to rs6000_init_stack_protect_guard.
35959 (rs6000_init_stack_protect_guard): New function.
35961 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35962 Yunqiang Su <yunqiang.su@imgtec.com>
35964 * config.gcc (supported_defaults): Add madd4.
35965 (with_madd4): Add validation.
35966 (all_defaults): Add madd4.
35967 * config/mips/mips.opt (mmadd4): New option.
35968 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35970 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35972 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
35973 (ISA_HAS_FUSED_MADD4): Likewise.
35974 * doc/invoke.texi (-mmadd4): Document the new option.
35975 * doc/install.texi (--with-madd4): Document the new option.
35977 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35979 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
35980 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
35981 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
35982 (aarch64_init_pauth_hint_builtins): New.
35983 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
35984 (aarch64_expand_builtin): Expand new builtins.
35986 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35988 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
35989 * combine-stack-adj.c (no_unhandled_cfa): Handle
35990 REG_CFA_TOGGLE_RA_MANGLE.
35991 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
35992 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
35993 info for return address signing.
35994 (aarch64_expand_epilogue): Likewise.
35996 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35998 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
35999 * config/aarch64/aarch64-protos.h
36000 (aarch64_return_address_signing_enabled): New declaration.
36001 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36003 (aarch64_expand_prologue): Sign return address before it's pushed onto
36005 (aarch64_expand_epilogue): Authenticate return address fetched from
36007 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36008 (aarch64_attributes): New function attributes for "sign-return-address".
36009 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36010 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36011 ("*do_return"): Generate combined instructions according to key index.
36012 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36013 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36015 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36016 * config/aarch64/aarch64.opt (msign-return-address=): New.
36017 * doc/extend.texi (AArch64 Function Attributes): Documents
36018 "sign-return-address=".
36019 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36021 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36023 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36024 overall option summary.
36026 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36028 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36029 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36030 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36031 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36033 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36035 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36036 -mpower9-minmax by default for -mcpu=power9.
36037 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36038 128-bit floating point.
36040 2017-01-20 Alan Modra <amodra@gmail.com>
36042 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36043 optimizing for size.
36045 2017-01-20 Alan Modra <amodra@gmail.com>
36048 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36049 for strcmp and strncmp from corresponding builtin decl.
36051 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36053 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36054 instead of i386/rtems-64.h.
36055 * config/i386/rtems-64.h: Remove.
36057 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36061 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36063 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36065 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36067 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36068 Change int to HOST_WIDE_INT.
36069 * config/aarch64/aarch64-protos.h
36070 (aarch64_simd_gen_const_vector_dup): Likewise.
36071 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36073 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36075 * langhooks-def.h (lhd_type_for_size): New decl.
36076 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36077 * langhooks.c (lhd_type_for_size): New function, taken from
36080 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36082 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36083 define_bypass for CR latency.
36084 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36085 (power9-alu2): Add define_bypass for CR latency.
36087 (power9-mul): Update insn latency.
36088 (power9-mul-compare): Update insn latency, bypass latency and remove
36091 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36093 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36095 * config/aarch64/aarch64.md
36096 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36097 aarch64_nopcrelative_literal_loads.
36098 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36100 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36102 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36103 TARGET_LOONGSON_3A.
36104 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36106 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36109 * config.gcc (supported_defaults): Add lxc1-sxc1.
36110 (with_lxc1_sxc1): Add validation.
36111 (all_defaults): Add lxc1-sxc1.
36112 * config/mips/mips.opt (mlxc1-sxc1): New option.
36113 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36115 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36116 __mips_no_lxc1_sxc1.
36117 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36118 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36119 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36121 2017-01-19 Richard Biener <rguenther@suse.de>
36123 PR tree-optimization/72488
36124 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36125 sure to restore SSA info.
36126 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36128 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36130 PR rtl-optimization/79121
36131 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36132 of the inner type when shifting an extended value.
36134 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36137 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36138 interposable aliases.
36140 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36143 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36144 Use the evmergelohi instruction.
36145 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36146 (mov_sitf_e500_subreg8_2_be): Likewise.
36147 (mov_sitf_e500_subreg12_2_le): Likewise.
36148 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36149 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36150 (mov_sitf_e500_subreg8_2_le): Likewise.
36151 (mov_sitf_e500_subreg12_2_be): Likewise.
36153 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36155 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36156 attribute from vecsimple to vecperm.
36157 (altivec_vbpermq2): Likewise.
36159 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36162 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36164 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36165 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36166 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36167 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36168 case where N arg is SIZE_MAX.
36169 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36170 (cmpstrsi): Add pattern.
36172 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36174 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36175 __builtin_vec_revb builtins.
36176 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36177 built-in functions to support generation of the ISA 3.0 XXBR<x>
36178 vector byte reverse instructions.
36179 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36180 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36181 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36182 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36183 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36184 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36185 (P9V_BUILTIN_VEC_REVB): Likewise.
36186 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36187 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36188 (p9_xxbrq_v16qi): Likewise.
36189 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36190 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36191 (p9_xxbrh_v8hi): Likewise.
36192 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36193 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36194 vec_revb built-in functions.
36196 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36198 PR rtl-optimization/78952
36199 * config/i386/i386.md (any_extract): New code iterator.
36200 (*insvqi_2): Use any_extract for source operand.
36201 (*insvqi_3): Use any_shiftrt for source operand.
36203 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36205 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36207 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36209 2017-01-18 Matthias Klose <doko@ubuntu.com>
36211 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
36213 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36215 * config/rs6000/altivec.h (vec_bperm): Change #define.
36216 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
36217 (altivec_vbpermq2): New define_insn.
36218 (altivec_vbpermd): Likewise.
36219 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
36220 function interface.
36221 (VBPERMD): Likewise.
36222 (VBPERM): New polymorphic function interface.
36223 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
36224 Add entries for P9V_BUILTIN_VEC_VBPERM.
36225 * doc/extend.texi: Add interfaces for vec_bperm.
36227 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36229 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
36230 first letter of error messages.
36231 (s390_resolve_overloaded_builtin): Likewise.
36232 * config/s390/s390.c (s390_expand_builtin): Likewise.
36233 (s390_invalid_arg_for_unprototyped_fn): Likewise.
36234 (s390_valid_target_attribute_inner_p): Likewise.
36235 * config/s390/s390.md ("tabort"): Likewise.
36237 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
36239 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
36240 (ISA_AVOID_DIV_HILO): New macro.
36241 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
36242 (ISA_HAS_DDIV): Likewise.
36244 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36246 * doc/invoke.texi (fabi-version): Correct number of occurrences.
36248 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36250 * doc/invoke.texi (fabi-version): Spelling fix.
36252 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36255 * doc/invoke.texi (fabi-version): Mention mangling fix for
36258 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36261 * doc/invoke.texi (fabi-version): Document discriminator mangling.
36263 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
36266 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
36267 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
36269 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
36271 (stack_protect_test): Ditto.
36272 * config/rs6000/rs6000.opt (mstack-protector-guard=,
36273 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
36275 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
36276 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
36277 -mstack-protector-guard-offset=.
36278 (RS/6000 and PowerPC Options): Ditto.
36280 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36282 * config/i386/i386.h (MASK_CLASS_P): New define.
36283 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
36284 there are no registers from different register sets also when
36285 mask registers are used. Update function comment.
36286 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
36287 to (*k/*r) and (*k/*km) alternatives.
36289 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
36291 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
36292 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
36293 (EH_RETURN_HANDLER_RTX): New define.
36294 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
36295 Force frame pointer in EH return functions.
36296 (aarch64_expand_epilogue): Add barrier for eh_return.
36297 (aarch64_final_eh_return_addr): Remove.
36298 (aarch64_eh_return_handler_rtx): New function.
36299 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
36301 (aarch64_eh_return_handler_rtx): New prototype.
36303 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36305 * config/rs6000/altivec.h (vec_rlmi): New #define.
36306 (vec_vrlnm): Likewise.
36307 (vec_rlnm): Likewise.
36308 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
36309 (UNSPEC_VRLNM): Likewise.
36310 (VIlong): New mode iterator.
36311 (altivec_vrl<VI_char>mi): New define_insn.
36312 (altivec_vrl<VI_char>nm): Likewise.
36313 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
36315 (VRLDNM): Likewise.
36316 (RLNM): New polymorphic function entry.
36317 (VRLWMI): New monomorphic function entry.
36318 (VRLDMI): Likewise.
36319 (RLMI): New polymorphic function entry.
36320 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
36321 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
36322 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
36325 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36328 * dwarf2out.c (field_byte_offset): Restore the
36329 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
36330 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
36331 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
36334 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36337 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
36339 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36342 * dwarf2out.c (add_data_member_location_attribute): For constant
36343 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
36344 instead of DW_AT_data_member_location, DW_AT_bit_offset and
36345 DW_AT_byte_size attributes.
36347 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36349 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
36350 after forcing to constant memory when the code model is medium.
36352 2017-01-17 Julia Koval <julia.koval@intel.com>
36355 * config/i386/avx512fintrin.h
36356 (_mm512_i32gather_ps): Change __addr type to void const*.
36357 (_mm512_mask_i32gather_ps): Ditto.
36358 (_mm512_i32gather_pd): Ditto.
36359 (_mm512_mask_i32gather_pd): Ditto.
36360 (_mm512_i64gather_ps): Ditto.
36361 (_mm512_mask_i64gather_ps): Ditto.
36362 (_mm512_i64gather_pd): Ditto.
36363 (_mm512_mask_i64gather_pd): Ditto.
36364 (_mm512_i32gather_epi32): Ditto.
36365 (_mm512_mask_i32gather_epi32): Ditto.
36366 (_mm512_i32gather_epi64): Ditto.
36367 (_mm512_mask_i32gather_epi64): Ditto.
36368 (_mm512_i64gather_epi32): Ditto.
36369 (_mm512_mask_i64gather_epi32): Ditto.
36370 (_mm512_i64gather_epi64): Ditto.
36371 (_mm512_mask_i64gather_epi64): Ditto.
36372 (_mm512_i32scatter_ps): Change __addr type to void*.
36373 (_mm512_mask_i32scatter_ps): Ditto.
36374 (_mm512_i32scatter_pd): Ditto.
36375 (_mm512_mask_i32scatter_pd): Ditto.
36376 (_mm512_i64scatter_ps): Ditto.
36377 (_mm512_mask_i64scatter_ps): Ditto.
36378 (_mm512_i64scatter_pd): Ditto.
36379 (_mm512_mask_i64scatter_pd): Ditto.
36380 (_mm512_i32scatter_epi32): Ditto.
36381 (_mm512_mask_i32scatter_epi32): Ditto.
36382 (_mm512_i32scatter_epi64): Ditto.
36383 (_mm512_mask_i32scatter_epi64): Ditto.
36384 (_mm512_i64scatter_epi32): Ditto.
36385 (_mm512_mask_i64scatter_epi32): Ditto.
36386 (_mm512_i64scatter_epi64): Ditto.
36387 (_mm512_mask_i64scatter_epi64): Ditto.
36388 * config/i386/avx512pfintrin.h
36389 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
36390 (_mm512_mask_prefetch_i32gather_ps): Ditto.
36391 (_mm512_mask_prefetch_i64gather_pd): Ditto.
36392 (_mm512_mask_prefetch_i64gather_ps): Ditto.
36393 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
36394 (_mm512_prefetch_i32scatter_ps): Ditto.
36395 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
36396 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
36397 (_mm512_prefetch_i64scatter_pd): Ditto.
36398 (_mm512_prefetch_i64scatter_ps): Ditto.
36399 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
36400 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
36401 * config/i386/avx512vlintrin.h
36402 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
36403 (_mm_mmask_i32gather_ps): Ditto.
36404 (_mm256_mmask_i32gather_pd): Ditto.
36405 (_mm_mmask_i32gather_pd): Ditto.
36406 (_mm256_mmask_i64gather_ps): Ditto.
36407 (_mm_mmask_i64gather_ps): Ditto.
36408 (_mm256_mmask_i64gather_pd): Ditto.
36409 (_mm_mmask_i64gather_pd): Ditto.
36410 (_mm256_mmask_i32gather_epi32): Ditto.
36411 (_mm_mmask_i32gather_epi32): Ditto.
36412 (_mm256_mmask_i32gather_epi64): Ditto.
36413 (_mm_mmask_i32gather_epi64): Ditto.
36414 (_mm256_mmask_i64gather_epi32): Ditto.
36415 (_mm_mmask_i64gather_epi32): Ditto.
36416 (_mm256_mmask_i64gather_epi64): Ditto.
36417 (_mm_mmask_i64gather_epi64): Ditto.
36418 (_mm256_i32scatter_ps): Change __addr type to void*.
36419 (_mm256_mask_i32scatter_ps): Ditto.
36420 (_mm_i32scatter_ps): Ditto.
36421 (_mm_mask_i32scatter_ps): Ditto.
36422 (_mm256_i32scatter_pd): Ditto.
36423 (_mm256_mask_i32scatter_pd): Ditto.
36424 (_mm_i32scatter_pd): Ditto.
36425 (_mm_mask_i32scatter_pd): Ditto.
36426 (_mm256_i64scatter_ps): Ditto.
36427 (_mm256_mask_i64scatter_ps): Ditto.
36428 (_mm_i64scatter_ps): Ditto.
36429 (_mm_mask_i64scatter_ps): Ditto.
36430 (_mm256_i64scatter_pd): Ditto.
36431 (_mm256_mask_i64scatter_pd): Ditto.
36432 (_mm_i64scatter_pd): Ditto.
36433 (_mm_mask_i64scatter_pd): Ditto.
36434 (_mm256_i32scatter_epi32): Ditto.
36435 (_mm256_mask_i32scatter_epi32): Ditto.
36436 (_mm_i32scatter_epi32): Ditto.
36437 (_mm_mask_i32scatter_epi32): Ditto.
36438 (_mm256_i32scatter_epi64): Ditto.
36439 (_mm256_mask_i32scatter_epi64): Ditto.
36440 (_mm_i32scatter_epi64): Ditto.
36441 (_mm_mask_i32scatter_epi64): Ditto.
36442 (_mm256_i64scatter_epi32): Ditto.
36443 (_mm256_mask_i64scatter_epi32): Ditto.
36444 (_mm_i64scatter_epi32): Ditto.
36445 (_mm_mask_i64scatter_epi32): Ditto.
36446 (_mm256_i64scatter_epi64): Ditto.
36447 (_mm256_mask_i64scatter_epi64): Ditto.
36448 (_mm_i64scatter_epi64): Ditto.
36449 (_mm_mask_i64scatter_epi64): Ditto.
36450 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
36451 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
36452 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
36453 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
36454 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
36455 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
36456 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
36457 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
36458 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
36459 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
36460 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
36461 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
36462 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
36463 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
36464 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
36465 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
36466 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
36467 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
36468 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
36469 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
36470 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
36471 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
36472 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
36473 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
36474 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
36475 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
36476 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
36477 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
36478 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
36479 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
36480 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
36481 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
36482 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
36483 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
36484 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
36485 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
36486 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
36487 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
36488 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
36489 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
36490 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
36491 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
36492 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
36493 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
36494 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
36495 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
36496 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
36497 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
36498 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
36499 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
36500 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
36501 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
36502 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
36503 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
36504 definitions accordingly.
36506 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
36507 Kuan-Lin Chen <kuanlinchentw@gmail.com>
36510 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
36513 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
36516 * ira-conflicts.c (ira_build_conflicts): Update total conflict
36517 hard regs for inner regno.
36519 2017-01-17 Martin Liska <mliska@suse.cz>
36522 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
36523 assumption and add comment.
36525 2017-01-17 Nathan Sidwell <nathan@acm.org>
36527 * ipa-visibility.c (localize_node): New function, broken out of ...
36528 (function_and_variable_visibility): ... here. Call it.
36530 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36532 PR middle-end/77445
36533 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
36534 correctly set frequency of oudgoing edge.
36535 (duplicate_thread_path): Fix profile updating.
36537 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36540 * configure.ac: Add GCC_BASE_VER.
36541 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
36542 version from BASE-VER file.
36543 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
36544 (gcc.o): Depend on $(BASEVER).
36545 * common.opt (dumpfullversion): New option.
36546 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
36547 * doc/invoke.texi: Document -dumpfullversion.
36548 * doc/install.texi: Document --with-gcc-major-version-only.
36549 * configure: Regenerated.
36551 2017-01-17 Richard Biener <rguenther@suse.de>
36553 PR tree-optimization/71433
36554 * tree-vrp.c (register_new_assert_for): Merge same asserts
36555 on all incoming edges.
36556 (process_assert_insertions_for): Handle insertions at the
36559 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
36561 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
36562 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
36564 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
36567 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
36570 2017-01-17 Alan Modra <amodra@gmail.com>
36573 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
36574 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
36575 symbolic stack limit when pic.
36577 2017-01-16 Martin Sebor <msebor@redhat.com>
36579 PR tree-optimization/78608
36580 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
36582 2017-01-16 Jeff Law <law@redhat.com>
36585 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
36586 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
36587 for several include directories that may be relative to sysroot.
36588 * config/i386/x-mingw32 (gplus_includedir): Define.
36589 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
36590 (native_system_includedir): Likewise.
36591 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
36592 override if TARGET_SYSTEM_ROOT is defined.
36593 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
36595 PR tree-optimization/79090
36596 PR tree-optimization/33562
36597 PR tree-optimization/61912
36598 PR tree-optimization/77485
36599 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
36600 and computed trims into the dump file.
36602 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36604 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
36606 2017-01-16 Jakub Jelinek <jakub@redhat.com>
36609 * gimplify.c (gimplify_init_constructor): If want_value and
36610 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
36614 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
36615 sequence. Formatting fixes.
36616 (doloop_optimize): Formatting fixes.
36619 * gcc.c (debug_level_greater_than_spec_func): New function.
36620 (static_spec_functions): Add debug-level-gt spec function.
36621 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
36623 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36624 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36625 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
36626 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
36627 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
36628 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
36630 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
36632 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
36633 QImode fixups to general and mask registers only.
36635 2017-01-16 Carl Love <cel@us.ibm.com>
36637 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
36638 for built-in functions
36639 vector signed char vec_nabs (vector signed char)
36640 vector signed short vec_nabs (vector signed short)
36641 vector signed int vec_nabs (vector signed int)
36642 vector signed long long vec_nabs (vector signed long long)
36643 vector float vec_nabs (vector float)
36644 vector double vec_nabs (vector double)
36645 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
36647 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
36648 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
36649 * doc/extend.texi: Update the documentation file for the new built-in
36652 2017-01-16 Martin Sebor <msebor@redhat.com>
36654 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
36657 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36659 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
36660 UNSPEC_VSX__XXSPLTD to require special splat handling.
36662 2017-01-16 David Malcolm <dmalcolm@redhat.com>
36665 * system.h: Poison strndup.
36667 2017-01-16 Alan Modra <amodra@gmail.com>
36670 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
36673 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
36675 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
36677 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
36679 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
36680 call recog here. Assert that INSN_CODE (insn) is non-negative.
36682 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36685 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
36687 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
36688 in the currently scheduled RTL fragment.
36690 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36692 PR rtl-optimization/78751
36693 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
36696 2017-01-14 Jeff Law <law@redhat.com>
36698 PR tree-optimization/79090
36699 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
36700 variable length stores.
36701 (compute_trims): Delete dead assignment to *trim_tail.
36702 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
36705 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
36707 PR rtl-optimization/78626
36708 PR rtl-optimization/78727
36709 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
36710 of a block, and split such blocks after everything else is finished.
36712 2017-01-14 Alan Modra <amodra@gmail.com>
36715 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
36716 target legitimate_combined_insn.
36717 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
36718 (rs6000_legitimate_combined_insn): New function.
36719 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
36721 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
36722 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
36723 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
36725 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
36727 * doc/frontends.texi (G++ and GCC): Remove references to Java.
36729 2017-01-13 Jeff Law <law@redhat.com>
36731 PR tree-optimization/33562
36732 PR tree-optimization/61912
36733 PR tree-optimization/77485
36734 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
36736 (delete_dead_assignment): Likewise.
36737 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
36738 statement to delete_dead_call and delete_dead_assignment.
36740 2017-01-13 David Malcolm <dmalcolm@redhat.com>
36743 * substring-locations.c (format_warning_va): Strengthen case 1 so
36744 that both endpoints of the substring must be within the format
36745 range for just the substring to be printed.
36747 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
36749 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
36750 * config/i386/i386.c (ix86_target_string): Add missing options
36751 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
36752 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
36753 flags_other and ix86_target_other to flags2_other. Display unknown
36755 (ix86_valid_target_attribute_inner_p): Add missing options and
36756 reorder options by implied ISAs, as in ix86_target_string.
36758 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36760 * hash-table.h (hash_table::too_empty_p): New function.
36761 (hash_table::expand): Use it.
36762 (hash_table::traverse): Likewise.
36763 (hash_table::empty_slot): Use sizeof (value_type) instead of
36764 sizeof (PTR) to convert bytes to elements. Shrink the table
36765 if the current size is excessive for the current number of
36768 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36770 * ira-costs.c (record_reg_classes): Break from the inner loop
36771 early once alt_fail is known to be true. Update outer loop
36772 handling accordingly.
36774 2017-01-13 Jeff Law <law@redhat.com>
36776 * tree-ssa-dse.c (decrement_count): New function.
36777 (increment_start_addr, maybe_trim_memstar_call): Likewise.
36778 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
36779 when we know the partially dead statement is a mem* function.
36781 PR tree-optimization/61912
36782 PR tree-optimization/77485
36783 * tree-ssa-dse.c: Include expr.h.
36784 (maybe_trim_constructor_store): New function.
36785 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
36787 PR tree-optimization/33562
36788 PR tree-optimization/61912
36789 PR tree-optimization/77485
36790 * doc/invoke.texi: Document new dse-max-object-size param.
36791 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
36792 * tree-ssa-dse.c: Include params.h.
36793 (dse_store_status): New enum.
36794 (initialize_ao_ref_for_dse): New, partially extracted from
36796 (valid_ao_ref_for_dse, normalize_ref): New.
36797 (setup_live_bytes_from_ref, compute_trims): Likewise.
36798 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
36799 (maybe_trim_partially_dead_store): Likewise.
36800 (maybe_trim_complex_store): Likewise.
36801 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
36802 Track what bytes live from the original store. Return tri-state
36803 for dead, partially dead or live.
36804 (dse_dom_walker): Add constructor, destructor and new private members.
36805 (delete_dead_call, delete_dead_assignment): New extracted from
36807 (dse_optimize_stmt): Make a member of dse_dom_walker.
36808 Use initialize_ao_ref_for_dse.
36810 PR tree-optimization/33562
36811 PR tree-optimization/61912
36812 PR tree-optimization/77485
36813 * sbitmap.h (bitmap_count_bits): Prototype.
36814 (bitmap_clear_range, bitmap_set_range): Likewise.
36815 * sbitmap.c (bitmap_clear_range): New function.
36816 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
36818 2017-01-13 Martin Liska <mliska@suse.cz>
36821 * function.c (set_cfun): Add new argument force.
36822 * function.h (set_cfun): Likewise.
36823 * ipa-inline-transform.c (inline_call): Use the function when
36824 strict alising from is dropped for function we inline to.
36826 2017-01-13 Richard Biener <rguenther@suse.de>
36828 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
36829 for dumping GIMPLE INTEGER_CSTs.
36831 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
36833 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
36834 to 201112L since C++17.
36836 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
36839 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
36840 if -fsanitize=kernel-address is present.
36842 2017-01-13 Richard Biener <rguenther@suse.de>
36844 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
36845 as _Literal ( type ) number in case usual suffixes do not
36846 preserve all information.
36848 2017-01-13 Richard Biener <rguenther@suse.de>
36850 PR tree-optimization/77283
36851 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
36852 and ssa-iterators.h.
36853 (is_feasible_trace): Implement a cost model based on joiner
36856 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
36859 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
36860 char or short to __float128/_Float128 directly.
36862 2017-01-12 Martin Sebor <msebor@redhat.com>
36864 to -Wformat-overflow.
36865 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
36866 (min_bytes_remaining): Same.
36867 (get_string_length): Same.
36868 (format_string): Same.
36869 (format_directive): Same.
36871 (pass_sprintf_length::handle_gimple_call): Same.
36873 2017-01-12 Jakub Jelinek <jakub@redhat.com>
36875 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
36876 info.nowrite calls with no lhs that can't throw. Return bool
36877 whether gsi_remove has been called or not.
36878 (pass_sprintf_length::handle_gimple_call): Return bool whether
36879 try_substitute_return_value called gsi_remove. Formatting fix.
36880 (pass_sprintf_length::execute): Don't use gsi_remove if
36881 handle_gimple_call returned true.
36884 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
36885 be removed due to side-effects, don't remove following barrier nor
36886 turn the successor edge into fallthru edge.
36888 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36891 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
36892 element-reversing loads and stores as not swappable.
36894 2017-01-12 Nathan Sidwell <nathan@acm.org>
36895 Nicolai Stange <nicstange@gmail.com>
36897 * combine.c (try_combine): Don't ignore result of overlap checking
36898 loop. Combine overlap & asm check into single loop.
36900 2017-01-12 Richard Biener <rguenther@suse.de>
36902 * tree-pretty-print.c (dump_generic_node): Provide -gimple
36903 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
36905 2017-01-12 Richard Biener <rguenther@suse.de>
36907 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
36908 and TS_TARGET_OPTION directly derive from TS_BASE.
36909 * tree-core.h (tree_optimization_option): Derive from tree_base.
36910 (tree_target_option): Likewise.
36912 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
36914 * config/i386/i386.c (memory_address_length): Increase len
36915 only when rip_relative_addr_p returns false.
36917 2017-01-11 Julia Koval <julia.koval@intel.com>
36919 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
36920 (OPTION_MASK_ISA_SGX_SET): New.
36921 (ix86_handle_option): Handle OPT_msgx.
36922 * config.gcc: Added sgxintrin.h.
36923 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
36924 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
36925 * config/i386/i386.c (ix86_target_string): Add -msgx.
36927 (ix86_option_override_internal): Handle new options.
36928 (ix86_valid_target_attribute_inner_p): Add sgx.
36929 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
36930 * config/i386/i386.opt: Add msgx.
36931 * config/i386/sgxintrin.h: New file.
36932 * config/i386/x86intrin.h: Add sgxintrin.h.
36934 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36937 * fold-const.c (maybe_nonzero_address): Return 1 for function
36939 (tree_single_nonzero_warnv_p): Don't handle function local objects
36943 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
36946 2017-01-11 David Malcolm <dmalcolm@redhat.com>
36949 * opts.c: Include "spellcheck.h"
36950 (struct string_fragment): New struct.
36951 (struct edit_distance_traits<const string_fragment &>): New
36953 (get_closest_sanitizer_option): New function.
36954 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
36956 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36958 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
36960 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
36961 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
36962 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
36963 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
36964 for initial die_offset if dwarf_split_debug_info.
36965 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
36966 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
36968 (output_skeleton_debug_sections): Formatting fix. Use
36969 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
36970 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
36972 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
36974 * config/arm/cortex-a53.md: Add bypasses for
36975 cortex_a53_r2f_cvt.
36976 (cortex_a53_r2f): Only use for transfers.
36977 (cortex_a53_f2r): Likewise.
36978 (cortex_a53_r2f_cvt): Add reservation for conversions.
36979 (cortex_a53_f2r_cvt): Likewise.
36981 2017-01-11 Tamar Christina <tamar.christina@arm.com>
36983 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
36984 to all inlined functions, change static to extern.
36986 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
36989 * config/arm/arm.c (legitimize_pic_address): Handle reference to
36991 (arm_assemble_integer): Likewise.
36993 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
36995 * config.gcc: Use new awk script to check CPU, FPU and architecture
36996 parameters for --with-... options.
36997 * config/arm/parsecpu.awk: New file
36998 * config/arm/arm-cpus.in: New file.
36999 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37001 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37003 * config/arm/t-arm: Update dependency rules.
37004 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37005 of processing .def files.
37006 * config/arm/genopt.sh: Deleted.
37007 * config/arm/gentune.sh: Deleted.
37008 * config/arm/arm-cores.def: Deleted.
37009 * config/arm/arm-arches.def: Deleted.
37010 * config/arm/arm-fpus.def: Deleted.
37011 * config/arm/arm-tune.md: Regenerated.
37012 * config/arm/arm-tables.opt: Regenerated.
37013 * config/arm/arm-cpu.h: New generated file.
37014 * config/arm/arm-cpu-data.h: New generated file.
37015 * config/arm/arm-cpu-cdata.h: New generated file.
37017 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37020 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37022 (input_varpool_node): Unpack dynamically_initialized bit.
37024 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37026 PR rtl-optimization/79032
37027 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37028 the alignment of the adjusted memory reference against that of MODE,
37029 instead of the alignment of the original memory reference.
37031 2017-01-11 Martin Jambor <mjambor@suse.cz>
37033 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37035 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37036 decorated functions.
37038 2017-01-11 Richard Biener <rguenther@suse.de>
37040 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37041 set range/nonnull info for PHI results. Do not set it on
37042 stmts marked for removal.
37044 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37046 * expr.c (store_field): In the bitfield case, fetch the return value
37047 from the registers before applying a single big-endian adjustment.
37048 Always do a final load for a BLKmode value not larger than a word.
37050 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37053 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37054 that we correctly handle column numbers greater than
37055 LINE_MAP_MAX_COLUMN_NUMBER.
37057 2017-01-10 Martin Sebor <msebor@redhat.com>
37059 PR middle-end/78245
37060 * gimple-ssa-sprintf.c (get_destination_size): Call
37061 {init,fini}object_sizes.
37062 * tree-object-size.c (addr_object_size): Adjust.
37063 (pass_through_call): Adjust.
37064 (pass_object_sizes::execute): Adjust.
37065 * tree-object-size.h (fini_object_sizes): Declare.
37067 2017-01-10 Martin Sebor <msebor@redhat.com>
37069 PR tree-optimization/78775
37070 * builtins.c (get_size_range): Move...
37071 * calls.c: ...to here.
37072 (alloc_max_size): Accept zero argument.
37073 (operand_signed_p): Remove.
37074 (maybe_warn_alloc_args_overflow): Call get_size_range.
37075 * calls.h (get_size_range): Declare.
37077 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37079 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37080 from TI's devices.csv file as of September 2016.
37081 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37083 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37085 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37086 * doc/invoke.texi: Likewise.
37087 * doc/md.texi: Likewise.
37088 * doc/objc.texi: Likewise.
37090 2017-01-10 Joshua Conner <joshconner@google.com>
37092 * config/arm/fuchsia-elf.h: New file.
37093 * config/fuchsia.h: New file.
37094 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37095 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37097 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37099 2016-01-10 Richard Biener <rguenther@suse.de>
37101 PR tree-optimization/79034
37102 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37103 Propagate out degenerate PHIs in the joiner.
37105 2017-01-10 Martin Liska <mliska@suse.cz>
37107 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37108 (sort_congruence_classes_by_decl_uid): Likewise.
37109 (sort_congruence_class_groups_by_decl_uid): Likewise.
37110 (sem_item_optimizer::merge_classes): Sort class, groups in these
37111 classes and members in the groups by DECL_UID of declarations.
37112 This would make merge operations stable.
37114 2017-01-10 Martin Liska <mliska@suse.cz>
37116 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37117 usage of m_classes_vec.
37118 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37119 (sem_item_optimizer::get_group_by_hash): Likewise.
37120 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37121 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37122 (sem_item_optimizer::verify_classes): Likewise.
37123 (sem_item_optimizer::process_cong_reduction): Likewise.
37124 (sem_item_optimizer::dump_cong_classes): Likewise.
37125 (sem_item_optimizer::merge_classes): Likewise.
37126 * ipa-icf.h (congruence_class_hash): Rename from
37127 congruence_class_group_hash. Remove declaration of m_classes_vec.
37129 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37131 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37132 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37133 * config.gcc: Add avx512vpopcntdqintrin.h.
37134 * config/i386/avx512vpopcntdqintrin.h: New.
37135 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37136 * config/i386/i386-builtin-types.def: Add new types.
37137 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37138 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37139 __builtin_ia32_vpopcountq_v8di_mask): New.
37140 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37141 __AVX512VPOPCNTDQ__.
37142 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37143 (PTA_AVX512VPOPCNTDQ): Define.
37144 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37145 TARGET_AVX512VPOPCNTDQ_P): Define.
37146 * config/i386/i386.opt: Add mavx512vpopcntdq.
37147 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37148 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37150 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37152 PR middle-end/77484
37153 * predict.def (PRED_CALL): Set to 67.
37155 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37157 * expr.c (store_field): In the bitfield case, if the value comes from
37158 a function call and is of an aggregate type returned in registers, do
37159 not modify the field mode; extract the value in all cases if the mode
37160 is BLKmode and the size is not larger than a word.
37162 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37165 * config/i386/cpuid.h: Fix undefined behavior.
37167 2017-01-04 Jeff Law <law@redhat.com>
37169 PR tree-optimization/79007
37170 PR tree-optimization/67955
37171 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37172 conservative for pt.null when flag_non_call_exceptions is on.
37174 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37176 PR translation/79019
37177 PR translation/79020
37178 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37179 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37180 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37182 * config/avr/avr.opt (maccumulate-args): Likewise.
37183 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37184 * common.opt (freport-bug): Likewise.
37185 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37186 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37187 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37188 translatable string.
37189 * config/i386/i386.c (function_value_32): Likewise.
37190 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37191 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37193 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37194 * common/config/msp430/msp430-common.c (msp430_handle_option):
37196 * symtab.c (symtab_node::verify_base): Likewise.
37197 * opts.c (set_debug_level): Likewise.
37198 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37199 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37200 missing whitespace to translatable strings.
37201 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37202 * config/sh/superh.h: Likewise.
37203 * config/i386/xopintrin.h: Likewise.
37204 * config/i386/znver1.md: Likewise.
37205 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37206 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37207 * double-int.h (struct double_int): Likewise.
37208 * double-int.c (div_and_round_double): Likewise.
37209 * wide-int.cc: Likewise.
37210 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
37211 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
37212 * cfgcleanup.c (crossjumps_occured): Renamed to ...
37213 (crossjumps_occurred): ... this.
37214 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
37217 PR tree-optimization/78899
37218 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
37219 returning bool return struct loop *, NULL for failure and the new
37221 (versionable_outer_loop_p): Don't version outer loop if it has
37222 dont_vectorized bit set.
37223 (tree_if_conversion): When versioning outer loop, ensure
37224 tree_if_conversion is performed also on the inner loop of the
37225 non-vectorizable outer loop copy.
37226 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
37227 LOOP_VECTORIZED in inner loop of the scalar outer loop and
37228 prevent vectorization of it.
37229 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
37230 the outer loop vectorization of the non-scalar version is attempted
37231 before vectorization of the inner loop in scalar version. If
37232 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
37233 vectorization of its inner loop.
37234 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
37235 has 2 inner loops, rename also on edges from bb whose single pred
37236 is outer_loop->header. Fix typo in function comment.
37238 2017-01-09 Martin Sebor <msebor@redhat.com>
37241 * asan.c (asan_emit_stack_protection): Increase local buffer size
37242 to avoid snprintf truncation warning.
37244 2017-01-09 Andrew Pinski <apinski@cavium.com>
37246 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
37247 to reference thunderx2t99 for the tuning structure
37248 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
37250 (thunderx2t99_extra_costs): This.
37251 * config/aarch64/aarch64-tune.md: Regenerate.
37252 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
37253 (vulcan_addrcost_table): This.
37254 (vulcan_regmove_cost): Rename to ...
37255 (thunderx2t99_regmove_cost): This.
37256 (vulcan_vector_cost): Rename to ...
37257 (thunderx2t99_vector_cost): this.
37258 (vulcan_branch_cost): Rename to ...
37259 (thunderx2t99_branch_cost): This.
37260 (vulcan_tunings): Rename to ...
37261 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
37262 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
37264 2017-01-09 Martin Jambor <mjambor@suse.cz>
37268 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
37269 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
37270 (propagate_vr_accross_jump_function): Use the above function for all
37271 value range computations for pass-through jump functions and type
37272 converasion from explicit value range values.
37273 (ipcp_propagate_stage): Do not attempt to deduce types of formal
37274 parameters from TYPE_ARG_TYPES.
37275 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
37276 (ipa_write_node_info): Stream type of the actual argument.
37277 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
37279 2017-01-09 Martin Liska <mliska@suse.cz>
37282 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
37283 (lookup_compiler): Do not show error message with have_E.
37285 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37287 PR tree-optimization/78938
37288 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
37289 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
37290 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
37291 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
37294 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37296 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
37299 2017-01-09 Richard Biener <rguenther@suse.de>
37301 PR tree-optimization/78997
37302 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
37303 name condition properly.
37305 2017-01-09 Richard Biener <rguenther@suse.de>
37308 * dwarf2out.c (is_cxx): New overload with context.
37309 (is_naming_typedef_decl): Use it.
37311 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37313 * invoke.texi (Option Summary): Correct spacing in option lists
37314 and add line breaks to fix over-long lines.
37316 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37318 PR middle-end/17660
37320 * extend.texi (Common Variable Attributes): Add xref to GCC
37321 Internals manual to explain mode attribute keywords.
37323 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37326 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
37327 and Preprocessor Options.
37328 (Options for Linking): Document -pthread here....
37329 (RS/6000 and PowerPC Options): ...not here.
37330 (Solaris 2 Options): ...or here.
37331 * doc/cppopts.texi: Document -pthread.
37333 2017-01-08 Martin Sebor <msebor@redhat.com>
37335 PR middle-end/77708
37336 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
37337 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
37338 New member functions.
37339 (format_directive): Used them.
37341 (pass_sprintf_length::handle_gimple_call): Same.
37342 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
37343 to avoid truncation for any argument.
37344 (extract_affine_mul): Same.
37345 * tree.c (get_file_function_name): Same.
37347 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37349 PR middle-end/77484
37350 * predict.def (PRED_INDIR_CALL): Set to 86.
37352 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37354 PR preprocessor/54124
37355 * doc/cppopts.texi: Reformat -d subtable to list the full name
37356 of the options. Add cross-reference to the docs for the general
37357 compiler -d options.
37358 * doc/invoke.texi (Developer Options): Add cross-reference to the
37359 preprocessor-specific -d option documentation.
37361 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37363 PR preprocessor/13498
37364 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
37365 redudant material, and reflect new command-line options.
37366 (System Headers): Likewise.
37368 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37370 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
37371 -isystem, and -idirafter. Copy-edit.
37372 * doc/cppopts.texi: Copy-edit. Remove contradiction about
37373 default for -ftrack-macro-expansion. Delete obsolete and
37374 badly-formatted implementation details about -fdebug-cpp output.
37375 * doc/cppwarnopts.texi: Copy-edit.
37377 2017-01-07 David Malcolm <dmalcolm@redhat.com>
37380 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37381 that the transition from a max line width >= 1<<10 to narrower
37382 lines works correctly.
37384 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
37386 * doc/options.texi (PerFunction): New.
37387 * opt-functions.awk (switch_flags): Map both Optimization and
37388 PerFunction to CL_OPTIMIZATION.
37389 * opth-gen.awk: Test for PerFunction flag along with
37391 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
37392 it only when the latter is present. Skip those that don't in
37393 the hash function generator.
37394 * common.opt (fvar-tracking): Mark as PerFunction instead of
37396 (fvar-tracking-assignments): Likewise.
37397 (fvar-tracking-assignments-toggle): Likewise.
37398 (fvar-tracking-uninit): Likewise.
37400 2017-01-07 Jakub Jelinek <jakub@redhat.com>
37402 PR translation/79018
37403 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
37406 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
37409 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
37410 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
37411 TARGET_LONG_JUMP_TABLE_OFFSETS.
37412 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
37413 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
37414 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
37415 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
37416 * config/m68k/m68k.md (tablejump expander): Likewise.
37417 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
37418 TARGET_LONG_JUMP_TABLE_OFFSETS.
37419 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
37420 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
37422 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
37423 David Holsgrove <david.holsgrove@xilinx.com>
37425 * common/config/microblaze/microblaze-common.c
37426 (TARGET_EXCEPT_UNWIND_INFO): Remove.
37427 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
37429 * config/microblaze/microblaze.c (microblaze_must_save_register)
37430 (microblaze_expand_epilogue, microblaze_return_addr): Handle
37432 (microblaze_eh_return): New function.
37433 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
37434 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
37435 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
37436 * config/microblaze/microblaze.md (eh_return): New pattern.
37438 2017-01-06 Jakub Jelinek <jakub@redhat.com>
37440 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
37441 GCC_DIAGNOSTIC_STRINGIFY): Define.
37443 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
37445 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37447 * config/arm/arm.md (<mcrr>): New.
37449 * config/arm/arm.c (arm_arch5te): New.
37450 (arm_option_override): Set arm_arch5te.
37451 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
37453 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
37454 (arm_mcrr_qualifiers): ... this. New.
37455 (MRRC_QUALIFIERS): Define to...
37456 (arm_mrrc_qualifiers): ... this. New.
37457 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
37459 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
37460 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
37461 (MRRCI, mrrc, MRRC): New.
37462 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
37463 VUNSPEC_MRRC2): New.
37465 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37467 * config/arm/arm.md (<mcr>): New.
37469 * config/arm/arm.c (arm_coproc_builtin_available): Add
37470 support for mcr, mrc, mcr2 and mrc2.
37471 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
37472 (arm_mcr_qualifiers): ... this. New.
37473 (MRC_QUALIFIERS): Define to ...
37474 (arm_mrc_qualifiers): ... this. New.
37475 (MCR_QUALIFIERS): Define to ...
37476 (arm_mcr_qualifiers): ... this. New.
37477 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
37479 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
37480 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
37481 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
37482 VUNSPEC_MRC2): New.
37484 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37486 * config/arm/arm.md (*ldc): New.
37490 * config/arm/arm.c (arm_coproc_builtin_available): Add
37491 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
37492 (arm_coproc_ldc_stc_legitimate_address): New.
37493 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
37494 'qualifier_const_pointer'.
37495 (LDC_QUALIFIERS): Define to...
37496 (arm_ldc_qualifiers): ... this. New.
37497 (STC_QUALIFIERS): Define to...
37498 (arm_stc_qualifiers): ... this. New.
37499 * config/arm/arm-protos.h
37500 (arm_coproc_ldc_stc_legitimate_address): New.
37501 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
37502 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
37503 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
37504 stc2, stcl, stc2l): New.
37505 * config/arm/constraints.md (Uz): New.
37506 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
37507 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
37508 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
37509 VUNSPEC_STC2L): New.
37511 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37513 * config/arm/arm.md (<cdp>): New.
37514 * config/arm/arm.c (neon_const_bounds): Rename this ...
37515 (arm_const_bounds): ... this.
37516 (arm_coproc_builtin_available): New.
37517 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
37518 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
37519 (CDP_QUALIFIERS): Define to...
37520 (arm_cdp_qualifiers): ... this. New.
37522 (arm_expand_builtin_args): Add case for 6 arguments.
37523 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
37524 (arm_const_bounds): ... this.
37525 (arm_coproc_builtin_available): New.
37526 * config/arm/arm_acle.h (__arm_cdp): New.
37528 * config/arm/arm_acle_builtins.def (cdp): New.
37530 * config/arm/iterators.md (CDPI,CDP,cdp): New.
37531 * config/arm/neon.md: Rename all 'neon_const_bounds' to
37532 'arm_const_bounds'.
37533 * config/arm/types.md (coproc): New.
37534 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
37535 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
37536 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
37537 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
37539 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37541 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
37542 (UBINOP_QUALIFIERS): New.
37544 (acle_builtin_data): New. Change comment.
37545 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
37546 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
37547 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
37548 arm_acle_builtins.def.
37549 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
37550 (arm_init_acle_builtins): New.
37551 (CRC32_BUILTIN): Remove.
37552 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
37553 crc32cb, crc32ch and crc32cw.
37554 (arm_init_crc32_builtins): Remove.
37555 (arm_init_builtins): Use arm_init_acle_builtins rather
37556 than arm_init_crc32_builtins.
37557 (arm_expand_acle_builtin): New.
37558 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
37559 * config/arm/arm_acle_builtins.def: New.
37561 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37563 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
37564 (arm_builtin_datum): ... this.
37565 (arm_init_neon_builtin): Rename to ...
37566 (arm_init_builtin): ... this. Add a new parameters PREFIX
37567 and USE_SIG_IN_NAME.
37568 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
37569 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
37570 'arm_builtin_datum'.
37571 (arm_init_vfp_builtins): Likewise.
37572 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
37573 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
37574 (arm_expand_neon_args): Rename to ...
37575 (arm_expand_builtin_args): ... this. Rename builtin_arg
37576 enum values and differentiate between ARG_BUILTIN_MEMORY
37577 and ARG_BUILTIN_NEON_MEMORY.
37578 (arm_expand_neon_builtin_1): Rename to ...
37579 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
37580 values, arm_expand_builtin_args and add bool parameter NEON.
37581 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
37582 (arm_expand_vfp_builtin): Likewise.
37583 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
37585 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37587 PR middle-end/77484
37588 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
37589 * predict.c (tree_estimate_probability_bb): Reverse direction of
37590 polymorphic call predictor.
37592 2017-01-06 David Malcolm <dmalcolm@redhat.com>
37594 * passes.c (execute_one_pass): Split out pass-skipping logic into...
37595 (determine_pass_name_match): ...this new function and...
37596 (should_skip_pass_p): ...this new function.
37598 2017-01-06 Nathan Sidwell <nathan@acm.org>
37600 * ipa-visibility.c (function_and_variable_visibility): Reformat
37601 comments and long lines. Remove extrneous if.
37602 * symtab.c (symtab_node::make_decl_local): Fix code format.
37603 (symtab_node::set_section_for_node): Fix comment typo.
37605 2017-01-06 Martin Liska <mliska@suse.cz>
37608 * lra-constraints.c: Rename invariant to lra_invariant.
37609 * predict.c (set_even_probabilities): Initialize e to NULL.
37611 2017-01-05 Martin Sebor <msebor@redhat.com>
37613 PR tree-optimization/78910
37614 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
37615 (format_integer): Correct off-by-one error in the handling
37616 of precision with negative numbers in signed conversions..
37618 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
37620 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
37622 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37624 PR tree-optimization/71016
37625 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
37626 factor_out_conditional_conversion. Formatting fix.
37627 (factor_out_conditional_conversion): Add cond_stmt argument.
37628 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
37629 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
37632 2017-01-05 David Malcolm <dmalcolm@redhat.com>
37634 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
37635 read-rtl-function.o, and selftest-rtl.o.
37636 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
37637 (selftest::aarch64_test_loading_full_dump): New function.
37638 (selftest::aarch64_run_selftests): New function.
37639 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
37640 selftest::aarch64_run_selftests.
37641 * config/i386/i386.c
37642 (selftest::ix86_test_loading_dump_fragment_1): New function.
37643 (selftest::ix86_test_loading_call_insn): New function.
37644 (selftest::ix86_test_loading_full_dump): New function.
37645 (selftest::ix86_test_loading_unspec): New function.
37646 (selftest::ix86_run_selftests): Call the new functions.
37647 * emit-rtl.c (maybe_set_max_label_num): New function.
37648 * emit-rtl.h (maybe_set_max_label_num): New decl.
37649 * function.c (instantiate_decls): Guard call to
37650 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
37651 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
37653 * gensupport.c (gen_reader::gen_reader): Pass "false"
37654 for new "compact" param of rtx_reader.
37655 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
37656 rather than an empty string for NULL strings.
37657 * read-md.c: Potentially include config.h rather than bconfig.h.
37658 Wrap include of errors.h with #ifdef GENERATOR_FILE.
37659 (have_error): New global, copied from errors.c.
37660 (md_reader::read_name): Rename to...
37661 (md_reader::read_name_1): ...this, adding "out_loc" param,
37662 and converting "missing name or number" to returning false, rather
37664 (md_reader::read_name): Reimplement in terms of read_name_1.
37665 (md_reader::read_name_or_nil): New function.
37666 (md_reader::read_string): Handle "(nil)" by returning NULL.
37667 (md_reader::md_reader): Add new param "compact".
37668 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
37669 (md_reader::read_file): New method.
37670 * read-md.h (md_reader::md_reader): Add new param "compact".
37671 (md_reader::read_file): New method.
37672 (md_reader::is_compact): New accessor.
37673 (md_reader::read_name): Convert return type from void to file_location.
37674 (md_reader::read_name_or_nil): New decl.
37675 (md_reader::read_name_1): New decl.
37676 (md_reader::m_compact): New field.
37677 (noop_reader::noop_reader): Pass "false" for new "compact" param
37679 (rtx_reader::rtx_reader): Add new "compact" param.
37680 (rtx_reader::read_rtx_operand): Make virtual and convert return
37681 type from void to rtx.
37682 (rtx_reader::read_until): New decl.
37683 (rtx_reader::handle_any_trailing_information): New virtual function.
37684 (rtx_reader::postprocess): New virtual function.
37685 (rtx_reader::finalize_string): New virtual function.
37686 (rtx_reader::m_in_call_function_usage): New field.
37687 (rtx_reader::m_reuse_rtx_by_id): New field.
37688 * read-rtl-function.c: New file.
37689 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
37690 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
37691 (selftest::verify_three_block_rtl_cfg): New decl.
37692 * read-rtl-function.h: New file.
37693 * read-rtl.c: Potentially include config.h rather than bconfig.h.
37694 For host, include function.h, memmodel.h, and emit-rtl.h.
37695 (one_time_initialization): New function.
37696 (struct compact_insn_name): New struct.
37697 (compact_insn_names): New array.
37698 (find_code): Handle insn codes in compact dumps.
37699 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
37700 (bind_subst_iter_and_attr): Likewise.
37701 (add_condition_to_string): Likewise.
37702 (add_condition_to_rtx): Likewise.
37703 (apply_attribute_uses): Likewise.
37704 (add_current_iterators): Likewise.
37705 (apply_iterators): Likewise.
37706 (initialize_iterators): Guard usage of apply_subst_iterator with
37707 #ifdef GENERATOR_FILE.
37708 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
37709 (md_reader::read_mapping): Likewise.
37710 (add_define_attr_for_define_subst): Likewise.
37711 (add_define_subst_attr): Likewise.
37712 (read_subst_mapping): Likewise.
37713 (check_code_iterator): Likewise.
37714 (rtx_reader::read_rtx): Likewise. Move one-time initialization
37716 (one_time_initialization): New function.
37717 (rtx_reader::read_until): New method.
37718 (read_flags): New function.
37719 (parse_reg_note_name): New function.
37720 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
37721 Handle reuse_rtx ids.
37722 Wrap iterator lookup within #ifdef GENERATOR_FILE.
37723 Add parsing support for RTL dumps, mirroring the special-cases in
37724 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
37725 values, and calling handle_any_trailing_information.
37726 (rtx_reader::read_rtx_operand): Convert return type from void
37727 to rtx, returning return_rtx. Handle case 'e'. Call
37728 finalize_string on XSTR and XTMPL fields.
37729 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
37730 "(nil)" values were omitted. Call the postprocess vfunc on the
37732 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
37733 class ctor. Initialize m_in_call_function_usage. Call
37734 one_time_initialization.
37735 * rtl-tests.c (selftest::test_uncond_jump): Call
37736 set_new_first_and_last_insn.
37737 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
37738 * selftest-rtl.c: New file.
37739 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
37740 (selftest::get_insn_by_uid): New decl.
37741 * selftest-run-tests.c (selftest::run_tests): Call
37742 read_rtl_function_c_tests.
37743 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
37744 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
37747 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
37749 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
37750 operands in a special way. Assert that pos+len <= mode precision.
37752 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37754 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
37755 3 argument Alias with unlimited for the negative form.
37756 (fno-vect-cost-model): Removed.
37758 2017-01-05 Martin Liska <mliska@suse.cz>
37760 * hsa-gen.c (gen_hsa_divmod): New function.
37761 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
37763 2017-01-05 Martin Liska <mliska@suse.cz>
37766 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
37769 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37771 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
37772 small constant length operands.
37774 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37776 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
37777 between loop iterations.
37779 2017-01-05 Martin Liska <mliska@suse.cz>
37782 * gimplify.c (gimplify_decl_expr): Compare to
37783 asan_poisoned_variables instread of checking flags.
37784 (gimplify_target_expr): Likewise.
37785 (gimplify_expr): Likewise.
37786 (gimplify_function_tree): Conditionally initialize
37787 asan_poisoned_variables.
37789 2017-01-04 Jeff Law <law@redhat.com>
37791 PR tree-optimizatin/78812
37792 * rtl.h (contains_mem_rtx_p): Prototype.
37793 * ifcvt.c (containts_mem_rtx_p): Move from here to...
37794 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
37795 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
37796 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
37797 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
37799 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37801 * input.c (assert_char_at_range): Default-initialize actual_range.
37803 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37805 * df-scan.c (df_ref_create_structure): Make regno unsigned,
37806 to match the caller.
37808 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37810 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
37811 insns after final jump in test to emit dummy move.
37813 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37815 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
37816 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
37818 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37820 * multiple_target.c (create_dispatcher_calls): Init e_next.
37821 * tree-ssa-loop-split.c (split_loop): Init border.
37822 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
37825 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
37830 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
37831 (altivec_register_operand): Do not return true if the operand
37832 contains a SUBREG mixing SImode and SFmode.
37833 (vsx_register_operand): Likewise.
37834 (vsx_reg_sfsubreg_ok): New predicate.
37835 (vfloat_operand): Do not return true if the operand contains a
37836 SUBREG mixing SImode and SFmode.
37837 (vint_operand): Likewise.
37838 (vlogical_operand): Likewise.
37839 (gpc_reg_operand): Likewise.
37840 (int_reg_operand): Likewise.
37841 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
37842 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
37843 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
37845 (rs6000_emit_move_si_sf_subreg): New helper function.
37846 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
37847 fixup SUBREGs involving SImode and SFmode.
37848 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
37849 numbers for the new peephole2 optimization.
37850 (peephole2 for SFmode unions): New peephole2 to optimize cases in
37851 the GLIBC math library that do AND/IOR/XOR operations on single
37852 precision floating point.
37853 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
37854 target macros to say whether we need to avoid SUBREGs mixing
37856 (TARGET_ALLOW_SF_SUBREG): Likewise.
37857 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
37858 (UNSPEC_SI_FROM_SF): Likewise.
37859 (iorxor): Change spacing.
37860 (and_ior_xor): New iterator for AND, IOR, and XOR.
37861 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
37862 (movdi_from_sf_zero_ext): Likewise.
37863 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
37864 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
37865 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
37866 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
37867 (fms<mode>4): Likewise.
37868 (fnma<mode>4): Likewise.
37869 (fnms<mode>4): Likewise.
37870 (nfma<mode>4): Likewise.
37871 (nfms<mode>4): Likewise.
37873 2017-01-04 Marek Polacek <polacek@redhat.com>
37876 * doc/invoke.texi: Document -Wpointer-compare.
37878 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37880 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
37883 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
37884 descriptions for -gdwarf-5 and emit them as uleb128 instead of
37887 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
37890 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
37891 documentation of the powerpc_popcntb_ok attribute.
37892 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
37893 code to issue warning messages if a requested CPU configuration is
37894 not supported by the binary (assembler and loader) toolchain.
37895 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
37896 made to define a built-in function that has been disabled.
37897 (paired_init_builtins): Add assertion to prevent ICE if attempt is
37898 made to define a built-in function that has been disabled.
37899 (altivec_init_builtins): Add comment explaining why definition
37900 of the DST built-in functions is not preceded by an assertion
37901 check. Add assertions to prevent ICE if attempts are made to
37902 define an altivec predicate or an abs* built-in function that has
37904 (htm_init_builtins): Add comment explaining why definition of the
37905 htm built-in functions is not preceded by an assertion check.
37907 2017-01-04 Jeff Law <law@redhat.com>
37909 PR tree-optimizatin/67955
37910 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
37911 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
37912 the points-to solution does not include pt_null. Use DECL_PT_UID
37915 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
37917 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
37918 Use gen_int_mode instead of gen_lopwart for const_int operands.
37920 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37922 PR tree-optimization/71563
37923 * match.pd: Simplify X << Y into X if Y is known to be 0 or
37924 out of range value - has low bits known to be zero.
37926 2017-01-04 Alan Modra <amodra@gmail.com>
37928 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
37929 * configure: Regenerate.
37930 * config.in: Regenerate.
37932 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37935 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
37936 a substring of the message, but strcmp with the whole message. Ifdef
37937 ENABLE_NLS, translate the message first using dgettext.
37939 2017-01-03 Jeff Law <law@redhat.com>
37941 PR tree-optimizatin/78856
37942 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
37943 (mark_threaded_blocks): Remove code to truncate thread paths that
37944 cross multiple loop headers. Instead invalidate the cached loop
37945 iteration information and handle case of a thread path walking
37946 into an irreducible region.
37948 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
37951 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
37952 assertions. Add support for doing the signbit if the IEEE 128-bit
37953 floating point value is in a GPR.
37954 * config/rs6000/rs6000.md (Fsignbit): Delete.
37955 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
37956 Update the length attribute if the value is in a GPR.
37957 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
37958 the sign or zero extension instruction, since the value is always 0/1.
37959 (signbit<mode>2_dm2): Delete using <Fsignbit>.
37962 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
37963 extracting SImode to a GPR register so that we can generate a
37964 store, limit the vector to be in a traditional Altivec register
37965 for the vextuwrx instruction.
37967 2017-01-03 Ian Lance Taylor <iant@google.com>
37969 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
37971 2017-01-03 Martin Sebor <msebor@redhat.com>
37973 PR tree-optimization/78696
37974 * gimple-ssa-sprintf.c (format_floating): Correct handling of
37975 precision. Use MPFR for %f for greater fidelity. Correct handling
37977 (pass_sprintf_length::compute_format_length): Set width and precision
37978 specified by asrerisk to void_node for vararg functions.
37979 (try_substitute_return_value): Adjust dump output.
37981 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
37983 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
37985 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
37987 * doc/invoke.texi (SPARC options): Document -mlra as the default.
37988 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
37989 -mlra/-mno-lra was passed to the compiler.
37991 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
37993 PR rtl-optimization/65618
37994 * emit-rtl.c (try_split): Move initialization of "before" and
37995 "after" to just before the call to emit_insn_after_setloc.
37997 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
37999 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38001 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38003 * dwarf2out.c (gen_enumeration_type_die): When
38004 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38006 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38008 PR tree-optimization/78965
38009 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38010 Change first argument from const call_info & to call_info &. For %n
38011 set info.nowrite to false.
38013 PR middle-end/78901
38014 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38015 possibly throwing calls.
38017 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38018 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38019 and fns handling, rather than in a separate case SSA_NAME.
38021 2017-01-02 Jeff Law <law@redhat.com>
38023 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38024 fixes for first_period and second_period variables.
38026 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38029 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38030 (*insvqi_1): New insn pattern.
38031 (*insvqi_1_mem_rex64): Ditto.
38032 (*insvqi_2): Ditto.
38033 (*insvqi_3): Rename from *insvqi.
38035 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38037 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38039 * doc/cfg.texi (Edges): Remove reference to Java.
38040 (Maintaining the CFG): Ditto.
38042 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38044 PR middle-end/77674
38045 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38046 transparent aliases.
38048 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38050 PR middle-end/77484
38051 * predict.def (PRED_CALL): Update hitrate.
38052 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38053 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38054 into direct/indirect/polymorphic variants.
38056 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38058 Update copyright years.
38060 * gcc.c (process_command): Update copyright notice dates.
38061 * gcov-dump.c (print_version): Ditto.
38062 * gcov.c (print_version): Ditto.
38063 * gcov-tool.c (print_version): Ditto.
38064 * gengtype.c (create_file): Ditto.
38065 * doc/cpp.texi: Bump @copying's copyright year.
38066 * doc/cppinternals.texi: Ditto.
38067 * doc/gcc.texi: Ditto.
38068 * doc/gccint.texi: Ditto.
38069 * doc/gcov.texi: Ditto.
38070 * doc/install.texi: Ditto.
38071 * doc/invoke.texi: Ditto.
38073 Copyright (C) 2017 Free Software Foundation, Inc.
38075 Copying and distribution of this file, with or without modification,
38076 are permitted in any medium without royalty provided the copyright
38077 notice and this notice are preserved.